Paperid:1
Authors:Changyeon Won; Hae-Gon Jeon
Title: Learning Depth from Focus in the Wild
Abstract:
For better photography, most recent commercial cameras including smartphones have either adopted large-aperture lens to collect more light or used a burst mode to take multiple images within short times. These interesting features lead us to examine depth from focus/defocus. In this work, we present a convolutional neural network-based depth estimation from single focal stacks. Our method differs from relevant state-of-the-art works with three unique features. First, our method allows depth maps to be inferred in an end-to-end manner even with image alignment. Second, we propose a sharp region detection module to reduce blur ambiguities in subtle focus changes and weakly texture-less regions. Third, we design an effective downsampling module to ease flows of focal information in feature extractions. In addition, for the generalization of the proposed network, we develop a simulator to realistically reproduce the features of commercial cameras, such as changes in field of view, focal length and principal points. By effectively incorporating these three unique features, our network achieves the top rank in the DDFF 12-Scene benchmark on most metrics. We also demonstrate the effectiveness of the proposed method on various quantitative evaluations and real-world images taken from various off-the-shelf cameras compared with state-of-the-art methods. Our source code is publicly available at https://github.com/wcy199705/DfFintheWild.



Paperid:2
Authors:Zheng Dang; Lizhou Wang; Yu Guo; Mathieu Salzmann
Title: Learning-Based Point Cloud Registration for 6D Object Pose Estimation in the Real World
Abstract:
In this work, we tackle the task of estimating the 6D pose of an object from point cloud data. While recent learning-based approaches to addressing this task have shown great success on synthetic datasets, we have observed them to fail in the presence of real-world data. We thus analyze the causes of these failures, which we trace back to the difference between the feature distributions of the source and target point clouds, and the sensitivity of the widely-used SVD-based loss function to the range of rotation between the two point clouds. We address the first challenge by introducing a new normalization strategy, Match Normalization, and the second via the use of a loss function based on the negative log likelihood of point correspondences. Our two contributions are general and can be applied to many existing learning-based 3D object registration frameworks, which we illustrate by implementing them in two of them, DCP and IDAM. Our experiments on the real-scene TUD-L, LINEMOD and Occluded-LINEMOD datasets evidence the benefits of our strategies. They allow for the first time learning-based 3D object registration methods to achieve meaningful results on real-world data. We therefore expect them to be key to the future development of point cloud registration methods.



Paperid:3
Authors:Dingkang Liang; Wei Xu; Xiang Bai
Title: An End-to-End Transformer Model for Crowd Localization
Abstract:
Crowd localization, predicting head positions, is a more practical and high-level task than simply counting. Existing methods employ pseudo-bounding boxes or pre-designed localization maps, relying on complex post-processing to obtain the head positions. In this paper, we propose an elegant, end-to-end Crowd Localization TRansformer named CLTR that solves the task in the regression-based paradigm. The proposed method views the crowd localization as a direct set prediction problem, taking extracted features and trainable embeddings as input of the transformer-decoder. To reduce the ambiguous points and generate more reasonable matching results, we introduce a KMO-based Hungarian matcher, which adopts the nearby context as the external matching cost. Extensive experiments conducted on five datasets in various data settings show the effectiveness of our method. In particular, the proposed method achieves the best localization performance on the NWPU-Crowd, UCF-QNRF, and ShanghaiTech Part A datasets.



Paperid:4
Authors:Zhen Xing; Yijiang Chen; Zhixin Ling; Xiangdong Zhou; Yu Xiang
Title: Few-Shot Single-View 3D Reconstruction with Memory Prior Contrastive Network
Abstract:
3D reconstruction of novel categories based on few-shot learning is appealing in real-world applications and attracts increasing research interests. Previous approaches mainly focus on how to design shape prior models for different categories. Their performance on unseen categories is not very competitive. In this paper, we present a Memory Prior Contrastive Network (MPCN) that can store shape prior knowledge in a few-shot learning based 3D reconstruction framework. With the shape memory, a multi-head attention module is proposed to capture different parts of a candidate shape prior and fuse these parts together to guide 3D reconstruction of novel categories. Besides, we introduce a 3D-aware contrastive learning method, which can not only complement the retrieval accuracy of memory network, but also better organize image features for downstream tasks. Compared with previous few-shot 3D reconstruction methods, MPCN can handle the inter-class variability without category annotations. Experimental results on a benchmark synthetic dataset and the Pascal3D+ real-world dataset show that our model outperforms the current state-of-the-art methods significantly.



Paperid:5
Authors:Liang Peng; Xiaopei Wu; Zheng Yang; Haifeng Liu; Deng Cai
Title: DID-M3D: Decoupling Instance Depth for Monocular 3D Object Detection
Abstract:
Monocular 3D detection has drawn much attention from the community due to its low cost and setup simplicity. It takes an RGB image as input and predicts 3D boxes in the 3D space. The most challenging sub-task lies in the instance depth estimation. Previous works usually use a direct estimation method. However, in this paper we point out that the instance depth on the RGB image is non-intuitive. It is coupled by visual depth clues and instance attribute clues, making it hard to be directly learned in the network. Therefore, we propose to reformulate the instance depth to the combination of the instance visual surface depth (visual depth) and the instance attribute depth (attribute depth). The visual depth is related to objects’ appearances and positions on the image. By contrast, the attribute depth relies on objects’ inherent attributes, which are invariant to the object affine transformation on the image. Correspondingly, we decouple the 3D location uncertainty into visual depth uncertainty and attribute depth uncertainty. By combining different types of depths and associated uncertainties, we can obtain the final instance depth. Furthermore, data augmentation in monocular 3D detection is usually limited due to the physical nature, hindering the boost of performance. Based on the proposed instance depth disentanglement strategy, we can alleviate this problem. Evaluated on KITTI, our method achieves new state-of-the-art results, and extensive ablation studies validate the effectiveness of each component in our method. The codes are released at https://github.com/SPengLiang/DID-M3D.



Paperid:6
Authors:Weisong Ren; Lijun Wang; Yongri Piao; Miao Zhang; Huchuan Lu; Ting Liu
Title: Adaptive Co-Teaching for Unsupervised Monocular Depth Estimation
Abstract:
Unsupervised depth estimation using photometric losses suffers from local minimum and training instability. We address this issue by proposing an adaptive co-teaching framework to distill the learned knowledge from unsupervised teacher networks to a student network. We design an ensemble architecture for our teacher networks, integrating a depth basis decoder with multiple depth coefficient decoders. Depth prediction can then be formulated as a combination of the predicted depth bases weighted by coefficients. By further constraining their correlations, multiple coefficient decoders can yield a diversity of depth predictions, serving as the ensemble teachers. During the co-teaching step, our method allows different supervision sources from not only ensemble teachers but also photometric losses to constantly compete with each other, and adaptively select the optimal ones to teach the student, which effectively improves the ability of the student to jump out of the local minimum. Our method is shown to significantly benefit unsupervised depth estimation and sets new state of the art on both KITTI and Nuscenes datasets.



Paperid:7
Authors:Chen Zhao; Yinlin Hu; Mathieu Salzmann
Title: Fusing Local Similarities for Retrieval-Based 3D Orientation Estimation of Unseen Objects
Abstract:
In this paper, we tackle the task of estimating the 3D orientation of previously-unseen objects from monocular images. This task contrasts with the one considered by most existing deep learning methods which typically assume that the testing objects have been observed during training. To handle the unseen objects, we follow a retrieval-based strategy and prevent the network from learning object-specific features by computing multi-scale local similarities between the query image and synthetically-generated reference images. We then introduce an adaptive fusion module that robustly aggregates the local similarities into a global similarity score of pairwise images. Furthermore, we speed up the retrieval process by developing a fast retrieval strategy. Our experiments on the LineMOD, LineMOD-Occluded, and T-LESS datasets show that our method yields a significantly better generalization to unseen objects than previous works. Our code and pre-trained models are available at https://sailor-z.github.io/projects/Unseen_Object_Pose.html.



Paperid:8
Authors:Liang Peng; Fei Liu; Zhengxu Yu; Senbo Yan; Dan Deng; Zheng Yang; Haifeng Liu; Deng Cai
Title: Lidar Point Cloud Guided Monocular 3D Object Detection
Abstract:
Monocular 3D object detection is a challenging task in the self-driving and computer vision community. As a common practice, most previous works use manually annotated 3D box labels, where the annotating process is expensive. In this paper, we find that the precisely and carefully annotated labels may be unnecessary in monocular 3D detection, which is an interesting and counterintuitive finding. Using rough labels that are randomly disturbed, the detector can achieve very close accuracy compared to the one using the ground-truth labels. We delve into this underlying mechanism and then empirically find that: concerning the label accuracy, the 3D location part in the label is preferred compared to other parts of labels. Motivated by the conclusions above and considering the precise LiDAR 3D measurement, we propose a simple and effective framework, dubbed LiDAR point cloud guided monocular 3D object detection (LPCG). This framework is capable of either reducing the annotation costs or considerably boosting the detection accuracy without introducing extra annotation costs. Specifically, It generates pseudo labels from unlabeled LiDAR point clouds. Thanks to accurate LiDAR 3D measurements in 3D space, such pseudo labels can replace manually annotated labels in the training of monocular 3D detectors, since their 3D location information is precise. LPCG can be applied into any monocular 3D detector to fully use massive unlabeled data in a self-driving system. As a result, in KITTI benchmark, we take the first place on both monocular 3D and BEV (bird’s-eye-view) detection with a significant margin. In Waymo benchmark, our method using 10% labeled data achieves comparable accuracy to the baseline detector using 100% labeled data. The codes are released at https://github.com/SPengLiang/LPCG.



Paperid:9
Authors:Weiyang Liu; Zhen Liu; Liam Paull; Adrian Weller; Bernhard Schölkopf
Title: Structural Causal 3D Reconstruction
Abstract:
This paper considers the problem of unsupervised 3D object reconstruction from in-the-wild single-view images. Due to ambiguity and intrinsic ill-posedness, this problem is inherently difficult to solve and therefore requires strong regularization to achieve disentanglement of different latent factors. Unlike existing works that introduce explicit regularizations into objective functions, we look into a different space for implicit regularization -- the structure of latent space. Specifically, we restrict the structure of latent space to capture a topological causal ordering of latent factors (i.e., representing causal dependency as a directed acyclic graph). We first show that different causal orderings matter for 3D reconstruction, and then explore several approaches to find a task-dependent causal factor ordering. Our experiments demonstrate that the latent space structure indeed serves as an implicit regularization and introduces an inductive bias beneficial for reconstruction.



Paperid:10
Authors:Niloofar Azizi; Horst Possegger; Emanuele Rodolà; Horst Bischof
Title: 3D Human Pose Estimation Using Möbius Graph Convolutional Networks
Abstract:
3D human pose estimation is fundamental to understanding human behavior. Recently, promising results have been achieved by graph convolutional networks(GCNs), which achieve state-of-the-art performance and provide rather light-weight architectures. However, a major limitation of GCNs is their inability to encode all the transformations between joints explicitly. To address this issue, we propose a novel spectral GCN using the Möbius transformation (MöbiusGCN). In particular, this allows us to directly and explicitly encode the transformation between joints, resulting in a significantly more compact representation. Compared to even the lightest architectures so far, our novel approach requires 90-98% fewer parameters, i.e. our lightest MöbiusGCN uses only 0.042M trainable parameters. Besides the drastic parameter reduction, explicitly encoding the transformation of joints also enables us to achieve state-of-the-art results. We evaluate our approach on the two challenging pose estimation benchmarks, Human3.6M and MPI-INF-3DHP, demonstrating both state-of-the-art results and the generalization capabilities of MöbiusGCN.



Paperid:11
Authors:Tianxin Huang; Xuemeng Yang; Jiangning Zhang; Jinhao Cui; Hao Zou; Jun Chen; Xiangrui Zhao; Yong Liu
Title: Learning to Train a Point Cloud Reconstruction Network without Matching
Abstract:
Reconstruction networks for well-ordered data such as 2D images and 1D continuous signals are easy to optimize through element-wised squared errors, while permutation-arbitrary point clouds cannot be constrained directly because their points permutations are not fixed. Though existing works design algorithms to match two point clouds and evaluate shape errors based on matched results, they are limited by pre-defined matching processes. In this work, we propose a novel framework named PCLossNet which learns to train a point cloud reconstruction network without any matching. By training through an adversarial process together with the reconstruction network, PCLossNet can better explore the differences between point clouds and create more precise reconstruction results. Experiments on multiple datasets prove the superiority of our method, where PCLossNet can help networks achieve much lower reconstruction errors and extract more representative features, with about 4 times faster training efficiency than the commonly-used EMD loss. Our codes can be found in https://github.com/Tianxinhuang/PCLossNet.



Paperid:12
Authors:Zhijie Shen; Chunyu Lin; Kang Liao; Lang Nie; Zishuo Zheng; Yao Zhao
Title: PanoFormer: Panorama Transformer for Indoor 360° Depth Estimation
Abstract:
Existing panoramic depth estimation methods based on convolutional neural networks (CNNs) focus on removing panoramic distortions, failing to perceive panoramic structures efficiently due to the fixed receptive field in CNNs. This paper proposes the panorama Transformer (named PanoFormer) to estimate the depth in panorama images, with tangent patches from spherical domain, learnable token flows, and panorama specific metrics. In particular, we divide patches on the spherical tangent domain into tokens to reduce the negative effect of panoramic distortions. Since the geometric structures are essential for depth estimation, a self-attention module is redesigned with an additional learnable token flow. In addition, considering the characteristic of the spherical domain, we present two panorama-specific metrics to comprehensively evaluate the panoramic depth estimation models’ performance. Extensive experiments demonstrate that our approach significantly outperforms the state-of-the-art methods. At last, the proposed method can be effectively extended to solve semantic panorama segmentation, a similar pixel2pixel task. Code will be released upon acceptance.



Paperid:13
Authors:Xuan Gong; Meng Zheng; Benjamin Planche; Srikrishna Karanam; Terrence Chen; David Doermann; Ziyan Wu
Title: Self-supervised Human Mesh Recovery with Cross-Representation Alignment
Abstract:
Fully supervised human mesh recovery methods are data-hungry and have poor generalizability due to the limited availability and diversity of 3D-annotated benchmark datasets. Recent progress in self-supervised human mesh recovery has been made using synthetic-data-driven training paradigms where the model is trained from synthetic paired 2D representation (e.g., 2D keypoints and segmentation masks) and 3D mesh. However, on synthetic dense correspondence maps (i.e., IUV) few have been explored since the domain gap between synthetic training data and real testing data is hard to address for 2D dense representation. To alleviate this domain gap on IUV, we propose cross-representation alignment utilizing the complementary information from the robust but sparse representation (2D keypoints). Specifically, the alignment errors between initial mesh estimation and both 2D representations are forwarded into regressor and dynamically corrected in the following mesh regression. This adaptive cross-representation alignment explicitly learns from the deviations and captures complementary information: robustness from sparse representation and richness from dense representation. We conduct extensive experiments on multiple standard benchmark datasets and demonstrate competitive results, helping take a step towards reducing the annotation effort needed to produce state-of-the-art models in human mesh estimation.



Paperid:14
Authors:Zerui Chen; Yana Hasson; Cordelia Schmid; Ivan Laptev
Title: AlignSDF: Pose-Aligned Signed Distance Fields for Hand-Object Reconstruction
Abstract:
Recent work achieved impressive progress towards joint reconstruction of hands and manipulated objects from monocular color images. Existing methods focus on two alternative representations in terms of either parametric meshes or signed distance fields (SDFs). On one side, parametric models can benefit from prior knowledge at the cost of limited shape deformations and mesh resolutions. Mesh models, hence, may fail to precisely reconstruct details such as contact surfaces of hands and objects. SDF-based methods, on the other side, can represent arbitrary details but are lacking explicit priors. In this work we aim to improve SDF models using priors provided by parametric representations. In particular, we propose a joint learning framework that disentangles the pose and the shape. We obtain hand and object poses from parametric models and use them to align SDFs in 3D space. We show that such aligned SDFs better focus on reconstructing shape details and improve reconstruction accuracy both for hands and objects. We evaluate our method and demonstrate significant improvements over the state of the art on the challenging ObMan and DexYCB benchmarks.



Paperid:15
Authors:Yiming Qian; James H. Elder
Title: A Reliable Online Method for Joint Estimation of Focal Length and Camera Rotation
Abstract:
Linear perspective cues deriving from regularities of the built environment can be used to recalibrate both intrinsic and extrinsic camera parameters online, but these estimates can be unreliable due to irregularities in the scene, uncertainties in line segment estimation and background clutter. Here we address this challenge through four initiatives. First, we use the PanoContext panoramic image dataset to curate a novel and realistic dataset of planar projections over a broad range of scenes, focal lengths and camera poses. Second, we use this novel dataset and the YorkUrbanDB to systematically evaluate the linear perspective deviation measures frequently found in the literature and show that the choice of deviation measure and likelihood model has a huge impact on reliability. Third, we use these findings to create a novel system for online camera calibration we call fR, and show that it outperforms the prior state of the art, substantially reducing error in estimated camera rotation and focal length. Our fourth contribution is a novel and efficient approach to estimating uncertainty that can dramatically improve online reliability for performance-critical applications by strategically selecting which frames to use for recalibration.



Paperid:16
Authors:Wenqi Yang; Guanying Chen; Chaofeng Chen; Zhenfang Chen; Kwan-Yee K. Wong
Title: PS-NeRF: Neural Inverse Rendering for Multi-View Photometric Stereo
Abstract:
Traditional multi-view photometric stereo (MVPS) methods are often composed of multiple disjoint stages, resulting in noticeable accumulated errors. In this paper, we present a neural inverse rendering method for MVPS based on implicit representation. Given multi-view images of a non-Lambertian object illuminated by multiple unknown directional lights, our method jointly estimates the geometry, materials, and lights. Our method first employs multi-light images to estimate per-view surface normal maps, which are used to regularize the normals derived from the neural radiance field. It then jointly optimizes the surface normals, spatially-varying BRDFs, and lights based on a shadow-aware differentiable rendering layer. After optimization, the reconstructed object can be used for novel-view rendering, relighting, and material editing. Experiments on both synthetic and real datasets demonstrate that our method achieves far more accurate shape reconstruction than existing MVPS and neural rendering methods. Our code and model can be found at https://ywq.github.io/psnerf.



Paperid:17
Authors:Tom Monnier; Matthew Fisher; Alexei A. Efros; Mathieu Aubry
Title: Share with Thy Neighbors: Single-View Reconstruction by Cross-Instance Consistency
Abstract:
Approaches for single-view reconstruction typically rely on viewpoint annotations, silhouettes, the absence of background, multiple views of the same instance, a template shape, or symmetry. We avoid all such supervision and assumptions by explicitly leveraging the consistency between images of different object instances. As a result, our method can learn from large collections of unlabelled images depicting the same object category. Our main contributions are two ways for leveraging cross-instance consistency: (i) progressive conditioning, a training strategy to gradually specialize the model from category to instances in a curriculum learning fashion; and (ii) neighbor reconstruction, a loss enforcing consistency between instances having similar shape or texture. Also critical to the success of our method are: our structured autoencoding architecture decomposing an image into explicit shape, texture, pose, and background; an adapted formulation of differential rendering; and a new optimization scheme alternating between 3D and pose learning. We compare our approach, UNICORN, both on the diverse synthetic ShapeNet dataset - the classical benchmark for methods requiring multiple views as supervision - and on standard real-image benchmarks (Pascal3D+ Car, CUB) for which most methods require known templates and silhouette annotations. We also showcase applicability to more challenging real-world collections (CompCars, LSUN), where silhouettes are not available and images are not cropped around the object.



Paperid:18
Authors:Jingyuan Ma; Xiangyu Lei; Nan Liu; Xian Zhao; Shiliang Pu
Title: Towards Comprehensive Representation Enhancement in Semantics-Guided Self-Supervised Monocular Depth Estimation
Abstract:
Semantics-guided self-supervised monocular depth estimation has been widely researched, owing to the strong cross-task correlation of depth and semantics. However, since depth estimation and semantic segmentation are fundamentally two types of tasks: one is regression while the other is classification, the distribution of depth feature and semantic feature are naturally different. Previous works that leverage semantic information in depth estimation mostly neglect such representational discrimination, which leads to insufficient representation enhancement of depth feature. In this work, we propose an attention-based module to enhance task-specific feature by addressing their feature uniqueness within instances. Additionally, we propose a metric learning based approach to accomplish comprehensive enhancement on depth feature by creating a separation between instances in feature space. Extensive experiments and analysis demonstrate the effectiveness of our proposed method. In the end, our method achieves the state-of-the-art performance on KITTI dataset.



Paperid:19
Authors:Zhe Li; Zerong Zheng; Hongwen Zhang; Chaonan Ji; Yebin Liu
Title: AvatarCap: Animatable Avatar Conditioned Monocular Human Volumetric Capture
Abstract:
To address the ill-posed problem caused by partial observations in monocular human volumetric capture, we present AvatarCap, a novel framework that introduces animatable avatars into the capture pipeline for high-fidelity reconstruction in both visible and invisible regions. Our method firstly creates an animatable avatar for the subject from a small number ( 20) of 3D scans as a prior. Then given a monocular RGB video of this subject, our method integrates information from both the image observation and the avatar prior, and accordingly reconstructs high-fidelity 3D textured models with dynamic details regardless of the visibility. To learn an effective avatar for volumetric capture from only few samples, we propose GeoTexAvatar, which leverages both geometry and texture supervisions to constrain the pose-dependent dynamics in a decomposed implicit manner. An avatar-conditioned volumetric capture method that involves a canonical normal fusion and a reconstruction network is further proposed to integrate both image observations and avatar dynamics for high-fidelity reconstruction in both observed and invisible regions. Overall, our method enables monocular human volumetric capture with detailed and pose-dependent dynamics, and the experiments show that our method outperforms state of the art.



Paperid:20
Authors:Junhyeong Cho; Kim Youwang; Tae-Hyun Oh
Title: Cross-Attention of Disentangled Modalities for 3D Human Mesh Recovery with Transformers
Abstract:
Transformer encoder architectures have recently achieved state-of-the-art results on monocular 3D human mesh reconstruction, but they require a substantial number of parameters and expensive computations. Due to the large memory overhead and slow inference speed, it is difficult to deploy such models for practical use. In this paper, we propose a novel transformer encoder-decoder architecture for 3D human mesh reconstruction from a single image, called FastMETRO. We identify the performance bottleneck in the encoder-based transformers is caused by the token design which introduces high complexity interactions among input tokens. We disentangle the interactions via an encoder-decoder architecture, which allows our model to demand much fewer parameters and shorter inference time. In addition, we impose the prior knowledge of human body’s morphological relationship via attention masking and mesh upsampling operations, which leads to faster convergence with higher accuracy. Our FastMETRO improves the Pareto-front of accuracy and efficiency, and clearly outperforms image-based methods on Human3.6M and 3DPW. Furthermore, we validate its generalizability on FreiHAND.



Paperid:21
Authors:Pan Ji; Qingan Yan; Yuxin Ma; Yi Xu
Title: GeoRefine: Self-Supervised Online Depth Refinement for Accurate Dense Mapping
Abstract:
We present a robust and accurate depth refinement system, named GeoRefine, for geometrically-consistent dense mapping from monocular sequences. GeoRefine consists of three modules: a hybrid SLAM module using learning-based priors, an online depth refinement module leveraging self-supervision, and a global mapping module via TSDF fusion. The proposed system is online by design and achieves great robustness and accuracy via: (i) a robustified hybrid SLAM that incorporates learning-based optical flow and/or depth; (ii) self-supervised losses that leverage SLAM outputs and enforce long-term geometric consistency; (iii) careful system design that avoids degenerate cases in online depth refinement. We extensively evaluate GeoRefine on multiple public datasets and reach as low as 5% absolute relative depth errors.



Paperid:22
Authors:Zhiqiang Yan; Xiang Li; Kun Wang; Zhenyu Zhang; Jun Li; Jian Yang
Title: Multi-modal Masked Pre-training for Monocular Panoramic Depth Completion
Abstract:
In this paper, we formulate a potentially valuable panoramic depth completion (PDC) task as panoramic 3D cameras often produce 360° depth with missing data in complex scenes. Its goal is to recover dense panoramic depths from raw sparse ones and panoramic RGB images. To deal with the PDC task, we train a deep network that takes both depth and image as inputs for the dense panoramic depth recovery. However, it needs to face a challenging optimization problem of the network parameters due to its non-convex objective function. To address this problem, we propose a simple yet effective approach termed M³PT: multi-modal masked pre-training. Specifically, during pre-training, we simultaneously cover up patches of the panoramic RGB image and sparse depth by shared random mask, then reconstruct the sparse depth in the masked regions. To our best knowledge, it is the first time that we show the effectiveness of masked pre-training in a multi-modal vision task, instead of the single-modal task resolved by masked autoencoders (MAE). Different from MAE where fine-tuning completely discards the decoder part of pre-training, there is no architectural difference between the pre-training and fine-tuning stages in our M³PT as they only differ in the prediction density, which potentially makes the transfer learning more convenient and effective. Extensive experiments verify the effectiveness of M³PT on three panoramic datasets. Notably, we improve the state-of-the-art baselines by averagely 29.2% in RMSE, 51.7% in MRE, 49.7% in MAE, and 37.5% in RMSElog on three benchmark datasets.



Paperid:23
Authors:Shi Gong; Xiaoqing Ye; Xiao Tan; Jingdong Wang; Errui Ding; Yu Zhou; Xiang Bai
Title: GitNet: Geometric Prior-Based Transformation for Birds-Eye-View Segmentation
Abstract:
Birds-eye-view (BEV) semantic segmentation is critical for autonomous driving for its powerful spatial representation ability. It is challenging to estimate the BEV semantic maps from monocular images due to the spatial gap, since it is implicitly required to realize both the perspective-to-BEV transformation and segmentation. We present a novel two-stage Geometry PrIor-based Transformation framework named GitNet, consisting of (i) the geometry-guided pre-alignment and (ii) ray-based transformer. In the first stage, we decouple the BEV segmentation into the perspective image segmentation and geometric prior-based mapping, with explicit supervision by projecting the BEV semantic labels onto the image plane to learn visibility-aware features and learnable geometry to translate into BEV space. Second, the pre-aligned coarse BEV features are further deformed by ray-based transformers to take visibility knowledge into account. GitNet achieves the leading performance on the challenging nuScenes and Argoverse Datasets. The code will be publicly available.



Paperid:24
Authors:Chun-Han Yao; Jimei Yang; Duygu Ceylan; Yi Zhou; Yang Zhou; Ming-Hsuan Yang
Title: Learning Visibility for Robust Dense Human Body Estimation
Abstract:
Estimating 3D human pose and shape from 2D images is a crucial yet challenging task. While prior methods with model-based representations can perform reasonably well on whole-body images, they often fail when parts of the body are occluded or outside the frame. Moreover, these results usually do not faithfully capture the human silhouettes due to their limited representation power of deformable models (e.g., representing only the naked body). An alternative approach is to estimate dense vertices of a predefined template body in the image space. Such representations are effective in localizing vertices within an image but cannot handle out-of-frame body parts. In this work, we learn dense human body estimation that is robust to partial observations. We explicitly model the visibility of human joints and vertices in the x, y, and z axes separately. The visibility in x and y axes help distinguishing out-of-frame cases, and the visibility in depth axis corresponds to occlusions (either self-occlusions or occlusions by other objects). We obtain pseudo ground-truths of visibility labels from dense UV correspondences and train a neural network to predict visibility along with 3D coordinates. We show that visibility can serve as 1) an additional signal to resolve depth ordering ambiguities of self-occluded vertices and 2) a regularization term when fitting a human body model to the predictions. Extensive experiments on multiple 3D human datasets demonstrate that visibility modeling significantly improves the accuracy of human body estimation, especially for partial-body cases. Our project page with code is at: https://github.com/chhankyao/visdb.



Paperid:25
Authors:Haolin Liu; Yujian Zheng; Guanying Chen; Shuguang Cui; Xiaoguang Han
Title: Towards High-Fidelity Single-View Holistic Reconstruction of Indoor Scenes
Abstract:
We present a new framework to reconstruct holistic 3D indoor scenes including both room background and indoor objects from single-view images. Existing methods can only produce 3D shapes of indoor objects with limited geometry quality because of the heavy occlusion of indoor scenes. To solve this, we propose an instance-aligned implicit function (InstPIFu) for detailed object reconstruction. Combining with instance-aligned attention module, our method is empowered to decouple mixed local features toward the occluded instances. Additionally, unlike previous methods that simply represents the room background as a 3D bounding box, depth map or a set of planes, we recover the fine geometry of the background via implicit representation. Extensive experiments on the SUN RGB-D, Pix3D, 3D-FUTURE, and 3D-FRONT datasets demonstrate that our method outperforms existing approaches in both background and foreground object reconstruction. Our code and model will be made publicly available.



Paperid:26
Authors:Zuoyue Li; Tianxing Fan; Zhenqiang Li; Zhaopeng Cui; Yoichi Sato; Marc Pollefeys; Martin R. Oswald
Title: CompNVS: Novel View Synthesis with Scene Completion
Abstract:
We introduce a scalable framework for novel view synthesis from RGB-D images with largely incomplete scene coverage. While generative neural approaches have demonstrated spectacular results on 2D images, they have not yet achieved similar photorealistic results in combination with scene completion where a spatial 3D scene understanding is essential. To this end, we propose a generative pipeline performing on a sparse grid-based neural scene representation to complete unobserved scene parts via a learned distribution of scenes in a 2.5D-3D-2.5D manner. We process encoded image features in 3D space with a geometry completion network and a subsequent texture inpainting network to extrapolate the missing area. Photorealistic image sequences can be finally obtained via consistency-relevant differentiable rendering. Comprehensive experiments show that the graphical outputs of our method outperform the state of the art, especially within unobserved scene parts.



Paperid:27
Authors:Chenjian Gao; Qian Yu; Lu Sheng; Yi-Zhe Song; Dong Xu
Title: SketchSampler: Sketch-Based 3D Reconstruction via View-Dependent Depth Sampling
Abstract:
Reconstructing a 3D shape based on a single sketch image is challenging due to the large domain gap between a sparse, irregular sketch and a regular, dense 3D shape. Existing works try to employ the global feature extracted from sketch to directly predict the 3D coordinates, but they usually suffer from losing fine details that are not faithful to the input sketch. Through analyzing the 3D-to-2D projection process, we notice that the density map that characterizes the distribution of 2D point clouds (i.e., the probability of points projected at each location of the projection plane) can be used as a proxy to facilitate the reconstruction process. To this end, we first translate a sketch via an image translation network to a more informative 2D representation that can be used to generate a density map. Next, a 3D point cloud is reconstructed via a two-stage probabilistic sampling process: first recovering the 2D points (i.e., the x and y coordinates) by sampling the density map; and then predicting the depth (i.e., the z coordinate) by sampling the depth values at the ray determined by each 2D point. Extensive experiments are conducted, and both quantitative and qualitative results show that our proposed approach significantly outperforms other baseline methods.



Paperid:28
Authors:Shariq Farooq Bhat; Ibraheem Alhashim; Peter Wonka
Title: LocalBins: Improving Depth Estimation by Learning Local Distributions
Abstract:
We propose a novel architecture for depth estimation from a single image. The architecture itself is based on the popular encoder-decoder architecture that is frequently used as a starting point for all dense regression tasks. We build on AdaBins which estimates a global distribution of depth values for the input image and evolve the architecture in two ways. First, instead of predicting global depth distributions, we predict depth distributions of local neighborhoods at every pixel. Second, instead of predicting depth distributions only towards the end of the decoder, we involve all layers of the decoder. We call this new architecture LocalBins. Our results demonstrate a clear improvement over the state-of-the-art in all metrics on the NYU-Depth V2 dataset. Code and pretrained models will be made publicly available.



Paperid:29
Authors:Feng Liu; Xiaoming Liu
Title: 2D GANs Meet Unsupervised Single-View 3D Reconstruction
Abstract:
Recent research has shown that controllable image generation based on pre-trained GANs can benefit a wide range of computer vision tasks. However, less attention has been devoted to 3D vision tasks. In light of this, we propose a novel image-conditioned neural implicit field, which can leverage 2D supervisions from GAN-generated multi-view images and perform the single-view reconstruction of generic objects. Firstly, a novel offline StyleGAN-based generator is presented to generate plausible pseudo images with full control over the viewpoint. Then, we propose to utilize a neural implicit function, along with a differentiable renderer to learn 3D geometry from pseudo images with object masks and rough pose initializations. To further detect the unreliable supervisions, we introduce a novel uncertainty module to predict uncertainty maps, which remedy the negative effect of uncertain regions in pseudo images, leading to a better reconstruction performance. The effectiveness of our approach is demonstrated through superior single-view 3D reconstruction results of generic objects.



Paperid:30
Authors:Zhengqi Li; Qianqian Wang; Noah Snavely; Angjoo Kanazawa
Title: InfiniteNature-Zero: Learning Perpetual View Generation of Natural Scenes from Single Images
Abstract:
We present a method for learning to generate unbounded flythrough videos of natural scenes starting from a single view. This capability is learned from a collection of single photographs, without requiring camera poses or even multiple views of each scene. To achieve this, we propose a novel self-supervised view generation training paradigm where we sample and render virtual camera trajectories, including cyclic camera paths, allowing our model to learn stable view generation from a collection of single views. At test time, despite never having seen a video, our approach can take a single image and generate long camera trajectories comprised of hundreds of new views with realistic and diverse content. We compare our approach with recent state-of-the-art supervised view generation methods that require posed multi-view videos and demonstrate superior performance and synthesis quality. Our project webpage, including video results, is at infinite-nature-zero.github.io.



Paperid:31
Authors:Zhen Xing; Hengduo Li; Zuxuan Wu; Yu-Gang Jiang
Title: Semi-Supervised Single-View 3D Reconstruction via Prototype Shape Priors
Abstract:
The performance of existing single-view 3D reconstruction methods heavily relies on large-scale of 3D annotations. However, such annotations are tedious and expensive to collect. Semi-supervised learning serves as an alternative way to mitigate the need for manual labels, but remains unexplored in 3D reconstruction. Inspired by the recent success of self-ensembling method in semi-supervised image classification task, we first propose SSP3D, a semi-supervised framework for 3D reconstruction. In particular, we introduce an attention-guided prototype shape prior module for guiding realistic object reconstruction. we further introduce a discriminator-guided module to incentivize better shape generation, as well as a regularizer to tolerate noisy training samples. On the ShapeNet benchmark, the proposed approach outperforms previous supervised methods by clear margins margin under various labeling ratios, ( i.e., 1%, 5%, 10% and 20%). Moreover, our approach also performs well when transferring to real-world Pix3D datasets under labeling ratios of 10%. We also demonstrate our method could transfer to novel categories with few novel supervised data. Experiments on the popular ShapeNet dataset show that our method outperforms the zero-shot baseline by over 12% and the current state-of-the-art by over 7% in the few-shot setting.



Paperid:32
Authors:Xu Cao; Hiroaki Santo; Boxin Shi; Fumio Okura; Yasuyuki Matsushita
Title: Bilateral Normal Integration
Abstract:
This paper studies the discontinuity preservation problem in recovering a surface from its surface normal map. To model discontinuities, we introduce the assumption that the surface to be recovered is semi-smooth, i.e., the surface is one-sided differentiable (hence one-sided continuous) everywhere in the horizontal and vertical directions. Under the semi-smooth surface assumption, we propose a bilaterally weighted functional for discontinuity preserving normal integration. The key idea is to relatively weight the one-sided differentiability at each point’s two sides based on the definition of one-sided depth discontinuity. As a result, our method effectively preserves discontinuities and alleviates the under- or over-segmentation artifacts in the recovered surfaces compared to existing methods. Further, we unify the normal integration problem in the orthographic and perspective cases in a new way and show effective discontinuity preservation results in both cases.



Paperid:33
Authors:Tze Ho Elden Tse; Zhongqun Zhang; Kwang In Kim; Aleš Leonardis; Feng Zheng; Hyung Jin Chang
Title: S2Contact: Graph-Based Network for 3D Hand-Object Contact Estimation with Semi-Supervised Learning
Abstract:
Being able to reason about the physical contacts between hands and objects is crucial in understanding hand-object manipulation. However, despite the efforts in accurate 3D annotations in hand and object datasets, there still exist gaps in 3D hand and object reconstructions. Recent works leverage contact maps to refine inaccurate hand-object pose estimations and generate grasps given object models. However, they require explicit 3D supervision which is seldom available and therefore, are limited to constrained settings, e.g., where thermal cameras observe residual heat left on manipulated objects. In this paper, we propose a novel semi-supervised framework that allows us to learn contact from monocular videos. Specifically, we leverage visual and geometric consistency constraints in large-scale datasets for generating pseudo-labels in semi-supervised learning and propose an efficient graph-based network to infer contact. Our semi-supervised learning framework achieves a favourable improvement over the existing supervised learning methods trained on data with ‘limited’ annotations. Notably, our proposed model is able to achieve superior results with less than half the network parameters and memory access cost when compared with the commonly-used PointNet-based approach. We show benefits from using a contact map that rules hand-object interactions to produce more accurate reconstructions. We further demonstrate that training with pseudo-labels can extend contact map estimations to out-of-domain objects and generalise better across multiple datasets.



Paperid:34
Authors:Xin Wu; Hao Zhao; Shunkai Li; Yingdian Cao; Hongbin Zha
Title: SC-wLS: Towards Interpretable Feed-Forward Camera Re-localization
Abstract:
Visual re-localization aims to recover camera poses in a known environment, which is vital for applications like robotics or augmented reality. Feed-forward absolute camera pose regression methods directly output poses by a network, but suffer from low accuracy. Meanwhile, scene coordinate based methods are accurate, but need iterative RANSAC post-processing, which brings challenges to efficient end-to-end training and inference. In order to have the best of both worlds, we propose a feed-forward method termed SC-wLS that exploits all scene coordinate estimates for weighted least squares pose regression. This differentiable formulation exploits a weight network imposed on 2D-3D correspondences, and requires pose supervision only. Qualitative results demonstrate the interpretability of learned weights. Evaluations on 7Scenes and Cambridge datasets show significantly promoted performance when compared with former feed-forward counterparts. Moreover, our SC-wLS method enables a new capability: self-supervised test-time adaptation on the weight network. Codes and models are publicly available.



Paperid:35
Authors:Andreas Meuleman; Hakyeong Kim; James Tompkin; Min H. Kim
Title: FloatingFusion: Depth from ToF and Image-Stabilized Stereo Cameras
Abstract:
High-accuracy per-pixel depth is vital for computational photography, so smartphones now have multimodal camera systems with time-of-flight (ToF) depth sensors and multiple color cameras. However, producing accurate high-resolution depth is still challenging due to the low resolution and limited active illumination power of ToF sensors. Fusing RGB stereo and ToF information is a promising direction to overcome these issues, but a key problem remains: to provide high-quality 2D RGB images, the main smartphone color sensor’s lens is optically stabilized, resulting in an unknown pose for the floating lens that breaks the geometric relationships between the multimodal image sensors. Leveraging ToF depth estimates and a wide-angle RGB camera, we design an automatic calibration technique based on dense 2D/3D matching that can estimate camera pose intrinsic and distortion parameters of a stabilized main RGB sensor from a single snapshot. This lets us fuse stereo and ToF cues via a correlation volume. For fusion, we apply deep learning via a real-world training dataset with depth supervision estimated by a neural reconstruction method. For evaluation, we acquire a test dataset using a commercial high-power depth camera and show that our approach achieves higher accuracy than existing baselines.



Paperid:36
Authors:Yijin Li; Xinyang Liu; Wenqi Dong; Han Zhou; Hujun Bao; Guofeng Zhang; Yinda Zhang; Zhaopeng Cui
Title: DELTAR: Depth Estimation from a Light-Weight ToF Sensor and RGB Image
Abstract:
Light-weight time-of-flight (ToF) depth sensors are small, cheap, low-energy and have been massively deployed on mobile devices for the purposes like autofocus, obstacle detection, etc. However, due to their specific measurements (depth distribution in a region instead of the depth value at a certain pixel) and extremely low resolution, they are insufficient for applications requiring high-fidelity depth such as 3D reconstruction. In this paper, we propose DELTAR, a novel method to empower light-weight ToF sensors with the capability of measuring high resolution and accurate depth by cooperating with a color image. As the core of DELTAR, a feature extractor customized for depth distribution and an attention-based neural architecture is proposed to fuse the information from the color and ToF domain efficiently. To evaluate our system in real-world scenarios, we design a data collection device and propose a new approach to calibrate the RGB camera and ToF sensor. Experiments show that our method produces more accurate depth than existing frameworks designed for depth completion and depth super-resolution and achieves on par performance with a commodity-level RGB-D sensor. Code and data are available on the project webpage: https://zju3dv.github.io/deltar.



Paperid:37
Authors:Yining Zhao; Chao Wen; Zhou Xue; Yue Gao
Title: 3D Room Layout Estimation from a Cubemap of Panorama Image via Deep Manhattan Hough Transform
Abstract:
Significant geometric structures can be compactly described by global wireframes in the estimation of 3D room layout from a single panoramic image. Based on this observation, we present an alternative approach to estimate the walls in 3D space by modeling long-range geometric patterns in a learnable Hough Transform block. We transform the image feature from a cubemap tile to the Hough space of a Manhattan world and directly map the feature to the geometric output. The convolutional layers not only learn the local gradient-like line features, but also utilize the global information to successfully predict occluded walls with a simple network structure. Unlike most previous work, the predictions are performed individually on each cubemap tile, and then assembled to get the layout estimation. Experimental results show that we achieve comparable results with recent state-of-the-art in prediction accuracy and performance. Code is available at https://github.com/Starrah/DMH-Net.



Paperid:38
Authors:Ruida Zhang; Yan Di; Zhiqiang Lou; Fabian Manhardt; Federico Tombari; Xiangyang Ji
Title: RBP-Pose: Residual Bounding Box Projection for Category-Level Pose Estimation
Abstract:
Category-level object pose estimation aims to predict the 6D pose as well as the 3D metric size of previously unseen objects from a known set of categories. Recent methods harness shape prior adaptation to map the observed point cloud into the canonical space and apply Umeyama’s algorithm to recover the pose and size. However, their shape prior integration strategy boosts pose estimation indirectly, which leads to insufficient pose-sensitive feature extraction and slow inference speed. To tackle this problem, in this paper, we propose a novel geometry-guided Residual Object Bounding Box Projection network RBP-Pose that jointly predicts object pose and residual vectors describing the displacements from the shape-prior-indicated object surface projections on the bounding box towards real surface projections. Such definition of residual vectors is inherently zero-mean and relatively small, and explicitly encapsulates spatial cues of the 3D object for robust and accurate pose regression. We enforce geometry-aware consistency terms to align the predicted pose and residual vectors to further boost performance. Finally, to avoid overfitting and enhance the generalization ability of RBP-Pose, we propose an online non-linear shape augmentation scheme to promote shape diversity during training. Extensive experiments on NOCS datasets demonstrate that RBP-Pose surpasses all existing methods by a large margin, whilst achieving a real-time inference speed.



Paperid:39
Authors:Junzhe Zhang; Daxuan Ren; Zhongang Cai; Chai Kiat Yeo; Bo Dai; Chen Change Loy
Title: Monocular 3D Object Reconstruction with GAN Inversion
Abstract:
Recovering a textured 3D mesh from a monocular image is highly challenging, particularly for in-the-wild objects that lack 3D ground truths. In this work, we present MeshInversion, a novel framework to improve the reconstruction by exploiting the generative prior of a 3D GAN pre-trained for 3D textured mesh synthesis. Reconstruction is achieved by searching for a latent space in the 3D GAN that best resembles the target mesh in accordance with the single view observation. Since the pre-trained GAN encapsulates rich 3D semantics in terms of mesh geometry and texture, searching within the GAN manifold thus naturally regularizes the realness and fidelity of the reconstruction. Importantly, such regularization is directly applied in the 3D space, providing crucial guidance of mesh parts that are unobserved in the 2D space. Experiments on standard benchmarks show that our framework obtains faithful 3D reconstructions with consistent geometry and texture across both observed and unobserved parts. Moreover, it generalizes well to meshes that are less commonly seen, such as the extended articulation of deformable objects. Code is released at https://github.com/junzhezhang/mesh-inversion.



Paperid:40
Authors:Eduardo Arnold; Jamie Wynn; Sara Vicente; Guillermo Garcia-Hernando; Aron Monszpart; Victor Prisacariu; Daniyar Turmukhambetov; Eric Brachmann
Title: Map-Free Visual Relocalization: Metric Pose Relative to a Single Image
Abstract:
Can we relocalize in a scene represented by a single reference image? Standard visual relocalization requires hundreds of images and scale calibration to build a scene-specific 3D map. In contrast, we propose Map-free Relocalization, i.e., using only one photo of a scene to enable instant, metric scaled relocalization. Existing datasets are not suitable to benchmark map-free relocalization, due to their focus on large scenes or their limited variability. Thus, we have constructed a new dataset of 655 small places of interest, such as sculptures, murals and fountains, collected worldwide. Each place comes with a reference image to serve as a relocalization anchor, and dozens of query images with known, metric camera poses. The dataset features changing conditions, stark viewpoint changes, high variability across places, and queries with low to no visual overlap with the reference image.We identify two viable families of existing methods to provide baseline results: relative pose regression, and feature matching combined with single-image depth prediction. While these methods show reasonable performance on some favorable scenes in our dataset, map-free relocalization proves to be a challenge that requires new, innovative solutions.



Paperid:41
Authors:Zhengming Zhou; Qiulei Dong
Title: Self-Distilled Feature Aggregation for Self-Supervised Monocular Depth Estimation
Abstract:
Self-supervised monocular depth estimation has received much attention recently in computer vision. Most of the existing works in literature aggregate multi-scale features for depth prediction via either straightforward concatenation or element-wise addition, however, such feature aggregation operations generally neglect the contextual consistency between multi-scale features. Addressing this problem, we propose the Self-Distilled Feature Aggregation (SDFA) module for simultaneously aggregating a pair of low-scale and high-scale features and maintaining their contextual consistency. The SDFA employs three branches to learn three feature offset maps respectively: one offset map for refining the input low-scale feature and the other two for refining the input high-scale feature under a designed self-distillation manner. Then, we propose an SDFA-based network for self-supervised monocular depth estimation, and design a self-distilled training strategy to train the proposed network with the SDFA module. Experimental results on the KITTI dataset demonstrate that the proposed method outperforms the comparative state-of-the-art methods in most cases. The code is available at https://github.com/ZM-Zhou/SDFA-Net_pytorch.



Paperid:42
Authors:Zixuan Huang; Stefan Stojanov; Anh Thai; Varun Jampani; James M. Rehg
Title: Planes vs. Chairs: Category-Guided 3D Shape Learning without Any 3D Cues
Abstract:
We present a novel 3D shape reconstruction method which learns to predict an implicit 3D shape representation from a single RGB image. Our approach uses a set of single-view images of multiple object categories without viewpoint annotation, forcing the model to learn across multiple object categories without 3D supervision. To facilitate learning with such minimal supervision, we use category labels to guide shape learning with a novel categorical metric learning approach. We also utilize adversarial and viewpoint regularization techniques to further disentangle the effects of viewpoint and shape. We obtain the first results for large-scale (more than 50 categories) single-viewpoint shape prediction using a single model. We are also the first to examine and quantify the benefit of class information in single-view supervised 3D shape reconstruction. Our method achieves superior performance over state-of-the-art methods on ShapeNet-13, ShapeNet-55 and Pascal3D+.



Paperid:43
Authors:Haitian Zeng; Xin Yu; Jiaxu Miao; Yi Yang
Title: MHR-Net: Multiple-Hypothesis Reconstruction of Non-rigid Shapes from 2D Views
Abstract:
We propose MHR-Net, a novel method for recovering Non-Rigid Shapes from Motion (NRSfM). MHR-Net aims to find a set of reasonable reconstructions for a 2D view, and it also selects the most likely reconstruction from the set. To deal with the challenging unsupervised generation of non-rigid shapes, we develop a new Deterministic Basis and Stochastic Deformation scheme in MHR-Net. The non-rigid shape is first expressed as the sum of a coarse shape basis and a flexible shape deformation, then multiple hypotheses are generated with uncertainty modeling of the deformation part. MHR-Net is optimized with reprojection loss on the basis and the best hypothesis. Furthermore, we design a new Procrustean Residual Loss, which reduces the rigid rotations between similar shapes and further improves the performance. Experiments show that MHR-Net achieves state-of-the-art reconstruction accuracy on Human3.6M, SURREAL and 300-VW datasets.



Paperid:44
Authors:Jinyoung Jun; Jae-Han Lee; Chul Lee; Chang-Su Kim
Title: Depth Map Decomposition for Monocular Depth Estimation
Abstract:
We propose a novel algorithm for monocular depth estimation that decomposes a metric depth map into a normalized depth map and scale features. The proposed network is composed of a shared encoder and three decoders, called G-Net, N-Net, and M-Net, which estimate gradient maps, a normalized depth map, and a metric depth map, respectively. M-Net learns to estimate metric depths more accurately using relative depth features extracted by G-Net and N-Net. The proposed algorithm has the advantage that it can use datasets without metric depth labels to improve the performance of metric depth estimation. Experimental results on various datasets demonstrate that the proposed algorithm not only provides competitive performance to state-of-the-art algorithms but also yields acceptable results even when only a small amount of metric depth data is available for its training.



Paperid:45
Authors:Tian Yu Liu; Parth Agrawal; Allison Chen; Byung-Woo Hong; Alex Wong
Title: Monitored Distillation for Positive Congruent Depth Completion
Abstract:
We propose a method to infer a dense depth map from a single image, its calibration, and the associated sparse point cloud. In order to leverage existing models (teachers) that produce putative depth maps, we propose an adaptive knowledge distillation approach that yields a positive congruent training process, wherein a student model avoids learning the error modes of the teachers. In the absence of ground truth for model selection and training, our method, termed Monitored Distillation, allows a student to exploit a blind ensemble of teachers by selectively learning from predictions that best minimize the reconstruction error for a given image. Monitored Distillation yields a distilled depth map and a confidence map, or “monitor"""", for how well a prediction from a particular teacher fits the observed image. The monitor adaptively weights the distilled depth where if all of the teachers exhibit high residuals, the standard unsupervised image reconstruction loss takes over as the supervisory signal. On indoor scenes (VOID), we outperform blind ensembling baselines by 17.53% and unsupervised methods by 24.25%; we boast a 79% model size reduction while maintaining comparable performance to the best supervised method. For outdoors (KITTI), we tie for 5th overall on the benchmark despite not using ground truth. Code available at: https://github.com/alexklwong/mondi-python.



Paperid:46
Authors:Tianxin Huang; Jiangning Zhang; Jun Chen; Yuang Liu; Yong Liu
Title: Resolution-Free Point Cloud Sampling Network with Data Distillation
Abstract:
Down-sampling algorithms are adopted to simplify the point clouds and save the computation cost on subsequent tasks. Existing learning-based sampling methods often need to train a big sampling network to support sampling under different resolutions, which must generate sampled points with the costly maximum resolution even if only low-resolution points need to be sampled. In this work, we propose a novel resolution-free point clouds sampling network to directly sample the original point cloud to different resolutions, which is conducted by optimizing non-learning-based initial sampled points to better positions. Besides, we introduce data distillation to assist the training process by considering the differences between task network outputs from original point clouds and sampled points. Experiments on point cloud reconstruction and recognition tasks demonstrate that our method can achieve SOTA performances with lower time and memory cost than existing learning-based sampling strategies. Codes are available at https://github.com/Tianxinhuang/PCDNet.



Paperid:47
Authors:Suryansh Kumar; Luc Van Gool
Title: Organic Priors in Non-rigid Structure from Motion
Abstract:
This paper advocates the use of organic priors in classical non-rigid structure from motion (NRSfM). By organic priors, we mean invaluable intermediate prior information intrinsic to the NRSfM matrix factorization theory. It is shown that such priors reside in the factorized matrices, and quite surprisingly, existing methods generally disregard them. The paper’s main contribution is to put forward a simple, methodical, and practical method that can effectively exploit such organic priors to solve NRSfM. The proposed method does not make assumptions other than the popular one on the low-rank shape and offers a reliable solution to NRSfM under orthographic projection. Our work reveals that the accessibility of organic priors is independent of the camera motion and shape deformation type. Besides that, the paper provides insights into the NRSfM factorization---both in terms of shape and motion---and is the first approach to show the benefit of single rotation averaging for NRSfM. Furthermore, we outline how to effectively recover motion and non-rigid 3D shape using the proposed organic prior based approach and demonstrate results that outperform prior-free NRSfM performance by a significant margin. Finally, we present the benefits of our method via extensive experiments and evaluations on several benchmark datasets.



Paperid:48
Authors:Yinlin Hu; Pascal Fua; Mathieu Salzmann
Title: Perspective Flow Aggregation for Data-Limited 6D Object Pose Estimation
Abstract:
Most recent 6D object pose estimation methods, including unsupervised ones, require many real training images. Unfortunately, for some applications, such as those in space or deep under water, acquiring real images, even unannotated, is virtually impossible. In this paper, we propose a method that can be trained solely on synthetic images, or optionally using a few additional real ones. Given a rough pose estimate obtained from a first network, it uses a second network to predict a dense 2D correspondence field between the image rendered using the rough pose and the real image and infers the required pose correction. This approach is much less sensitive to the domain shift between synthetic and real images than state-of-the-art methods. It performs on par with methods that require annotated real images for training when not using any, and outperforms them considerably when using as few as twenty real images.



Paperid:49
Authors:Shih-Yang Su; Timur Bagautdinov; Helge Rhodin
Title: DANBO: Disentangled Articulated Neural Body Representations via Graph Neural Networks
Abstract:
Deep learning greatly improved the realism of animatable human models by learning geometry and appearance from collections of 3D scans, template meshes, and multi-view imagery. High-resolution models enable photo-realistic avatars but at the cost of requiring studio settings not available to end users. Our goal is to create avatars directly from raw images without relying on expensive studio setups and surface tracking. While a few such approaches exist, those have limited generalization capabilities and are prone to learning spurious (chance) correlations between irrelevant body parts, resulting in implausible deformations and missing body parts on unseen poses. We introduce a three-stage method that induces two inductive biases to better disentangled pose-dependent deformation. First, we model correlations of body parts explicitly with a graph neural network. Second, to further reduce the effect of chance correlations, we introduce localized per-bone features that use a factorized volumetric representation and a new aggregation function. We demonstrate that our model produces realistic body shapes under challenging unseen poses and shows high-quality image synthesis. Our proposed representation strikes a better trade-off between model capacity, expressiveness, and robustness than competing methods. Project website: https://lemonatsu.github.io/danbo.



Paperid:50
Authors:Xianghui Xie; Bharat Lal Bhatnagar; Gerard Pons-Moll
Title: "CHORE: Contact, Human and Object REconstruction from a Single RGB Image"
Abstract:
Most prior works in perceiving 3D humans from images reason human in isolation without their surroundings. However, humans are constantly interacting with the surrounding objects, thus calling for models that can reason about not only the human but also the object and their interaction. The problem is extremely challenging due to heavy occlusions between humans and objects, diverse interaction types and depth ambiguity. In this paper, we introduce CHORE, a novel method that learns to jointly reconstruct the human and the object from a single RGB image. CHORE takes inspiration from recent advances in implicit surface learning and classical model-based fitting. We compute a neural reconstruction of human and object represented implicitly with two unsigned distance fields, a correspondence field to a parametric body and an object pose field. This allows us to robustly fit a parametric body model and a 3D object template, while reasoning about interactions. Furthermore, prior pixel-aligned implicit learning methods use synthetic data and make assumptions that are not met in the real data. We propose a elegant depth-aware scaling that allows more efficient shape learning on real data. Experiments show that our joint reconstruction learned with the proposed strategy significantly outperforms the SOTA. Our code and models are available at https://virtualhumans.mpi-inf.mpg.de/chore"



Paperid:51
Authors:Enric Corona; Gerard Pons-Moll; Guillem Alenyà; Francesc Moreno-Noguer
Title: Learned Vertex Descent: A New Direction for 3D Human Model Fitting
Abstract:
We propose a novel optimization-based paradigm for 3D human shape fitting on images. In contrast to existing approaches that directly regress the parameters of a low-dimensional statistical body model (e.g. SMPL) from input images, we propose training a deep network that, given solely image features and an unfit mesh, predicts the directions of the vertices towards the 3D body mesh. At inference, we employ this network, dubbed LVD, within a gradient-descent optimization pipeline until its convergence, which typically occurs in a fraction of a second even when initializing all vertices into a single point. An exhaustive evaluation demonstrates that our approach is able to capture the underlying body of clothed people with very different body shapes, achieving a significant improvement compared to state-of-the-art. Additionally, the proposed formulation can generalize to other sources of input data, which we experimentally show on fitting 3D scans of full bodies and hands.



Paperid:52
Authors:Junxuan Li; Hongdong Li
Title: Self-Calibrating Photometric Stereo by Neural Inverse Rendering
Abstract:
This paper tackles the task of uncalibrated photometric stereo for 3D object reconstruction, where both the object shape, object reflectance, and lighting directions are unknown. This is an extremely difficult task, and the challenge is further compounded with the existence of the well-known generalized bas-relief (GBR) ambiguity in photometric stereo. Previous methods to resolve this ambiguity either rely on an overly simplified reflectance model, or assume special light distribution. We propose a new method that jointly optimizes object shape, light directions, and light intensities, all under general surfaces and lights assumptions. The specularities are used explicitly to resolve the GBR ambiguity via a neural inverse rendering process. We gradually fit specularities from shiny to rough using novel progressive specular bases. Our method leverages a physically based rendering equation by minimizing the reconstruction error on a per-object-basis. Our method demonstrates state-of-the-art accuracy in light estimation and shape recovery on real-world datasets.



Paperid:53
Authors:Gyeongsik Moon; Hyeongjin Nam; Takaaki Shiratori; Kyoung Mu Lee
Title: 3D Clothed Human Reconstruction in the Wild
Abstract:
Although much progress has been made in 3D clothed human reconstruction, most of the existing methods fail to produce robust results from in-the-wild images, which contain diverse human poses and appearances. This is mainly due to the large domain gap between training datasets and in-the-wild datasets. The training datasets are usually synthetic ones, which contain rendered images from GT 3D scans. However, such datasets contain simple human poses and less natural image appearances compared to those of real in-the-wild datasets, which makes generalization of it to in-the-wild images extremely challenging. To resolve this issue, in this work, we propose ClothWild, a 3D clothed human reconstruction framework that firstly addresses the robustness on in-thewild images. First, for the robustness to the domain gap, we propose a weakly supervised pipeline that is trainable with 2D supervision targets of in-the-wild datasets. Second, we design a DensePose-based loss function to reduce ambiguities of the weak supervision. Extensive empirical tests on several public in-the-wild datasets demonstrate that our proposed ClothWild produces much more accurate and robust results than the state-of-the-art methods. The codes are available in here: https://github.com/hygenie1228/ClothWild_RELEASE.



Paperid:54
Authors:Nilesh Kulkarni; Justin Johnson; David F. Fouhey
Title: Directed Ray Distance Functions for 3D Scene Reconstruction
Abstract:
We present an approach for full 3D scene reconstruction from a single new image that can be trained on realistic non-watertight scans. Our approach uses a predicted distance function, since these have shown promise in handling complex topologies and large spaces. We identify and analyze two key challenges for predicting these implicit functions from an image that have prevented their success on 3D scenes from a single image. First, we show that predicting a conventional scene distance from an image requires reasoning over a large receptive field. Second, we analytically show that the optimal output of a network that predicts these distance functions is often not a distance function. We propose an alternate approach, the Direct Ray Distance Function (DRDF), that avoids both challenges. We show that a deep network trained to predict DRDFs outperforms all other methods quantitatively and qualitatively on 3D reconstruction on Matterport3D, 3DFront, and ScanNet"



Paperid:55
Authors:Zhaoxin Fan; Zhenbo Song; Jian Xu; Zhicheng Wang; Kejian Wu; Hongyan Liu; Jun He
Title: Object Level Depth Reconstruction for Category Level 6D Object Pose Estimation from Monocular RGB Image
Abstract:
Recently, RGBD-based category-level 6D object pose estimation has achieved promising improvement in performance, however, the requirement of depth information prohibits broader applications. In order to relieve this problem, this paper proposes a novel approach named Object Level Depth reconstruction Network (OLD-Net) taking only RGB images as input for category-level 6D object pose estimation. We propose to directly predict object-level depth from a monocular RGB image by deforming the category-level shape prior into object-level depth and the canonical NOCS representation. Two novel modules named Normalized Global Position Hints (NGPH) and Shape-aware Decoupled Depth Reconstruction (SDDR) module are introduced to learn high fidelity object-level depth and delicate shape representations. At last, the 6D object pose is solved by aligning the predicted canonical representation with the back-projected object-level depth. Extensive experiments on the challenging CAMERA25 and REAL275 datasets indicate that our model, though simple, achieves state-of-the-art performance.



Paperid:56
Authors:Dongting Hu; Liuhua Peng; Tingjin Chu; Xiaoxing Zhang; Yinian Mao; Howard Bondell; Mingming Gong
Title: Uncertainty Quantification in Depth Estimation via Constrained Ordinal Regression
Abstract:
Monocular Depth Estimation (MDE) is a task to predict a dense depth map from a single image. Despite the recent progress brought by deep learning, existing methods are still prone to errors due to the ill-posed nature of MDE. Hence depth estimation systems must be self-aware of possible mistakes to avoid disastrous consequences. This paper provides an uncertainty quantification method for supervised MDE models. From a frequentist view, we capture the uncertainty by predictive variance that consists of two terms: error variance and estimation variance. The former represents the noise of a depth value, and the latter measures the randomness in the depth regression model due to training on finite data. To estimate error variance, we perform constrained ordinal regression (ConOR) on discretized depth to estimate the conditional distribution of depth given image, and then compute the corresponding conditional mean and variance as the predicted depth and error variance estimator, respectively. Our work also leverages bootstrapping methods to infer estimation variance from re-sampled data. We perform experiments on both simulated and real data to validate the effectiveness of the proposed method. The results show that our approach produces accurate uncertainty estimates while maintaining high depth prediction accuracy.



Paperid:57
Authors:Jaewon Kam; Jungeon Kim; Soongjin Kim; Jaesik Park; Seungyong Lee
Title: CostDCNet: Cost Volume Based Depth Completion for a Single RGB-D Image
Abstract:
Successful depth completion from a single RGB-D image requires both extracting plentiful 2D and 3D features and merging these heterogeneous features appropriately. We propose a novel depth completion framework, CostDCNet, based on the cost volume-based depth estimation approach that has been successfully employed for multi-view stereo (MVS). The key to high-quality depth map estimation in the approach is constructing an accurate cost volume. To produce a quality cost volume tailored to single-view depth completion, we present a simple but effective architecture that can fully exploit the 3D information, three options to make an RGB-D feature volume, and per-plane pixel shuffle for efficient volume upsampling. Our CostDCNet framework consists of lightweight deep neural networks ( 1.8M parameters), running in real time ( 30ms). Nevertheless, thanks to our simple but effective design, CostDCNet demonstrates depth completion results comparable to or better than the state-of-the-art methods.



Paperid:58
Authors:Muhammad Zubair Irshad; Sergey Zakharov; Rareș Ambruș; Thomas Kollar; Zsolt Kira; Adrien Gaidon
Title: "ShAPO: Implicit Representations for Multi-Object Shape, Appearance, and Pose Optimization"
Abstract:
Our method studies the complex task of object-centric 3D understanding from a single RGB-D observation. As it is an ill-posed problem, existing methods suffer from low performance for both 3D shape and 6D pose and size estimation in complex multi-object scenarios with occlusions. We present ShAPO, a method for joint multi-object detection, 3D textured reconstruction, 6D object pose and size estimation. Key to ShAPO is a single-shot pipeline to regress shape, appearance and pose latent codes along with the masks of each object instance, which is then further refined in a sparse-to-dense fashion. A novel disentangled shape and appearance database of priors is first learned to embed objects in their respective shape and appearance space. We also propose a novel, octree-based differentiable optimization step, allowing us to further improve object shape, pose and appearance simultaneously under the learned latent space, in an analysis-by synthesis fashion. Our novel joint implicit textured object representation allows us to accurately identify and reconstruct novel unseen objects without having access to their 3D meshes. Through extensive experiments, we show that our method, trained on simulated indoor scenes, accurately regresses the shape, appearance and pose of novel objects in the real-world with minimal fine-tuning. Our method significantly out-performs all baselines on the NOCS dataset with an 8% absolute improvement in mAP for 6D pose estimation.



Paperid:59
Authors:Le Hui; Lingpeng Wang; Linghua Tang; Kaihao Lan; Jin Xie; Jian Yang
Title: 3D Siamese Transformer Network for Single Object Tracking on Point Clouds
Abstract:
Siamese network based trackers formulate 3D single object tracking as cross-correlation learning between point features of a template and a search area. Due to the large appearance variation between the template and search area during tracking, how to learn the robust cross correlation between them for identifying the potential target in the search area is still a challenging problem. In this paper, we explicitly use Transformer to form a 3D Siamese Transformer network for learning robust cross correlation between the template and the search area of point clouds. Specifically, we develop a Siamese point Transformer network to learn shape context information of the target. Its encoder uses self-attention to capture non-local information of point clouds to characterize the shape information of the object, and the decoder utilizes cross-attention to upsample discriminative point features. After that, we develop an iterative coarse-to-fine correlation network to learn the robust cross correlation between the template and the search area. It formulates the cross-feature augmentation to associate the template with the potential target in the search area via cross attention. To further enhance the potential target, it employs the ego-feature augmentation that applies self-attention to the local k-NN graph of the feature space to aggregate target features. Experiments on the KITTI, nuScenes, and Waymo datasets show that our method achieves state-of-the-art performance on the 3D single object tracking task.



Paperid:60
Authors:Ji Yang; Xinxin Zuo; Sen Wang; Zhenbo Yu; Xingyu Li; Bingbing Ni; Minglun Gong; Li Cheng
Title: Object Wake-Up: 3D Object Rigging from a Single Image
Abstract:
Given a single chair image, could we wake it up by reconstructing its 3D shape and skeleton, as well as animating its plausible articulations and motions, similar to that of human modeling? It is a new problem that not only goes beyond image-based object reconstruction but also involves articulated animation of generic objects in 3D, which could give rise to numerous downstream augmented and virtual reality applications. In this paper, we propose an automated approach to tackle the entire process of reconstruct such generic 3D objects, rigging and animation, all from single images. A two-stage pipeline has thus been proposed, which specifically contains a multi-head structure to utilize the deep implicit functions for skeleton prediction. Two in-house 3D datasets of generic objects with high-fidelity rendering and annotated skeletons have also been constructed. Empirically our approach demonstrated promising results; when evaluated on the related sub-tasks of 3D reconstruction and skeleton prediction, our results surpass those of the state-of-the-arts by a noticeable margin. Our code and datasets are made publicly available at the dedicated project website.



Paperid:61
Authors:Kennard Yanting Chan; Guosheng Lin; Haiyu Zhao; Weisi Lin
Title: IntegratedPIFu: Integrated Pixel Aligned Implicit Function for Single-View Human Reconstruction
Abstract:
We propose IntegratedPIFu, a new pixel-aligned implicit model that builds on the foundation set by PIFuHD. IntegratedPIFu shows how depth and human parsing information can be predicted and capitalized upon in a pixel-aligned implicit model. In addition, IntegratedPIFu introduces depth-oriented sampling, a novel training scheme that improve any pixel-aligned implicit model’s ability to reconstruct important human features without noisy artefacts. Lastly, IntegratedPIFu presents a new architecture that, despite using less model parameters than PIFuHD, is able to improves the structural correctness of reconstructed meshes. Our results show that IntegratedPIFu significantly outperforms existing state-of-the-arts methods on single-view human reconstruction. We provide the code in our supplementary materials. Our code is available at https://github.com/kcyt/IntegratedPIFu.



Paperid:62
Authors:Taras Khakhulin; Vanessa Sklyarova; Victor Lempitsky; Egor Zakharov
Title: Realistic One-Shot Mesh-Based Head Avatars
Abstract:
We present a system for the creation of realistic one-shot mesh-based (ROME) human head avatars. From a single photograph, our system estimates the head mesh (with person-specific details in both the facial and non-facial head parts) as well as the neural texture encoding local photometric and geometric details. The resulting avatars are rigged and can be rendered using a deep rendering network, which is trained alongside the mesh and texture estimators on a dataset of in-the-wild videos. In the experiments, we observe that our system performs competitively both in terms of head geometry recovery and the quality of renders, especially for strong pose and expression changes.



Paperid:63
Authors:Martha Paskin; Daniel Baum; Mason N. Dean; Christoph von Tycowicz
Title: A Kendall Shape Space Approach to 3D Shape Estimation from 2D Landmarks
Abstract:
3D shapes provide substantially more information than 2D images. However, the acquisition of 3D shapes is sometimes very difficult or even impossible in comparison with acquiring 2D images, making it necessary to derive the 3D shape from 2D images. Although this is, in general, a mathematically ill-posed problem, it might be solved by constraining the problem formulation using prior information. Here, we present a new approach based on Kendall’s shape space to reconstruct 3D shapes from single monocular 2D images. The work is motivated by an application to study the feeding behavior of the basking shark, an endangered species whose massive size and mobility render 3D shape data nearly impossible to obtain, hampering understanding of their feeding behaviors and ecology. 2D images of these animals in feeding position, however, are readily available. We compare our approach with state-of-the-art shape-based approaches, both on human stick models and on shark head skeletons. Using a small set of training shapes, we show that the Kendall shape space approach is substantially more robust than previous methods and results in plausible shapes. This is essential for the motivating application in which specimens are rare and therefore only few training shapes are available.



Paperid:64
Authors:Zian Wang; Wenzheng Chen; David Acuna; Jan Kautz; Sanja Fidler
Title: Neural Light Field Estimation for Street Scenes with Differentiable Virtual Object Insertion
Abstract:
We consider the challenging problem of outdoor lighting estimation for the goal of photorealistic virtual object insertion into photographs. Existing works on outdoor lighting estimation typically simplify the scene lighting into an environment map which cannot capture the spatially-varying lighting effects in outdoor scenes. In this work, we propose a neural approach that estimates the 5D HDR light field from a single image, and a differentiable object insertion formulation that enables end-to-end training with image-based losses that encourage realism. Specifically, we design a hybrid lighting representation tailored to outdoor scenes, which contains an HDR sky dome that handles the extreme intensity of the sun, and a volumetric lighting representation that models the spatially-varying appearance of the surrounding scene. With the estimated lighting, our shadow-aware object insertion is fully differentiable, which enables adversarial training over the composited image to provide additional supervisory signal to the lighting prediction. We experimentally demonstrate that our hybrid lighting representation is more performant than existing outdoor lighting estimation methods. We further show the benefits of our AR object insertion in an autonomous driving application, where we obtain performance gains for a 3D object detector when trained on our augmented data.



Paperid:65
Authors:Guangcheng Chen; Li He; Yisheng Guan; Hong Zhang
Title: Perspective Phase Angle Model for Polarimetric 3D Reconstruction
Abstract:
Current polarimetric 3D reconstruction methods, including those in the well-established shape from polarization literature, are all developed under the orthographic projection assumption. In the case of a large field of view, however, this assumption does not hold and may result in significant reconstruction errors in methods that make this assumption. To address this problem, we present the perspective phase angle (PPA) model that is applicable to perspective cameras. Compared with the orthographic model, the proposed PPA model accurately describes the relationship between polarization phase angle and surface normal under perspective projection. In addition, the PPA model makes it possible to estimate surface normals from only one single-view phase angle map and does not suffer from the so-called π-ambiguity problem. Experiments on real data show that the PPA model is more accurate for surface normal estimation with a perspective camera than the orthographic model.



Paperid:66
Authors:Asaf Karnieli; Ohad Fried; Yacov Hel-Or
Title: DeepShadow: Neural Shape from Shadow
Abstract:
This paper presents ‘DeepShadow’, a one-shot method for recovering the depth map and surface normals from photometric stereo shadow maps. Previous works that try to recover the surface normals from photometric stereo images treat cast shadows as a disturbance. We show that the self and cast shadows not only do not disturb 3D reconstruction, but can be used alone, as a strong learning signal, to recover the depth map and surface normals. We demonstrate that 3D reconstruction from shadows can even outperform shape-from-shading in certain cases. To the best of our knowledge, our method is the first to reconstruct 3D shape-from-shadows using neural networks. The method does not require any pre-training or expensive labeled data, and is optimized during inference time.



Paperid:67
Authors:Yu Liu; Hui Zhang
Title: Camera Auto-Calibration from the Steiner Conic of the Fundamental Matrix
Abstract:
This paper addresses the problem of camera auto-calibration from the fundamental matrix under general motion. The fundamental matrix can be decomposed into a symmetric part (a Steiner conic) and a skew-symmetric part (a fixed point), which we find useful for fully calibrating camera parameters. We first obtain a fixed line from the image of the symmetric, skew-symmetric parts of the fundamental matrix and the image of the absolute conic. Then the properties of this fixed line are presented and proved, from which new constraints on general eigenvectors between the Steiner conic and the image of the absolute conic are derived. We thus propose a method to fully calibrate the camera. First, the three camera intrinsic parameters, i.e., the two focal lengths and the skew, can be solved from our new constraints on the imaged absolute conic obtained from at least three images. On this basis, we can initialize and then iteratively restore the optimal pair of projection centers of the Steiner conic, thereby obtaining the corresponding vanishing lines and images of circular points. Finally, all five camera parameters are fully calibrated using images of circular points obtained from at least three images. Experimental results on synthetic and real data demonstrate that our method achieves state-of-the-art performance in terms of accuracy.



Paperid:68
Authors:Marco Pesavento; Marco Volino; Adrian Hilton
Title: Super-Resolution 3D Human Shape from a Single Low-Resolution Image
Abstract:
We propose a novel framework to reconstruct super-resolution human shape from a single low-resolution input image. The approach overcomes limitations of existing approaches that reconstruct 3D human shape from a single image, which require high-resolution images together with auxiliary data such as surface normal or a parametric model to reconstruct high-detail shape. The proposed framework represents the reconstructed shape with a high-detail implicit function. Analogous to the objective of 2D image super-resolution, the approach learns the mapping from a low-resolution shape to its high-resolution counterpart and it is applied to reconstruct 3D shape detail from low-resolution images. The approach is trained end-to-end employing a novel loss function which estimates the information lost between a low and high-resolution representation of the same 3D surface shape. Evaluation for single image reconstruction of clothed people demonstrates that our method achieves high-detail surface reconstruction from low-resolution images without auxiliary data. Extensive experiments show that the proposed approach can estimate super-resolution human geometries with a significantly higher level of detail than that obtained with previous approaches when applied to low-resolution images.



Paperid:69
Authors:Weng Fei Low; Gim Hee Lee
Title: Minimal Neural Atlas: Parameterizing Complex Surfaces with Minimal Charts and Distortion
Abstract:
Explicit neural surface representations allow for exact and efficient extraction of the encoded surface at arbitrary precision, as well as analytic derivation of differential geometric properties such as surface normal and curvature. Such desirable properties, which are absent in its implicit counterpart, makes it ideal for various applications in computer vision, graphics and robotics. However, SOTA works are limited in terms of the topology it can effectively describe, distortion it introduces to reconstruct complex surfaces and model efficiency. In this work, we present Minimal Neural Atlas, a novel atlas-based explicit neural surface representation. At its core is a fully learnable parametric domain, given by an implicit probabilistic occupancy field defined on an open square of the parametric space. In contrast, prior works generally predefine the parametric domain. The added flexibility enables charts to admit arbitrary topology and boundary. Thus, our representation can learn a minimal atlas of 3 charts with distortion-minimal parameterization for surfaces of arbitrary topology, including closed and open surfaces with arbitrary connected components. Our experiments support the hypotheses and show that our reconstructions are more accurate in terms of the overall geometry, due to the separation of concerns on topology and geometry.



Paperid:70
Authors:Daxuan Ren; Jianmin Zheng; Jianfei Cai; Jiatong Li; Junzhe Zhang
Title: ExtrudeNet: Unsupervised Inverse Sketch-and-Extrude for Shape Parsing
Abstract:
Sketch-and-extrude is a common and intuitive modeling process in computer aided design. This paper studies the problem of learning the shape given in the form of point clouds by “inverse” sketch-and-extrude. We present ExtrudeNet, an unsupervised end-to-end network for discovering sketch and extrude from point clouds. Behind ExtrudeNet are two new technical components: 1) the use of a specially-designed rational Bézier representation for sketch and extrude, which can model extrusion with freeform sketches and conventional cylinder and box primitives as well; and 2) a numerical method for computing the signed distance field which is used in the network learning. This is the first attempt that uses machine learning to reverse engineer the sketch-and-extrude modeling process of a shape in an unsupervised fashion. ExtrudeNet not only outputs a compact, editable and interpretable representation of the shape that can be seamlessly integrated into modern CAD software, but also aligns with the standard CAD modeling process facilitating various editing applications, which distinguishes our work from existing shape parsing research. Code will be open-sourced upon acceptance.



Paperid:71
Authors:Xingyu Liu; Gu Wang; Yi Li; Xiangyang Ji
Title: CATRE: Iterative Point Clouds Alignment for Category-Level Object Pose Refinement
Abstract:
While category-level 9DoF object pose estimation has emerged recently, previous correspondence-based or direct regression methods are both limited in accuracy due to the huge intra-category variances in object shape and color, etc. Orthogonal to them, this work presents a category-level object pose and size refiner CATRE, which is able to iteratively enhance pose estimate from point clouds to produce accurate results. Given an initial pose estimate, CATRE predicts a relative transformation between the initial pose and ground truth by means of aligning the partially observed point cloud and an abstract shape prior. In specific, we propose a novel disentangled architecture being aware of the inherent distinctions between rotation and translation/size estimation. Extensive experiments show that our approach remarkably outperforms state-of-the-art methods on REAL275, CAMERA25, and LM benchmarks up to a speed of approximately 85.32Hz, and achieves competitive results on category-level tracking. We further demonstrate that CATRE can perform pose refinement on unseen category. Code and trained models are available.



Paperid:72
Authors:Jingyu Gong; Fengqi Liu; Jiachen Xu; Min Wang; Xin Tan; Zhizhong Zhang; Ran Yi; Haichuan Song; Yuan Xie; Lizhuang Ma
Title: Optimization over Disentangled Encoding: Unsupervised Cross-Domain Point Cloud Completion via Occlusion Factor Manipulation
Abstract:
Recently, studies considering domain gaps in shape completion attracted more attention, due to the undesirable performance of supervised methods on real scans. They only noticed the gap in input scans, but ignored the gap in output prediction, which is specific for completion. In this paper, we disentangle partial scans into three (domain, shape, and occlusion) factors to handle the output gap in cross-domain completion. For factor learning, we design view-point prediction and domain classification tasks in a self-supervised manner and bring a factor permutation consistency regularization to ensure factor independence. Thus, scans can be completed by simply manipulating occlusion factors while preserving domain and shape information. To further adapt to instances in the target domain, we introduce an optimization stage to maximize the consistency between completed shapes and input scans. Extensive experiments on real scans and synthetic datasets show that ours outperforms previous methods by a large margin and is encouraging for the following works. Code is available at https://github.com/azuki-miho/OptDE.



Paperid:73
Authors:Haocheng Yuan; Chen Zhao; Shichao Fan; Jiaxi Jiang; Jiaqi Yang
Title: Unsupervised Learning of 3D Semantic Keypoints with Mutual Reconstruction
Abstract:
Semantic 3D keypoints are category-level semantic consistent points on 3D objects. Detecting 3D semantic keypoints is a foundation for a number of 3D vision tasks but remains challenging, due to the ambiguity of semantic information, especially when the objects are represented by unordered 3D point clouds. Existing unsupervised methods tend to generate category-level keypoints in implicit manners, making it difficult to extract high-level information, such as semantic labels and topology. From a novel mutual reconstruction perspective, we present an unsupervised method to generate consistent semantic keypoints from point clouds explicitly. To achieve this, we train our unsupervised model to reconstruct both the input object and other objects from the same category based on predicted keypoints. To the best of our knowledge, the proposed method is the first to mine 3D semantic consistent keypoints from a mutual reconstruction view. Experiments under various evaluation metrics as well as comparisons with the state-of-the-arts have verified the efficacy of our new solution to mining semantic consistent keypoints with mutual reconstruction.



Paperid:74
Authors:Gopal Sharma; Kangxue Yin; Subhransu Maji; Evangelos Kalogerakis; Or Litany; Sanja Fidler
Title: MvDeCor: Multi-View Dense Correspondence Learning for Fine-Grained 3D Segmentation
Abstract:
We propose to utilize self-supervised techniques in the 2D domain for fine-grained 3D shape segmentation tasks. This is inspired by the observation that view-based surface representations are more effective at modeling high-resolution surface details and texture than their 3D counterparts based on point clouds or voxel occupancy. Specifically, given a 3D shape, we render it from multiple views, and set up a dense correspondence learning task within the contrastive learning framework. As a result, the learned 2D representations are view-invariant and geometrically consistent, leading to better generalization when trained on a limited number of labeled shapes than alternatives based on self-supervision in 2D or 3D alone. Experiments on textured (RenderPeople) and untextured (PartNet) 3D datasets show that our method outperforms state-of-the-art alternatives in fine-grained part segmentation. The improvements over baselines are greater when only a sparse set of views is available for training or when shapes are textured, indicating that \mvd benefits from both 2D processing and 3D geometric reasoning.



Paperid:75
Authors:Ahmed A. A. Osman; Timo Bolkart; Dimitrios Tzionas; Michael J. Black
Title: SUPR: A Sparse Unified Part-Based Human Representation
Abstract:
Statistical 3D shape models of the head, hands, and full body are widely used in computer vision and graphics. Despite their wide use, we show that existing models of the head and hands fail to capture the full range of motion for these parts. Moreover, existing work largely ignores the feet, which are crucial for modeling human movement and have applications in biomechanics, animation, and the footwear industry. The problem is that previous body part models are trained using 3D scans that are isolated to the individual parts. Such data does not capture the full range of motion for such parts, e.g. the motion of head relative to the neck. Our observation is that full-body scans provide important information about the motion of the body parts. Consequently, we propose a new learning scheme that jointly trains a full-body model and specific part models using a federated dataset of full-body and body-part scans. Specifically, we train an expressive human body model called SUPR (Sparse Unified Part-Based Representation), where each joint strictly influences a sparse set of model vertices. The factorized representation enables separating SUPR into an entire suite of body part models: an expressive head (SUPR-Head), an articulated hand (SUPR-Hand), and a novel foot (SUPR-Foot). Note that feet have received little attention and existing 3D body models have highly under-actuated feet. Using novel 4D scans of feet, we train a model with an extended kinematic tree that captures the range of motion of the toes. Additionally, feet deform due to ground contact. To model this, we include a novel non-linear deformation function that predicts foot deformation conditioned on the foot pose, shape, and ground contact. We train SUPR on an unprecedented number of scans: 1.2 million body, head, hand and foot scans. We quantitatively compare SUPR and the separate body parts to existing expressive human body models and body-part models and find that our suite of models generalizes better and captures the body parts’ full range of motion. SUPR is publicly available for research purposes.



Paperid:76
Authors:Rolandos Alexandros Potamias; Giorgos Bouritsas; Stefanos Zafeiriou
Title: Revisiting Point Cloud Simplification: A Learnable Feature Preserving Approach
Abstract:
The recent advances in 3D sensing technology have made possible the capture of point clouds in significantly high resolution. However, increased detail usually comes at the expense of high storage, as well as computational costs in terms of processing and visualization operations. Mesh and Point Cloud simplification methods aim to reduce the complexity of 3D models while retaining visual quality and relevant salient features. Traditional simplification techniques usually rely on solving a time-consuming optimization problem, hence they are impractical for large-scale datasets. In an attempt to alleviate this computational burden, we propose a fast point cloud simplification method by learning to sample salient points. The proposed method relies on a graph neural network architecture trained to select an arbitrary, user-defined, number of points according to their latent encodings and re-arrange their positions so as to minimize the visual perception error. The approach is extensively evaluated on various datasets using several perceptual metrics. Importantly, our method is able to generalize to out-of-distribution shapes, hence demonstrating zero-shot capabilities.



Paperid:77
Authors:Yatian Pang; Wenxiao Wang; Francis E.H. Tay; Wei Liu; Yonghong Tian; Li Yuan
Title: Masked Autoencoders for Point Cloud Self-Supervised Learning
Abstract:
As a promising scheme of self-supervised learning, masked autoencoding has significantly advanced natural language processing and computer vision. Inspired by this, we propose a neat scheme of masked autoencoders for point cloud self-supervised learning, addressing the challenges posed by point cloud’s properties, including leakage of location information and uneven information density. Concretely, we divide the input point cloud into irregular point patches and randomly mask them at a high ratio. Then, a standard Transformer based autoencoder, with an asymmetric design and a shifting mask tokens operation, learns high-level latent features from unmasked point patches, aiming to reconstruct the masked point patches. Extensive experiments show that our approach is efficient during pre-training and generalizes well on various downstream tasks. The pre-trained models achieve 85.18% accuracy on ScanObjectNN and 94.04% accuracy on ModelNet40, outperforming all the other self-supervised learning methods. We show with our scheme, a simple architecture entirely based on standard Transformers can surpass dedicated Transformer models from supervised learning. Our approach also advances state-of-the-art accuracies by 1.5%-2.3% in the few-shot classification. Furthermore, our work inspires the feasibility of applying unified architectures from languages and images to the point cloud. Codes are available at https://github.com/Pang-Yatian/Point-MAE.



Paperid:78
Authors:Lukas Koestler; Daniel Grittner; Michael Moeller; Daniel Cremers; Zorah Lähner
Title: Intrinsic Neural Fields: Learning Functions on Manifolds
Abstract:
Neural fields have gained significant attention in the computer vision community due to their excellent performance in novel view synthesis, geometry reconstruction, and generative modeling. Some of their advantages are a sound theoretic foundation and an easy implementation in current deep learning frameworks. While neural fields have been applied to signals on manifolds, e.g., for texture reconstruction, their representation has been limited to extrinsically embedding the shape into Euclidean space. The extrinsic embedding ignores known intrinsic manifold properties and is inflexible wrt. transfer of the learned function. To overcome these limitations, this work introduces intrinsic neural fields, a novel and versatile representation for neural fields on manifolds. Intrinsic neural fields combine the advantages of neural fields with the spectral properties of the Laplace-Beltrami operator. We show theoretically that intrinsic neural fields inherit many desirable properties of the extrinsic neural field framework but exhibit additional intrinsic qualities, like isometry invariance. In experiments, we show intrinsic neural fields can reconstruct high-fidelity textures from images with state-of-the-art quality and are robust to the discretization of the underlying manifold. We demonstrate the versatility of intrinsic neural fields by tackling various applications: texture transfer between deformed shapes & different shapes, texture reconstruction from real-world images with view dependence, and discretization-agnostic learning on meshes and point clouds.



Paperid:79
Authors:Zhouyingcheng Liao; Jimei Yang; Jun Saito; Gerard Pons-Moll; Yang Zhou
Title: Skeleton-Free Pose Transfer for Stylized 3D Characters
Abstract:
We present the first method that automatically transfers poses between stylized 3D characters without skeletal rigging. In contrast to previous attempts to learn pose transformations on fixed or topology-equivalent skeleton templates, our method focuses on a novel scenario to handle skeleton-free characters with diverse shapes, topologies, and mesh connectivities. The key idea of our method is to represent the characters in a unified articulation model so that the pose can be transferred through the correspondent parts. To achieve this, we propose a novel pose transfer network that predicts the character skinning weights and deformation transformations jointly to articulate the target character to match the desired pose. Our method is trained in a semi-supervised manner absorbing all existing character data with paired/unpaired poses and stylized shapes. It generalizes well to unseen stylized characters and inanimate objects. We conduct extensive experiments and demonstrate the effectiveness of our method on this novel task.



Paperid:80
Authors:Haotian Liu; Mu Cai; Yong Jae Lee
Title: Masked Discrimination for Self-Supervised Learning on Point Clouds
Abstract:
Masked autoencoding has achieved great success for self-supervised learning in the image and language domains. However, mask based pretraining has yet to show benefits for point cloud understanding, likely due to standard backbones like PointNet being unable to properly handle the training versus testing distribution mismatch introduced by masking during training. In this paper, we bridge this gap by proposing a discriminative mask pretraining Transformer framework, MaskPoint, for point clouds. Our key idea is to represent the point cloud as discrete occupancy values (1 if part of the point cloud; 0 if not), and perform simple binary classification between masked object points and sampled noise points as the proxy task. In this way, our approach is robust to the point sampling variance in point clouds, and facilitates learning rich representations. We evaluate our pretrained models across several downstream tasks, including 3D shape classification, segmentation, and real-word object detection, and demonstrate state-of-the-art results while achieving a significant pretraining speedup (e.g., 4.1x on ScanNet) compared to the prior state-of-the-art Transformer baseline. Code is available at https://github.com/haotian-liu/MaskPoint.



Paperid:81
Authors:Xuejun Yan; Hongyu Yan; Jingjing Wang; Hang Du; Zhihong Wu; Di Xie; Shiliang Pu; Li Lu
Title: FBNet: Feedback Network for Point Cloud Completion
Abstract:
The rapid development of point cloud learning has driven point cloud completion into a new era. However, the information flows of most existing completion methods are solely feedforward, and high-level information is rarely reused to improve low-level feature learning. To this end, we propose a novel Feedback Network (FBNet) for point cloud completion, in which present features are efficiently refined by rerouting subsequent fine-grained ones. Firstly, partial inputs are fed to a Hierarchical Graph-based Network (HGNet) to generate coarse shapes. Then, we cascade several Feedback-Aware Completion (FBAC) Blocks and unfold them across time recurrently. Feedback connections between two adjacent time steps exploit fine-grained features to improve present shape generations. The main challenge of building feedback connections is the dimension mismatching between present and subsequent features. To address this, the elaborately designed point Cross Transformer exploits efficient information from feedback features via cross attention strategy and then refines present features with the enhanced feedback features. Quantitative and qualitative experiments on several datasets demonstrate the superiority of proposed FBNet compared to state-of-the-art methods on point completion task. The source code and model are available at https://github.com/hikvision-research/3DVision/tree/main/PointCompletion/FBNet.



Paperid:82
Authors:Ta-Ying Cheng; Qingyong Hu; Qian Xie; Niki Trigoni; Andrew Markham
Title: Meta-Sampler: Almost-Universal yet Task-Oriented Sampling for Point Clouds
Abstract:
Sampling is a key operation in point-cloud task and acts to increase computational efficiency and tractability by discarding redundant points. Universal sampling algorithms (e.g., Farthest Point Sampling) work without modification across different tasks, models, and datasets, but by their very nature are agnostic about the downstream task/model. As such, they have no implicit knowledge about which points would be best to keep and which to reject. Recent work has shown how task-specific point cloud sampling (e.g., SampleNet) can be used to outperform traditional sampling approaches by learning which points are more informative. However, these learnable samplers face two inherent issues: i) overfitting to a model rather than a task, and ii) requiring training of the sampling network from scratch, in addition to the task network, somewhat countering the original objective of down-sampling to increase efficiency. In this work, we propose an almost-universal sampler, in our quest for a sampler that can learn to preserve the most useful points for a particular task, yet be inexpensive to adapt to different tasks, models or datasets. We first demonstrate how training over multiple models for the same task (e.g., shape reconstruction) significantly outperforms the vanilla SampleNet in terms of accuracy by not overfitting the sample network to a particular task network. Second, we show how we can train an almost-universal meta-sampler across multiple tasks. This meta-sampler can then be rapidly fine-tuned when applied to different datasets, networks, or even different tasks, thus amortizing the initial cost of training.



Paperid:83
Authors:Ishit Mehta; Manmohan Chandraker; Ravi Ramamoorthi
Title: A Level Set Theory for Neural Implicit Evolution under Explicit Flows
Abstract:
Coordinate-based neural networks parameterizing implicit surfaces have emerged as efficient representations of geometry. They effectively act as parametric level sets with the zero-level set defining the surface of interest. We present a framework that allows applying deformation operations defined for triangle meshes onto such implicit surfaces. Several of these operations can be viewed as energy-minimization problems that induce an instantaneous flow field on the explicit surface. Our method uses the flow field to deform parametric implicit surfaces by extending the classical theory of level sets. We also derive a consolidated view for existing methods on differentiable surface extraction and rendering, by formalizing connections to the level-set theory. We show that these methods drift from the theory and that our approach exhibits improvements for applications like surface smoothing, mean-curvature flow, inverse rendering and user-defined editing on implicit geometry.



Paperid:84
Authors:Wanli Chen; Xinge Zhu; Guojin Chen; Bei Yu
Title: Efficient Point Cloud Analysis Using Hilbert Curve
Abstract:
Previous state-of-the-art research on analyzing point cloud mainly rely on the voxelization quantization because it keeps the better spatial locality and geometry. However, these 3D voxelization methods and subsequent 3D convolution networks often bring the large computational overhead and GPU occupation. A straightforward alternative is to flatten 3D voxelization into 2D structure or utilize the pillar representation to perform the dimension reduction, while all of them would inevitably alter the spatial locality and 3D geometric information. In this way, we propose the HilbertNet to maintain the locality advantage of voxel-based methods while significantly reducing the computational cost. Here the key component is a new flattening mechanism based on Hilbert curve, which is a famous locality and geometry preserving function. Namely, if flattening 3D voxels using Hilbert curve encoding, the resulting structure will have similar spatial topology compared with original voxels. Through the Hilbert flattening, we can not only use 2D convolution (more lightweight than 3D convolution) to process voxels, but also incorporate technologies suitable in 2D space, such as transformer, to boost the performance. Our proposed HilbertNet achieves state-of-the-art performance on ShapeNet and ModelNet40 datasets with smaller cost and GPU occupation.



Paperid:85
Authors:Keyang Zhou; Bharat Lal Bhatnagar; Jan Eric Lenssen; Gerard Pons-Moll
Title: TOCH: Spatio-Temporal Object-to-Hand Correspondence for Motion Refinement
Abstract:
We present TOCH, a method for refining incorrect 3D hand-object interaction sequences using a data prior. Existing hand trackers, especially those that rely on very few cameras, often produce visually unrealistic results with hand-object intersection or missing contacts. Although correcting such errors requires reasoning about temporal aspects of interaction, most previous works focus on static grasps and contacts. The core of our method are TOCH fields, a novel spatio-temporal representation for modeling correspondences between hands and objects during interaction. TOCH fields are a point-wise, object-centric representation, which encode the hand position relative to the object. Leveraging this novel representation, we learn a latent manifold of plausible TOCH fields with a temporal denoising auto-encoder. Experiments demonstrate that TOCH outperforms state-of-the-art 3D hand-object interaction models, which are limited to static grasps and contacts. More importantly, our method produces smooth interactions even before and after contact. Using a single trained TOCH model, we quantitatively and qualitatively demonstrate its usefulness for correcting erroneous sequences from off-the-shelf RGB/RGB-D hand-object reconstruction methods and transferring grasps across objects.



Paperid:86
Authors:Ashkan Mirzaei; Yash Kant; Jonathan Kelly; Igor Gilitschenski
Title: LaTeRF: Label and Text Driven Object Radiance Fields
Abstract:
Obtaining 3D object representations is important for creating photo-realistic simulators and collecting assets for AR/VR applications. Neural fields have shown their effectiveness in learning a continuous volumetric representation of a scene from 2D images, but acquiring object representations from these models with weak supervision remains an open challenge. In this paper we introduce LaTeRF, a method for extracting an object of interest from a scene given 2D images of the entire scene and known camera poses, a natural language description of the object, and a small number of point-labels of object and non-object points in the input images. To faithfully extract the object from the scene, LaTeRF extends the NeRF formulation with an additional ‘objectness’ probability at each 3D point. Additionally, we leverage the rich latent space of a pre-trained CLIP model combined with our differentiable object renderer, to inpaint the occluded parts of the object. We demonstrate high-fidelity object extraction on both synthetic and real datasets and justify our design choices through an extensive ablation study.



Paperid:87
Authors:Yaqian Liang; Shanshan Zhao; Baosheng Yu; Jing Zhang; Fazhi He
Title: MeshMAE: Masked Autoencoders for 3D Mesh Data Analysis
Abstract:
Recently, self-supervised pre-training has advanced Vision Transformers on various tasks w.r.t. different data modalities, e.g., image and 3D point cloud data. In this paper, we explore this learning paradigm for 3D mesh data analysis based on Transformers. Since applying Transformer architectures to new modalities is usually non-trivial, we first adapt Vision Transformer to 3D mesh data processing, i.e., Mesh Transformer. In specific, we divide a mesh into several non-overlapping local patches with each containing the same number of faces and use the 3D position of each patch’s center point to form positional embeddings. Inspired by MAE, we explore how pre-training on 3D mesh data with the Transformer-based structure benefits downstream 3D mesh analysis tasks. We first randomly mask some patches of the mesh and feed the corrupted mesh into Mesh Transformers. Then, through reconstructing the information of masked patches, the network is capable of learning discriminative representations for mesh data. Therefore, we name our method MeshMAE, which can yield state-of-the-art or comparable performance on mesh analysis tasks, i.e., classification and segmentation. In addition, we also conduct comprehensive ablation studies to show the effectiveness of key designs in our method.



Paperid:88
Authors:Dongliang Cao; Florian Bernard
Title: Unsupervised Deep Multi-Shape Matching
Abstract:
3D shape matching is a long-standing problem in computer vision and computer graphics. While deep neural networks were shown to lead to state-of-the-art results in shape matching, existing learning-based approaches are limited in the context of multi-shape matching: (i) either they focus on matching pairs of shapes only and thus suffer from cycle-inconsistent multi-matchings, or (ii) they require an explicit template shape to address the matching of a collection of shapes. In this paper, we present a novel approach for deep multi-shape matching that ensures cycle-consistent multi-matchings while not depending on an explicit template shape. To this end, we utilise a shape-to-universe multi-matching representation that we combine with powerful functional map regularisation, so that our multi-shape matching neural network can be trained in a fully unsupervised manner. While the functional map regularisation is only considered during training time, functional maps are not computed for predicting correspondences, thereby allowing for fast inference. We demonstrate that our method achieves state-of-the-art results on several challenging benchmark datasets, and, most remarkably, that our unsupervised method even outperforms recent supervised methods.



Paperid:89
Authors:Yawar Siddiqui; Justus Thies; Fangchang Ma; Qi Shan; Matthias Nießner; Angela Dai
Title: Texturify: Generating Textures on 3D Shape Surfaces
Abstract:
Texture cues on 3D objects are key to compelling visual representations, with the possibility to create high visual fidelity with inherent spatial consistency across different views. Since the availability of textured 3D shapes remains very limited, learning a 3D-supervised data-driven method that predicts a texture based on the 3D input is very challenging. We thus propose Texturify, a GAN-based method that leverages a 3D shape dataset of an object class and learns to reproduce the distribution of appearances observed in real images by generating high-quality textures. In particular, our method does not require any 3D color supervision or correspondence between shape geometry and images to learn the texturing of 3D objects. Texturify operates directly on the surface of the 3D objects by introducing face convolutional operators on a hierarchical 4-RoSy parametrization to generate plausible object-specific textures. Employing differentiable rendering and adversarial losses that critique individual views and consistency across views, we effectively learn the high-quality surface texturing distribution from real-world images. Experiments on car and chair shape collections show that our approach outperforms state of the art by an average of 22% in FID score.



Paperid:90
Authors:An-Chieh Cheng; Xueting Li; Sifei Liu; Min Sun; Ming-Hsuan Yang
Title: Autoregressive 3D Shape Generation via Canonical Mapping
Abstract:
With the capacity of modeling long-range dependencies in sequential data, transformers have shown remarkable performances in a variety of generative tasks such as image, audio, and text generation. Yet, taming them in generating less structured and voluminous data formats such as high-resolution point clouds have seldom been explored due to ambiguous sequentialization processes and infeasible computation burden. In this paper, we aim to further exploit the power of transformers and employ them for the task of 3D point cloud generation. The key idea is to decompose point clouds of one category into semantically aligned sequences of shape compositions, via a learned canonical space. These shape compositions can then be quantized and used to learn a context-rich composition codebook for point cloud generation. Experimental results on point cloud reconstruction and unconditional generation show that our model performs favorably against state-of-the-art approaches. Furthermore, our model can be easily extended to multi-modal shape completion as an application for conditional shape generation.



Paperid:91
Authors:Jun-Kun Chen; Yu-Xiong Wang
Title: PointTree: Transformation-Robust Point Cloud Encoder with Relaxed K-D Trees
Abstract:
Being able to learn an effective semantic representation directly on raw point clouds has become a central topic in 3D understanding. Despite rapid progress, state-of-the-art encoders are restrictive to canonicalized point clouds, and have weaker than necessary performance when encountering geometric transformation distortions. To overcome this challenge, we propose PointTree, a general-purpose point cloud encoder that is robust to transformations based on relaxed K-D trees. Key to our approach is the design of the division rule in K-D trees by using principal component analysis (PCA). We use the structure of the relaxed K-D tree as our computational graph, and model the features as border descriptors which are merged with pointwise-maximum operation. In addition to this novel architecture design, we further improve the robustness by introducing pre-alignment -- a simple yet effective PCA-based normalization scheme. Our PointTree encoder combined with pre-alignment consistently outperforms state-of-the-art methods by large margins, for applications from object classification to semantic segmentation on various transformed versions of the widely-benchmarked datasets. Code and pre-trained models are available at https://github.com/immortalCO/PointTree.



Paperid:92
Authors:Shenhan Qian; Jiale Xu; Ziwei Liu; Liqian Ma; Shenghua Gao
Title: UNIF: United Neural Implicit Functions for Clothed Human Reconstruction and Animation
Abstract:
We propose united implicit functions (UNIF), a part-based method for clothed human reconstruction and animation with raw scans and skeletons as the input. Previous part-based methods for human reconstruction rely on ground-truth part labels from SMPL and thus are limited to minimal-clothed humans. In contrast, our method learns to separate parts from body motions instead of part supervision, thus can be extended to clothed humans and other articulated objects. Our Partition-from-Motion is achieved by a bone-centered initialization, a bone limit loss, and a section normal loss that ensure stable part division even when the training poses are limited. We also present a minimal perimeter loss for SDF to suppress extra surfaces and part overlapping. Another core of our method is an adjacent part seaming algorithm that produces non-rigid deformations to maintain the connection between parts which significantly relieves the part-based artifacts. Under this algorithm, we further propose ""Competing Parts”, a method that defines blending weights by the relative position of a point to bones instead of the absolute position, avoiding the generalization problem of neural implicit functions with inverse LBS (linear blend skinning). We demonstrate the effectiveness of our method by clothed human body reconstruction and animation on the CAPE and the ClothSeq datasets.



Paperid:93
Authors:Brandon Y. Feng; Yinda Zhang; Danhang Tang; Ruofei Du; Amitabh Varshney
Title: PRIF: Primary Ray-Based Implicit Function
Abstract:
We introduce a new implicit shape representation called Primary Ray-based Implicit Function (PRIF). In contrast to most existing approaches based on the signed distance function (SDF) which handles spatial locations, our representation operates on oriented rays. Specifically, PRIF is formulated to directly produce the surface hit point of a given input ray, without the expensive sphere-tracing operations, hence enabling efficient shape extraction and differentiable rendering. We demonstrate that neural networks trained to encode PRIF achieve successes in various tasks including single shape representation, category-wise shape generation, shape completion from sparse or noisy observations, inverse rendering for camera pose estimation, and neural rendering with color.



Paperid:94
Authors:Hanxue Liang; Hehe Fan; Zhiwen Fan; Yi Wang; Tianlong Chen; Yu Cheng; Zhangyang Wang
Title: Point Cloud Domain Adaptation via Masked Local 3D Structure Prediction
Abstract:
The superiority of deep learning based point cloud representations relies on large-scale labeled datasets, while the annotation of point clouds is notoriously expensive. One of the most effective solutions is to transfer the knowledge from existing labeled source data to unlabeled target data. However, domain bias typically hinders knowledge transfer and leads to accuracy degradation. In this paper, we propose a Masked Local Structure Prediction (MLSP) method to encode target data. Along with the supervised learning on the source domain, our method enables models to embed source and target data in a shared feature space. Specifically, we predict masked local structure via estimating point cardinality, position and normal. Our design philosophies lie in: 1) Point cardinality reflects basic structures (e.g., line, edge and plane) that are invariant to specific domains. 2) Predicting point positions in masked areas generalizes learned representations so that they are robust to incompletion-caused domain bias. 3) Point normal is generated by neighbors and thus robust to noise across domains. We conduct experiments on shape classification and semantic segmentation with different transfer permutations and the results demonstrate the effectiveness of our method. Code is available at https://github.com/VITA-Group/MLSP.



Paperid:95
Authors:Kim Youwang; Kim Ji-Yeon; Tae-Hyun Oh
Title: CLIP-Actor: Text-Driven Recommendation and Stylization for Animating Human Meshes
Abstract:
We propose CLIP-Actor, a text-driven motion recommendation and neural mesh stylization system for human mesh animation. CLIP-Actor animates a 3D human mesh to conform to a text prompt by recommending a motion sequence and optimizing mesh style attributes. We build a text-driven human motion recommendation system by leveraging a large-scale human motion dataset with language labels. Given a natural language prompt, CLIP-Actor suggests a text-conforming human motion in a coarse-to-fine manner. Then, our novel zero-shot neural style optimization detailizes and texturizes the recommended mesh sequence to conform to the prompt in a temporally-consistent and pose-agnostic manner. This is distinctive in that prior work fails to generate plausible results when the pose of an artist-designed mesh does not conform to the text from the beginning. We further propose the spatio-temporal view augmentation and mask-weighted embedding attention, which stabilize the optimization process by leveraging multi-frame human motion and rejecting poorly rendered views. We demonstrate that CLIP-Actor produces plausible and human-recognizable style 3D human mesh in motion with detailed geometry and texture solely from a natural language prompt.



Paperid:96
Authors:Samir Agarwala; Linyi Jin; Chris Rockwell; David F. Fouhey
Title: PlaneFormers: From Sparse View Planes to 3D Reconstruction
Abstract:
We present an approach for the planar surface reconstruction of a scene from images with limited overlap. This reconstruction task is challenging since it requires jointly reasoning about single image 3D reconstruction, correspondence between images, and the relative camera pose between images. Past work has proposed optimization-based approaches. We introduce a simpler approach, the PlaneFormer, that uses a transformer applied to 3D-aware plane tokens to perform 3D reasoning. Our experiments show that our approach is substantially more effective than prior work, and that several 3D-specific design decisions are crucial for its success.



Paperid:97
Authors:Siyou Lin; Hongwen Zhang; Zerong Zheng; Ruizhi Shao; Yebin Liu
Title: Learning Implicit Templates for Point-Based Clothed Human Modeling
Abstract:
We present FITE, a First-Implicit-Then-Explicit framework for modeling human avatars in clothing. Our framework first learns implicit surface templates representing the coarse clothing topology, and then employs the templates to guide the generation of point sets which further capture pose-dependent clothing deformations such as wrinkles. Our pipeline incorporates the merits of both implicit and explicit representations, namely, the ability to handle varying topology and the ability to efficiently capture fine details. We also propose diffused skinning to facilitate template training especially for loose clothing, and projection-based pose-encoding to extract pose information from mesh templates without predefined UV map or connectivity. Our code is publicly available at https://github.com/jsnln/fite.



Paperid:98
Authors:Qianjiang Hu; Daizong Liu; Wei Hu
Title: Exploring the Devil in Graph Spectral Domain for 3D Point Cloud Attacks
Abstract:
With the maturity of depth sensors, point clouds have received increasing attention in various applications such as autonomous driving, robotics, surveillance, \etc., while deep point cloud learning models have shown to be vulnerable to adversarial attacks. Existing attack methods generally add/delete points or perform point-wise perturbation over point clouds to generate adversarial examples in the data space, which may neglect the geometric characteristics of point clouds. Instead, we propose point cloud attacks from a new perspective---Graph Spectral Domain Attack (GSDA), aiming to perturb transform coefficients in the graph spectral domain that corresponds to varying certain geometric structure. In particular, we naturally represent a point cloud over a graph, and adaptively transform the coordinates of points into the graph spectral domain via graph Fourier transform (GFT) for compact representation. We then analyze the influence of different spectral bands on the geometric structure of the point cloud, based on which we propose to perturb the GFT coefficients in a learnable manner guided by an energy constraint loss function. Finally, the adversarial point cloud is generated by transforming the perturbed spectral representation back to the data domain via the inverse GFT (IGFT). Experimental results demonstrate the effectiveness of the proposed GSDA in terms of both imperceptibility and attack success rates under a variety of defense strategies.



Paperid:99
Authors:Jingwang Ling; Zhibo Wang; Ming Lu; Quan Wang; Chen Qian; Feng Xu
Title: Structure-Aware Editable Morphable Model for 3D Facial Detail Animation and Manipulation
Abstract:
Morphable models are essential for the statistical modeling of 3D faces. Previous works on morphable models mostly focus on large-scale facial geometry but ignore facial details. This paper augments morphable models in representing facial details by learning a Structure-aware Editable Morphable Model (SEMM). SEMM introduces a detail structure representation based on the distance field of wrinkle lines, jointly modeled with detail displacements to establish better correspondences and enable intuitive manipulation of wrinkle structure. Besides, SEMM introduces two transformation modules to translate expression blendshape weights and age values into changes in latent space, allowing effective semantic detail editing while maintaining identity. Extensive experiments demonstrate that the proposed model compactly represents facial details, outperforms previous methods in expression animation qualitatively and quantitatively, and achieves effective age editing and wrinkle line editing of facial details. Code and model are available at https://github.com/gerwang/facial-detail-manipulation.



Paperid:100
Authors:Yiyu Zhuang; Hao Zhu; Xusen Sun; Xun Cao
Title: MoFaNeRF: Morphable Facial Neural Radiance Field
Abstract:
We propose a parametric model that maps free-view images into a vector space of coded facial shape, expression and appearance with a neural radiance field, namely Morphable Facial NeRF. Specifically, MoFaNeRF takes the coded facial shape, expression and appearance along with space coordinate and view direction as input to an MLP, and outputs the radiance of the space point for photo-realistic image synthesis. Compared with conventional 3D morphable models (3DMM), MoFaNeRF shows superiority in directly synthesizing photo-realistic facial details even for eyes, mouths, and beards. Also, continuous face morphing can be easily achieved by interpolating the input shape, expression and appearance codes. By introducing identity-specific modulation and texture encoder, our model synthesizes accurate photometric details and shows strong representation ability. Our model shows strong ability on multiple applications including image-based fitting, random generation, face rigging, face editing, and novel view synthesis. Experiments show that our method achieves higher representation ability than previous parametric models, and achieves competitive performance in several applications. To the best of our knowledge, our work is the first facial parametric model built upon a neural radiance field that can be used in fitting, generation and manipulation. The code and data is available at https://github.com/zhuhao-nju/mofanerf.



Paperid:101
Authors:Tong He; Wei Yin; Chunhua Shen; Anton van den Hengel
Title: PointInst3D: Segmenting 3D Instances by Points
Abstract:
The current state-of-the-art methods in 3D instance segmentation typically involve a clustering step, despite the tendency towards heuristics, greedy algorithms, and a lack of robustness to the changes in data statistics. In contrast, we propose a fully convolutional 3D point cloud instance segmentation method that works in a per-point prediction fashion. In doing so it avoids the challenges that clustering-based methods face: introducing dependencies among different tasks of the model. We find the key to its success is assigning a suitable target to each sampled point. Instead of the commonly used static or distance-based assignment strategies, we propose to use an Optimal Transport approach to optimally assign target masks to the sampled points according to the dynamic matching costs. Our approach achieves promising results on both ScanNet and S3DIS benchmarks. The proposed approach removes intertask dependencies and thus represents a simpler and more flexible 3D instance segmentation framework than other competing methods, while achieving improved segmentation accuracy.



Paperid:102
Authors:Zezhou Cheng; Menglei Chai; Jian Ren; Hsin-Ying Lee; Kyle Olszewski; Zeng Huang; Subhransu Maji; Sergey Tulyakov
Title: Cross-Modal 3D Shape Generation and Manipulation
Abstract:
Creating and editing the shape and color of 3D objects require tremendous human effort and expertise. Compared to direct manipulation in 3D interfaces, 2D interactions such as sketches and scribbles are usually much more natural and intuitive for the users. In this paper, we propose a generic multi-modal generative model that couples the 2D modalities and implicit 3D representations through shared latent spaces. With the proposed model, versatile 3D generation and manipulation are enabled by simply propagating the editing from a specific 2D controlling modality through the latent spaces. For example, editing the 3D shape by drawing a sketch, re-colorizing the 3D surface via painting color scribbles on the 2D rendering, or generating 3D shapes of a certain category given one or a few reference images. Unlike prior works, our model does not require re-training or fine-tuning per editing task and is also conceptually simple, easy to implement, robust to input domain shifts, and flexible to diverse reconstruction on partial 2D inputs. We evaluate our framework on two representative 2D modalities of grayscale line sketches and rendered color images, and demonstrate that our method enables various shape manipulation and generation tasks with these 2D modalities.



Paperid:103
Authors:Chao Chen; Yu-Shen Liu; Zhizhong Han
Title: Latent Partition Implicit with Surface Codes for 3D Representation
Abstract:
Deep implicit functions have shown remarkable shape modeling ability in various 3D computer vision tasks. One drawback is that it is hard for them to represent a 3D shape as multiple parts. Current solutions learn various primitives and blend the primitives directly in the spatial space, which still struggle to approximate the 3D shape accurately. To resolve this problem, we introduce a novel implicit representation to represent a single 3D shape as a set of parts in the latent space, towards both highly accurate and plausibly interpretable shape modeling. Our insight here is that both the part learning and the part blending can be conducted much easier in the latent space than in the spatial space. We name our method Latent Partition Implicit (LPI), because of its ability of casting the global shape modeling into multiple local part modeling, which partitions the global shape unity. LPI represents a shape as Signed Distance Functions (SDFs) using surface codes. Each surface code is a latent code representing a part whose center is on the surface, which enables us to flexibly employ intrinsic attributes of shapes or additional surface properties. Eventually, LPI can reconstruct both the shape and the parts on the shape, both of which are plausible meshes. LPI is a multi-level representation, which can partition a shape into different numbers of parts after training. LPI can be learned without ground truth signed distances, point normals or any supervision for part partition. LPI outperforms the state-of-the-art methods under the widely used benchmarks in terms of reconstruction accuracy and modeling interpretability.



Paperid:104
Authors:Ramana Sundararaman; Gautam Pai; Maks Ovsjanikov
Title: Implicit Field Supervision for Robust Non-rigid Shape Matching
Abstract:
Establishing a correspondence between two non-rigidly deforming shapes is one of the most fundamental problems in visual computing. Existing methods often show weak resilience when presented with challenges innate to real-world data such as noise, outliers, self-occlusion etc. On the other hand, auto-decoders have demonstrated strong expressive power in learning geometrically meaningful latent embeddings. However, their use in shape analysis has been limited. In this paper, we introduce an approach based on an auto-decoder framework, that learns a continuous shape-wise deformation field over a fixed template. By supervising the deformation field for points on-surface and regularizing for points off-surface through a novel Signed Distance Regularization (SDR), we learn an alignment between the template and shape volumes. Trained on clean water-tight meshes, without any data-augmentation, we demonstrate compelling performance on compromised data and real-world scans.



Paperid:105
Authors:Shota Hattori; Tatsuya Yatagawa; Yutaka Ohtake; Hiromasa Suzuki
Title: Learning Self-Prior for Mesh Denoising Using Dual Graph Convolutional Networks
Abstract:
This study proposes a deep-learning framework for mesh denoising from a single noisy input, where two graph convolutional networks are trained jointly to filter vertex positions and facet normals apart. The prior obtained only from a single input is particularly referred to as a self-prior. The proposed method leverages the framework of the deep image prior (DIP), which obtains the self-prior for image restoration using a convolutional neural network (CNN). Thus, we obtain a denoised mesh without any ground-truth noise-free meshes. Compared to the original DIP that transforms a fixed random code into a noise-free image by the neural network, we reproduce vertex displacement from a fixed random code and reproduce facet normals from feature vectors that summarize local triangle arrangements. After tuning several hyperparameters with a few validation samples, our method achieved significantly higher performance than traditional approaches working with a single noisy input mesh. Moreover, its performance is better than the other methods using deep neural networks trained with a large-scale shape dataset. The independence of our method of either large-scale datasets or ground-truth noise-free mesh will allow us to easily denoise meshes whose shapes are rarely included in the shape datasets.



Paperid:106
Authors:Manxi Lin; Aasa Feragen
Title: diffConv: Analyzing Irregular Point Clouds with an Irregular View
Abstract:
Standard spatial convolutions assume input data with a regular neighborhood structure. Existing methods typically generalize convolution to the irregular point cloud domain by fixing a regular ""view"" through e.g. a fixed neighborhood size, where the convolution kernel size remains the same for each point. However, since point clouds are not as structured as images, the fixed neighbor number gives an unfortunate inductive bias. We present a novel graph convolution named Difference Graph Convolution (diffConv), which does not rely on a regular view. diffConv operates on spatially-varying and density-dilated neighborhoods, which are further adapted by a learned masked attention mechanism. Experiments show that our model is very robust to the noise, obtaining state-of-the-art performance in 3D shape classification and scene understanding tasks, along with a faster inference speed.



Paperid:107
Authors:Aihua Mao; Zihui Du; Yu-Hui Wen; Jun Xuan; Yong-Jin Liu
Title: PD-Flow: A Point Cloud Denoising Framework with Normalizing Flows
Abstract:
Point cloud denoising aims to restore clean point clouds from raw observations corrupted by noise and outliers while preserving the fine-grained details. We present a novel deep learning-based denoising model, that incorporates normalizing flows and noise disentanglement techniques to achieve high denoising accuracy. Unlike the existing works that extract features of point clouds for point-wise correction, we formulate the denoising process from the perspective of distribution learning and feature disentanglement. By considering noisy point clouds as a joint distribution of clean points and noise, the denoised results can be derived from disentangling the noise counterpart from latent point representation, whereas the mapping between Euclidean and latent spaces is modeled by normalizing flows. We evaluate our method on synthesized 3D models and real-world datasets with various noise settings. Qualitative and quantitative results show that our method surpasses previous state-of-the-art deep learning-based approaches in terms of detail preservation and distribution uniformity. The source code is available at https://github.com/unknownue/pdflow.



Paperid:108
Authors:Haoran Zhou; Yun Cao; Wenqing Chu; Junwei Zhu; Tong Lu; Ying Tai; Chengjie Wang
Title: SeedFormer: Patch Seeds Based Point Cloud Completion with Upsample Transformer
Abstract:
Point cloud completion has become increasingly popular among generation tasks of 3D point clouds, as it is a challenging yet indispensable problem to recover the complete shape of a 3D object from its partial observation. In this paper, we propose a novel SeedFormer to improve the ability of detail preservation and recovery in point cloud completion. Unlike previous methods based on a global feature vector, we introduce a new shape representation, namely Patch Seeds, which not only captures general structures from partial inputs but also preserves regional information of local patterns. Then, by integrating seed features into the generation process, we can recover faithful details for complete point clouds in a coarse-to-fine manner. Moreover, we devise an Upsample Transformer by extending the transformer structure into basic operations of point generators, which effectively incorporates spatial and semantic relationships between neighboring points. Qualitative and quantitative evaluations demonstrate that our method outperforms state-of-the-art completion networks on several benchmark datasets. Our code is available at https://github.com/hrzhou2/seedformer.



Paperid:109
Authors:Nikolas Lamb; Sean Banerjee; Natasha Kholgade Banerjee
Title: DeepMend: Learning Occupancy Functions to Represent Shape for Repair
Abstract:
We present DeepMend, a novel approach to reconstruct restorations to fractured shapes using learned occupancy functions. Existing shape repair approaches predict low-resolution voxelized restorations or smooth restorations, or require symmetries or access to a pre-existing complete oracle. We represent the occupancy of a fractured shape as the conjunction of the occupancy of an underlying complete shape and a break surface, which we model as functions of latent codes using neural networks. Given occupancy samples from a fractured shape, we estimate latent codes using an inference loss augmented with novel penalties to avoid empty or voluminous restorations. We use the estimated codes to reconstruct a restoration shape. We show results with simulated fractures on synthetic and real-world scanned objects, and with scanned real fractured mugs. Compared to existing approaches and to two baseline methods, our work shows state-of-the-art results in accuracy and avoiding restoration artifacts over non-fracture regions of the fractured shape.



Paperid:110
Authors:Qingyang Tan; Yi Zhou; Tuanfeng Wang; Duygu Ceylan; Xin Sun; Dinesh Manocha
Title: A Repulsive Force Unit for Garment Collision Handling in Neural Networks
Abstract:
Despite recent success, deep learning-based methods for predicting 3D garment deformation under body motion suffer from interpenetration problems between the garment and the body. To address this problem, we propose a novel collision handling neural network layer called Repulsive Force Unit (ReFU). Based on the signed distance function (SDF) of the underlying body and the current garment vertex positions, ReFU predicts the per-vertex offsets that push any interpenetrating vertex to a collision-free configuration while preserving the fine geometric details. We show that ReFU is differentiable with trainable parameters and can be integrated into different network backbones that predict 3D garment deformations. Our experiments show that ReFU significantly reduces the number of collisions between the body and the garment and better preserves geometric details compared to prior methods based on collision loss or post-processing optimization.



Paperid:111
Authors:Oren Katzir; Dani Lischinski; Daniel Cohen-Or
Title: Shape-Pose Disentanglement Using SE(3)-Equivariant Vector Neurons
Abstract:
We introduce an unsupervised technique for encoding point clouds into a canonical shape representation, by disentangling shape and pose. Our encoder is stable and consistent, meaning that the shape encoding is purely pose-invariant, while the extracted rotation and translation are able to semantically align different input shapes of the same class to a common canonical pose. Specifically, we design an auto-encoder based on Vector Neuron Networks, a rotation-equivariant neural network, whose layers we extend to provide translation-equivariance in addition to rotation-equivariance only. The resulting encoder produces pose-invariant shape encoding by construction, enabling our approach to focus on learning a consistent canonical pose for a class of objects. Quantitative and qualitative experiments validate the superior stability and consistency of our approach.



Paperid:112
Authors:Yunlu Chen; Basura Fernando; Hakan Bilen; Matthias Nießner; Efstratios Gavves
Title: 3D Equivariant Graph Implicit Functions
Abstract:
In recent years, neural implicit representations have made remarkable progress in modeling of 3D shapes with arbitrary topology. In this work, we address two key limitations of such representations, in failing to capture local 3D geometric fine details, and to learn from and generalize to shapes with unseen 3D transformations. To this end, we introduce a novel family of graph implicit functions with equivariant layers that facilitates modeling fine local details and guaranteed robustness to various groups of geometric transformations, through local k-NN graph embeddings with sparse point set observations at multiple resolutions. Our method improves over the existing rotation-equivariant implicit function from 0.69 to 0.89 (IoU) on the ShapeNet reconstruction task. We also show that our equivariant implicit function can be extended to other types of similarity transformations and generalizes to unseen translations and scaling.



Paperid:113
Authors:Bo Sun; Vladimir G. Kim; Noam Aigerman; Qixing Huang; Siddhartha Chaudhuri
Title: PatchRD: Detail-Preserving Shape Completion by Learning Patch Retrieval and Deformation
Abstract:
This paper introduces a data-driven shape completion approach that focuses on completing geometric details of missing regions of 3D shapes. We observe that existing generative methods do not have enough training data and representation capacity to synthesize plausible, fine-grained details with complex geometry and topology. Thus, our key insight is to copy and deform the patches from the partial input to complete the missing regions. This enables us to preserve the style of local geometric features, even if it is drastically different from the training data. Our fully automatic approach proceeds in two stages. First, we learn to retrieve candidate patches from the input shape. Second, we select and deform some of the retrieved candidates to seamlessly blend them into the complete shape. This method combines the advantages of the two most common completion methods: similarity-based single-instance completion, and completion by learning a shape space. We leverage repeating patterns by retrieving patches from the partial input, and learn global structural priors by using a neural network to guide the retrieval and deformation steps. Experimental results show that our approach considerably outperforms baseline approaches across multiple datasets and shape categories.



Paperid:114
Authors:Emery Pierson; Mohamed Daoudi; Sylvain Arguillere
Title: 3D Shape Sequence of Human Comparison and Classification Using Current and Varifolds
Abstract:
In this paper we address the task of the comparison and the classification of 3D shape sequences of human. The non-linear dynamics of the human motion and the changing of the surface parametrization over the time make this task very challenging. To tackle this issue, we propose to embed the 3D shape sequences in an infinite dimensional space, the space of varifolds, endowed with an inner product that comes from a given positive definite kernel. More specifically, our approach involves two steps: 1) the surfaces are represented as varifolds, this representation induces metrics equivariant to rigid motions and invariant to parametrization; 2) the sequences of 3D shapes are represented by Gram matrices derived from their infinite dimensional Hankel matrices, and we use Frobenius distance between two Symmetric Positive definite (SPD) matrices to compare two sequences. Extensive experiments show that our method is competitive with state-of-the-art in 3D sequence motion retrieval.



Paperid:115
Authors:Jianxiong Shen; Antonio Agudo; Francesc Moreno-Noguer; Adria Ruiz
Title: Conditional-Flow NeRF: Accurate 3D Modelling with Reliable Uncertainty Quantification
Abstract:
A critical limitation of current methods based on Neural Radiance Fields (NeRF) is that they are unable to quantify the uncertainty associated with the learned appearance and geometry of the scene. This information is paramount in real applications such as medical diagnosis or autonomous driving where, to reduce potentially catastrophic failures, the confidence on the model outputs must be included into the decision-making process. In this context, we introduce Conditional-Flow NeRF (CF-NeRF), a novel probabilistic framework to incorporate uncertainty quantification into NeRF-based approaches. For this purpose, our method learns a distribution over all possible radiance fields modelling the scene which is used to quantify the uncertainty associated with the modelled scene. In contrast to previous approaches enforcing strong constraints over the radiance field distribution, CF-NeRF learns it in a flexible and fully data-driven manner by coupling Latent Variable Modelling and Conditional Normalizing Flows. This strategy allows to obtain reliable uncertainty estimation while preserving model expressivity. Compared to previous state-of-the-art methods proposed for uncertainty quantification in NeRF, our experiments show that the proposed method achieves significantly lower prediction errors and more reliable uncertainty values for synthetic novel view and depth-map estimation.



Paperid:116
Authors:Yuki Kawana; Yusuke Mukuta; Tatsuya Harada
Title: Unsupervised Pose-Aware Part Decomposition for Man-Made Articulated Objects
Abstract:
Man-made articulated objects exist widely in the real world. However, previous methods for unsupervised part decomposition are unsuitable for such objects because they assume a spatially fixed part location, resulting in inconsistent part parsing. In this paper, we propose PPD (unsupervised Pose-aware Part Decomposition) to address a novel setting that explicitly targets man-made articulated objects with mechanical joints, considering the part poses in part parsing. As an analysis-by-synthesis approach, We show that category-common prior learning for both part shapes and poses facilitates the unsupervised learning of (1) part parsing with abstracted part shapes, and (2) part poses as joint parameters under single-frame shape supervision. We evaluate our method on synthetic and real datasets, and we show that it outperforms previous works in consistent part parsing of the articulated objects based on comparable part pose estimation performance to the supervised baseline.



Paperid:117
Authors:Benoît Guillard; Federico Stella; Pascal Fua
Title: MeshUDF: Fast and Differentiable Meshing of Unsigned Distance Field Networks
Abstract:
Unsigned Distance Fields (UDFs) can be used to represent non-watertight surfaces. However, current approaches to converting them into explicit meshes tend to either be expensive or to degrade the accuracy. Here, we extend the marching cube algorithm to handle UDFs, both fast and accurately. Moreover, our approach to surface extraction is differentiable, which is key to using pretrained UDF networks to fit sparse data.



Paperid:118
Authors:Zhaofan Qiu; Yehao Li; Yu Wang; Yingwei Pan; Ting Yao; Tao Mei
Title: SPE-Net: Boosting Point Cloud Analysis via Rotation Robustness Enhancement
Abstract:
In this paper, we propose a novel deep architecture tailored for 3D point cloud applications, named as SPE-Net. The embedded ""Selective Position Encoding (SPE)"" procedure relies on an attention mechanism that can effectively attend to the underlying rotation condition of the input. Such encoded rotation condition then determines which part of the network parameters to be focused on, and is shown to efficiently help reduce the degree of freedom of the optimization during training. This mechanism henceforth can better leverage the rotation augmentations through reduced training difficulties, making SPE-Net robust against rotated data both during training and testing. The new findings in our paper also urge us to rethink the relationship between the extracted rotation information and the actual test accuracy. Intriguingly, we reveal evidences that by locally encoding the rotation information through SPE-Net, the rotation-invariant features are still of critical importance in benefiting the test samples without any actual global rotation. We empirically demonstrate the merits of the SPE-Net and the associated hypothesis on four benchmarks, showing evident improvements on both rotated and unrotated test data over SOTA methods. Source code is available at https://github.com/ZhaofanQiu/SPE-Net.



Paperid:119
Authors:Kai Wang; Paul Guerrero; Vladimir G. Kim; Siddhartha Chaudhuri; Minhyuk Sung; Daniel Ritchie
Title: The Shape Part Slot Machine: Contact-Based Reasoning for Generating 3D Shapes from Parts
Abstract:
We present the Shape Part Slot Machine, a new method for assembling novel 3D shapes from existing parts by performing contact-based reasoning. Our method represents each shape as a graph of ""slots,"" where each slot is a region of contact between two shape parts. Based on this representation, we design a graph-neural-network-based model for generating new slot graphs and retrieving compatible parts, as well as a gradient-descent-based optimization scheme for assembling the retrieved parts into a complete shape that respects the generated slot graph. This approach does not require any semantic part labels; interestingly, it also does not require complete part geometries---reasoning about the regions where parts connect proves sufficient to generate novel, high-quality 3D shapes. We demonstrate that our method generates shapes that outperform existing modeling-by-assembly approaches in terms of quality, diversity, and structural complexity.



Paperid:120
Authors:Wangmeng Xiang; Chao Li; Biao Wang; Xihan Wei; Xian-Sheng Hua; Lei Zhang
Title: Spatiotemporal Self-Attention Modeling with Temporal Patch Shift for Action Recognition
Abstract:
Transformer-based methods have recently achieved great advancement on 2D image-based vision tasks. For 3D video-based tasks such as action recognition, however, directly applying spatiotemporal transformers on video data will bring heavy computation and memory burdens due to the largely increased number of patches and the quadratic complexity of self-attention computation. How to efficiently and effectively model the 3D self-attention of video data has been a great challenge for transformers. In this paper, we propose a Temporal Patch Shift (TPS) method for efficient 3D self-attention modeling in transformers for video-based action recognition. TPS shifts part of patches with a specific mosaic pattern in the temporal dimension, thus converting a vanilla spatial self-attention operation to a spatiotemporal one with little additional cost. As a result, we can compute 3D self-attention using nearly the same computation and memory cost as 2D self-attention. TPS is a plug-and-play module and can be inserted into existing 2D transformer models to enhance spatiotemporal feature learning. The proposed method achieves competitive performance with state-of-the-arts on Something-something V1 \& V2, Diving-48, and Kinetics400 while being much more efficient on computation and memory cost. The source code of TPS can be found at https://github.com/MartinXM/TPS.



Paperid:121
Authors:Sauradip Nag; Xiatian Zhu; Yi-Zhe Song; Tao Xiang
Title: Proposal-Free Temporal Action Detection via Global Segmentation Mask Learning
Abstract:
Existing temporal action detection (TAD) methods rely on generating an overwhelmingly large number of proposals per video. This leads to complex model designs due to proposal generation and/or per-proposal action instance evaluation and the resultant high computational cost. In this work, for the first time, we propose a proposal-free Temporal Action detection model with Global Segmentation mask (TAGS). Our core idea is to learn a global segmentation mask of each action instance jointly at the full video length. The TAGS model differs significantly from the conventional proposal-based methods by focusing on global temporal representation learning to directly detect local start and end points of action instances without proposals. Further, by modeling TAD holistically rather than locally at the individual proposal level, TAGS needs a much simpler model architecture with lower computational cost. Extensive experiments show that despite its simpler design, TAGS outperforms existing TAD methods, achieving new state-of-the-art performance on two benchmarks. Importantly, it is 20x faster to train and 1.6x more efficient for inference. Our PyTorch implementation of TAGS is available at https://github.com/sauradip/TAGS"



Paperid:122
Authors:Sauradip Nag; Xiatian Zhu; Yi-Zhe Song; Tao Xiang
Title: Semi-Supervised Temporal Action Detection with Proposal-Free Masking
Abstract:
Existing temporal action detection (TAD) methods rely on a large number of training data with segment-level annotations. Collecting and annotating such a training set is thus highly expensive and unscalable. Semi-supervised TAD (SS-TAD) alleviates this problem by leveraging unlabeled videos freely available at scale. However, SS-TAD is also a much more challenging problem than supervised TAD, and consequently much under-studied. Prior SS-TAD methods directly combine an existing proposal-based TAD method and a SSL method. Due to their sequential localization (e.g, proposal generation) and classification design, they are prone to proposal error propagation. To overcome this limitation, in this work we propose a novel Semi-supervised Temporal action detection model based on PropOsal-free Temporal mask (SPOT) with a parallel localization (mask generation) and classification architecture. Such a novel design effectively eliminates the dependence between localization and classification by cutting off the route for error propagation in-between. We further introduce an interaction mechanism between classification and localization for prediction refinement, and a new pretext task for self-supervised model pre-training. Extensive experiments on two standard benchmarks show that our SPOT outperforms state-of-the-art alternatives, often by a large margin. The PyTorch implementation of SPOT is available at https://github.com/sauradip/SPOT"



Paperid:123
Authors:Sauradip Nag; Xiatian Zhu; Yi-Zhe Song; Tao Xiang
Title: Zero-Shot Temporal Action Detection via Vision-Language Prompting
Abstract:
Existing temporal action detection (TAD) methods rely on large training data including segment-level annotations, limited to recognizing previously seen classes alone during inference. Collecting and annotating a large training set for each class of interest is costly and hence unscalable. Zero-shot TAD (ZS-TAD) resolves this obstacle by enabling a pre-trained model to recognize any unseen action classes. Meanwhile, ZS-TAD is also much more challenging with significantly less investigation. Inspired by the success of zero-shot image classification aided by vision-language (ViL) models such as CLIP, we aim to tackle the more complex TAD task. An intuitive method is to integrate an off-the-shelf proposal detector with CLIP style classification. However, due to the sequential localization (e.g, proposal generation) and classification design, it is prone to localization error propagation. To overcome this problem, in this paper we propose a novel zero-Shot Temporal Action detection model via vision-LanguagE prompting (STALE). Such a novel design effectively eliminates the dependence between localization and classification by breaking the route for error propagation in-between. We further introduce an interaction mechanism between classification and localization for improved optimization. Extensive experiments on standard ZS-TAD video benchmarks show that our STALE significantly outperforms state-of-the-art alternatives. Besides, our model also yields superior results on supervised TAD over recent strong competitors. The PyTorch implementation of STALE is available on https://github.com/sauradip/STALE.



Paperid:124
Authors:Wei Lin; Anna Kukleva; Kunyang Sun; Horst Possegger; Hilde Kuehne; Horst Bischof
Title: CycDA: Unsupervised Cycle Domain Adaptation to Learn from Image to Video
Abstract:
Although action recognition has achieved impressive results over recent years, both collection and annotation of video training data are still time-consuming and cost intensive. Therefore, image-to-video adaptation has been proposed to exploit labeling-free web image source for adapting on unlabeled target videos. This poses two major challenges: (1) spatial domain shift between web images and video frames; (2) modality gap between image and video data. To address these challenges, we propose Cycle Domain Adaptation (CycDA), a cycle-based approach for unsupervised image-to-video domain adaptation. We leverage the joint spatial information in images and videos on the one hand and, on the other hand, train an independent spatio-temporal model to bridge the modality gap. We alternate between the spatial and spatio-temporal learning with knowledge transfer between the two in each cycle. We evaluate our approach on benchmark datasets for image-to-video as well as for mixed-source domain adaptation achieving state-of-the-art results and demonstrating the benefits of our cyclic adaptation.



Paperid:125
Authors:Zengyu Wan; Yang Wang; Ganchao Tan; Yang Cao; Zheng-Jun Zha
Title: S2N: Suppression-Strengthen Network for Event-Based Recognition under Variant Illuminations
Abstract:
The emerging event-based sensors have demonstrated out-standing potential in visual tasks thanks to their high speed and high dynamic range. However, the event degradation due to imaging under low illumination obscures the correlation between event signals and brings uncertainty into event representation. Targeting this issue, we present a novel suppression-strengthen network (S2N) to augment the event feature representation after suppressing the influence of degradation. Specifically, a suppression sub-network is devised to obtain intensity mapping between the degraded and denoised enhancement frames by unsupervised learning. To further restrain the degradation’s influence, a strengthen sub-network is presented to generate robust event representation by adaptively perceiving the local variations between the center and surrounding regions. After being trained on a single illumination condition, our S2N can be directly generalized to other illuminations to boost the recognition performance. Experimental results on three challenging recognition tasks demonstrate the superiority of our method. The codes and datasets could refer to https://github.com/wanzengy/S2N-Suppression-Strengthen-Network.



Paperid:126
Authors:Yunyao Mao; Wengang Zhou; Zhenbo Lu; Jiajun Deng; Houqiang Li
Title: CMD: Self-Supervised 3D Action Representation Learning with Cross-Modal Mutual Distillation
Abstract:
In 3D action recognition, there exists rich complementary information between skeleton modalities. Nevertheless, how to model and utilize this information remains a challenging problem for self-supervised 3D action representation learning. In this work, we formulate the cross-modal interaction as a bidirectional knowledge distillation problem. Different from classic distillation solutions that transfer the knowledge of a fixed and pre-trained teacher to the student, in this work, the knowledge is continuously updated and bidirectionally distilled between modalities. To this end, we propose a new Cross-modal Mutual Distillation (CMD) framework with the following designs. On the one hand, the neighboring similarity distribution is introduced to model the knowledge learned in each modality, where the relational information is naturally suitable for the contrastive frameworks. On the other hand, asymmetrical configurations are used for teacher and student to stabilize the distillation process and to transfer high-confidence information between modalities. By derivation, we find that the cross-modal positive mining in previous works can be regarded as a degenerated version of our CMD. We perform extensive experiments on NTU RGB+D 60, NTU RGB+D 120, and PKU-MMD II datasets. Our approach outperforms existing self-supervised methods and sets a series of new records. The code is available at https://github.com/maoyunyao/CMD"



Paperid:127
Authors:Bolin Ni; Houwen Peng; Minghao Chen; Songyang Zhang; Gaofeng Meng; Jianlong Fu; Shiming Xiang; Haibin Ling
Title: Expanding Language-Image Pretrained Models for General Video Recognition
Abstract:
Contrastive language-image pretraining has shown great success in learning visual-textual joint representation from web-scale data, demonstrating remarkable “zero-shot” generalization ability for various image tasks. However, how to effectively expand such new language-image pretraining methods to video domains is still an open problem. In this work, we present a simple yet effective approach that adapts the pretrained language-image models to video recognition directly, instead of pretraining a new model from scratch. More concretely, to capture the long-range dependencies of frames along the temporal dimension, we propose a cross-frame attention mechanism that explicitly exchanges information across frames. Such module is lightweight and can be plugged into pretrained language-image models seamlessly. Moreover, we propose a video-specific prompting scheme, which leverages video content information for generating discriminative textual prompts. Extensive experiments demonstrate that our approach is effective and can be generalized to different video recognition scenarios. In particular, under fully-supervised settings, our approach achieves a top-1 accuracy of 87.1% on Kinectics-400, while using 12× fewer FLOPs compared with Swin-L and ViViT-H. In zero-shot experiments, our approach surpasses the current state-of-the-art methods by +7.6% and +14.9% in terms of top-1 accuracy under two popular protocols. In few-shot scenarios, our approach outperforms previous best methods by +32.1% and +23.1% when the labeled data is extremely limited. Code and models are available at https://github.com/microsoft/VideoX/tree/master/X-CLIP"



Paperid:128
Authors:Masato Tamura; Rahul Vishwakarma; Ravigopal Vennelakanti
Title: Hunting Group Clues with Transformers for Social Group Activity Recognition
Abstract:
This paper presents a novel framework for social group activity recognition. As an expanded task of group activity recognition, social group activity recognition requires recognizing multiple sub-group activities and identifying group members. Most existing methods tackle both tasks by refining region features and then summarizing them into activity features. Such heuristic feature design renders the effectiveness of features susceptible to incomplete person localization and disregards the importance of scene contexts. Furthermore, region features are sub-optimal to identify group members because the features may be dominated by those of people in the regions and have different semantics. To overcome these drawbacks, we propose to leverage attention modules in transformers to generate effective social group features. Our method is designed in such a way that the attention modules identify and then aggregate features relevant to social group activities, generating an effective feature for each social group. Group member information is embedded into the features and thus accessed by feed-forward networks. The outputs of feed-forward networks represent groups so concisely that group members can be identified with simple Hungarian matching between groups and individuals. Experimental results show that our method outperforms state-of-the-art methods on the Volleyball and Collective Activity datasets.



Paperid:129
Authors:Haoyuan Zhang; Yonghong Hou; Wenjing Zhang; Wanqing Li
Title: Contrastive Positive Mining for Unsupervised 3D Action Representation Learning
Abstract:
Recent contrastive based 3D action representation learning has made great progress. However, the strict positive/negative constraint is yet to be relaxed and the use of non-self positive is yet to be explored. In this paper, a Contrastive Positive Mining (CPM) framework is proposed for unsupervised skeleton 3D action representation learning. The CPM identifies non-self positives in a contextual queue to boost learning. Specifically, the siamese encoders are adopted and trained to match the similarity distributions of the augmented instances in reference to all instances in the contextual queue. By identifying the non-self positive instances in the queue, a positive-enhanced learning strategy is proposed to leverage the knowledge of mined positives to boost the robustness of the learned latent space against intra-class and inter-class diversity. Experimental results have shown that the proposed CPM is effective and outperforms the existing state-of-the-art unsupervised methods on the challenging NTU and PKU-MMD datasets.



Paperid:130
Authors:Zhibo Yang; Sounak Mondal; Seoyoung Ahn; Gregory Zelinsky; Minh Hoai; Dimitris Samaras
Title: Target-Absent Human Attention
Abstract:
The prediction of human gaze behavior is important for building human-computer interactive systems that can anticipate a user’s attention. Computer vision models have been developed to predict the fixations made by people as they search for target objects. But what about when the image has no target? Equally important is to know how people search when they cannot find a target, and when they would stop searching. In this paper, we propose a data-driven computational model that addresses the search-termination problem and predicts the scanpath of search fixations made by people searching for targets that do not appear in images. We model visual search as an imitation learning problem and represent the internal knowledge that the viewer acquires through fixations using a novel state representation that we call Foveated Feature Maps (FFMs). FFMs integrate a simulated foveated retina into a pretrained ConvNet that produces an in-network feature pyramid, all with minimal computational overhead. Our method integrates FFMs as the state representation in inverse reinforcement learning. Experimentally, we improve the state of the art in predicting human target-absent search behavior on the COCO-Search18 dataset.



Paperid:131
Authors:Hongji Guo; Zhou Ren; Yi Wu; Gang Hua; Qiang Ji
Title: Uncertainty-Based Spatial-Temporal Attention for Online Action Detection
Abstract:
Online action detection aims at detecting the ongoing action in a streaming video. In this paper, we proposed an uncertainty-based spatial-temporal attention for online action detection. By explicitly modeling the distribution of model parameters, we extend the baseline models in a probabilistic manner. Then we quantify the predictive uncertainty and use it to generate spatial-temporal attention that focus on large mutual information regions and frames. For inference, we introduce a two-stream framework that combines the baseline model and the probabilistic model based on the input uncertainty. We validate the effectiveness of our method on three benchmark datasets: THUMOS-14, TVSeries, and HDD. Furthermore, we demonstrate that our method generalizes better under different views and occlusions, and is more robust when training with small-scale data.



Paperid:132
Authors:Danyang Tu; Xiongkuo Min; Huiyu Duan; Guodong Guo; Guangtao Zhai; Wei Shen
Title: Iwin: Human-Object Interaction Detection via Transformer with Irregular Windows
Abstract:
This paper presents a new vision Transformer, named Iwin Transformer, which is specifically designed for human-object interaction (HOI) detection, a detailed scene understanding task involving a sequential process of human/object detection and interaction recognition. Iwin Transformer is a hierarchical Transformer which progressively performs token representation learning and token agglomeration within irregular windows. The irregular windows, achieved by augmenting regular grid locations with learned offsets, 1) eliminate redundancy in token representation learning, which leads to efficient humans/objects detection, and 2) enable the agglomerated tokens to align with humans/objects with different shapes, which facilitates the acquisition of highly-abstracted visual semantics for interaction recognition. The effectiveness and efficiency of Iwin Transformer are verified on the two standard HOI detection benchmark datasets, HICO-DET and V-COCO. Results show our method outperforms existing Transformers-based methods by large margins (3.7 mAP gain on HICO-DET and 2.0 mAP gain on V-COCO) with fewer training epochs (0.5 ×).



Paperid:133
Authors:Yijun Qian; Lijun Yu; Wenhe Liu; Alexander G. Hauptmann
Title: Rethinking Zero-Shot Action Recognition: Learning from Latent Atomic Actions
Abstract:
To avoid the time-consuming annotating and retraining cycle in applying supervised action recognition models, Zero-Shot Action Recognition (ZSAR) has become a thriving direction. ZSAR requires models to recognize actions that never appear in the training set through bridging visual features and semantic representations. However, due to the complexity of actions, it remains challenging to transfer knowledge learned from source to target action domains. Previous ZSAR methods mainly focus on mitigating representation variance between source and target actions through integrating or applying new action-level features. However, the action-level features are coarse-grained and make the learned one-to-one bridge fragile to similar target actions. Meanwhile, integration or application of features usually requires extra computation or annotation. These methods didn’t notice that two actions with different names may still share the same atomic action components. It enables humans to quickly understand an unseen action given a bunch of atomic actions learned from seen actions. Inspired by this, we propose Jigsaw Network (JigsawNet) which recognizes complex actions through unsupervisedly decomposing them into combinations of atomic actions and bridging group-to-group relationships between visual features and semantic representations. To enhance the robustness of the learned group-to-group bridge, we propose Group Excitation (GE) module to model intra-sample knowledge and Consistency Loss to enforce the model learn from inter-sample knowledge. Our JigsawNet achieves state-of-the-art performance on three benchmarks and surpasses previous works with noticeable margins.



Paperid:134
Authors:Xiaoqian Wu; Yong-Lu Li; Xinpeng Liu; Junyi Zhang; Yuzhe Wu; Cewu Lu
Title: Mining Cross-Person Cues for Body-Part Interactiveness Learning in HOI Detection
Abstract:
Human-Object Interaction (HOI) detection plays a crucial role in activity understanding. Though significant progress has been made, interactiveness learning remains a challenging problem in HOI detection: existing methods usually generate redundant negative H-O pair proposals and fail to effectively extract interactive pairs. Though interactiveness has been studied in both whole body- and part- level and facilitates the H-O pairing, previous works only focus on the target person once (i.e., in a local perspective) and overlook the information of the other persons. In this paper, we argue that comparing body-parts of multi-person simultaneously can afford us more useful and supplementary interactiveness cues. That said, to learn body-part interactiveness from a global perspective: when classifying a target person’s body-part interactiveness, visual cues are explored not only from herself/himself but also from other persons in the image. We construct body-part saliency maps based on self-attention to mine cross-person informative cues and learn the holistic relationships between all the body-parts. We evaluate the proposed method on widely-used benchmarks HICO-DET and V-COCO. With our new perspective, the holistic global-local body-part interactiveness learning achieves significant improvements over state-of-the-art. Our code is available at https://github.com/enlighten0707/Body-Part-Map-for-Interactiveness.



Paperid:135
Authors:Qinying Liu; Zilei Wang
Title: Collaborating Domain-Shared and Target-Specific Feature Clustering for Cross-Domain 3D Action Recognition
Abstract:
In this work, we consider the problem of cross-domain 3D action recognition in the open-set setting, which has been rarely explored before. Specifically, there is a source domain and a target domain that contain the skeleton sequences with different styles and categories, and our purpose is to cluster the target data by utilizing the labeled source data and unlabeled target data. For such a challenging task, this paper presents a novel approach dubbed CoDT to collaboratively cluster the domain-shared features and target-specific features. CoDT consists of two parallel branches. One branch aims to learn domain-shared features with supervised learning in the source domain, while the other is to learn target-specific features using contrastive learning in the target domain. To cluster the features, we propose an online clustering algorithm that enables simultaneous promotion of robust pseudo label generation and feature clustering. Furthermore, to leverage the complementarity of domain-shared features and target-specific features, we propose a novel collaborative clustering strategy to enforce pair-wise relationship consistency between the two branches. We conduct extensive experiments on multiple cross-domain 3D action recognition datasets, and the results demonstrate the effectiveness of our method.



Paperid:136
Authors:Filip Ilic; Thomas Pock; Richard P. Wildes
Title: Is Appearance Free Action Recognition Possible?
Abstract:
Intuition might suggest that motion and dynamic information are key to video-based action recognition. In contrast, there is evidence that state-of-the-art deep-learning video understanding architectures are biased toward static information available in single frames. Presently, a methodology and corresponding dataset to isolate the effects of dynamic information in video are missing. Their absence makes it difficult to understand how well contemporary architectures capitalize on dynamic vs. static information. We respond with a novel Appearance Free Dataset (AFD) for action recognition. AFD is devoid of static information relevant to action recognition in a single frame. Modeling of the dynamics is necessary for solving the task, as the action is only apparent through consideration of the temporal dimension. We evaluated 11 contemporary action recognition architectures on AFD as well as its related RGB video. Our results show a notable decrease in performance for all architectures on AFD compared to RGB. We also conducted a complimentary study with humans that shows their recognition accuracy on AFD and RGB is very similar and much better than the evaluated architectures on AFD. Our results motivate a novel architecture that revives explicit recovery of optical flow, within a contemporary design for best performance on AFD and RGB.



Paperid:137
Authors:Ning Ma; Hongyi Zhang; Xuhui Li; Sheng Zhou; Zhen Zhang; Jun Wen; Haifeng Li; Jingjun Gu; Jiajun Bu
Title: Learning Spatial-Preserved Skeleton Representations for Few-Shot Action Recognition
Abstract:
Few-shot action recognition aims to recognize few-labeled novel action classes and attracts growing attentions due to practical significance. Human skeletons provide explainable and data-efficient representation for this problem by explicitly modeling spatial-temporal relations among skeleton joints. However, existing skeleton-based spatial-temporal models tend to deteriorate the positional distinguishability of joints, which leads to fuzzy spatial matching and poor explainability. To address these issues, we propose a novel spatial matching strategy consisting of spatial disentanglement and spatial activation. The motivation behind spatial disentanglement is that we find more spatial information for leaf nodes (e.g., the “hand” joint ) is beneficial to increase representation diversity for skeleton matching. To achieve spatial disentanglement, we encourage the skeletons to be represented in a full rank space with rank maximization constraint. Finally, an attention based spatial activation mechanism is introduced to incorporate the disentanglement, by adaptively adjusting the disentangled joints according to matching pairs. Extensive experiments on three skeleton benchmarks demonstrate that the proposed spatial matching strategy can be effectively inserted into existing temporal alignment frameworks, achieving considerable performance improvements as well as inherent explainability.



Paperid:138
Authors:Mengyuan Chen; Junyu Gao; Shicai Yang; Changsheng Xu
Title: Dual-Evidential Learning for Weakly-Supervised Temporal Action Localization
Abstract:
Weakly-supervised temporal action localization (WS-TAL) aims to localize the action instances and recognize their categories with only video-level labels. Despite great progress, existing methods suffer from severe action-background ambiguity, which mainly comes from background noise introduced by aggregation operations and large intra-action variations caused by the task gap between classification and localization. To address this issue, we propose a generalized evidential deep learning (EDL) framework for WS-TAL, called Dual-Evidential Learning for Uncertainty modeling (DELU), which extends the traditional paradigm of EDL to adapt to the weakly-supervised multi-label classification goal. Specifically, targeting at adaptively excluding the undesirable background snippets, we utilize the video-level uncertainty to measure the interference of background noise to video-level prediction. Then, the snippet-level uncertainty is further induced for progressive learning, which gradually focuses on the entire action instances in an “easy-to-hard” manner. Extensive experiments show that DELU achieves state-of-the-art performance on THUMOS14 and ActivityNet1.2 benchmarks. Our code is available in github.com/MengyuanChen21/ECCV2022-DELU.



Paperid:139
Authors:Boeun Kim; Hyung Jin Chang; Jungho Kim; Jin Young Choi
Title: Global-Local Motion Transformer for Unsupervised Skeleton-Based Action Learning
Abstract:
We propose a new transformer model for the task of unsupervised learning of skeleton motion sequences. The existing transformer model utilized for unsupervised skeleton-based action learning is learned the instantaneous velocity of each joint from adjacent frames without global motion information. Thus, the model has difficulties in learning the attention globally over whole-body motions and temporally distant joints. In addition, person-to-person interactions have not been considered in the model. To tackle the learning of whole-body motion, long-range temporal dynamics, and person-to-person interactions, we design a global and local attention mechanism, where, global body motions and local joint motions pay attention to each other. In addition, we propose a novel pretraining strategy, multi-interval pose displacement prediction, to learn both global and local attention in diverse time ranges. The proposed model successfully learns local dynamics of the joints and captures global context from the motion sequences. Our model outperforms state-of-the-art models by notable margins in the representative benchmarks.



Paperid:140
Authors:Yulin Wang; Yang Yue; Xinhong Xu; Ali Hassani; Victor Kulikov; Nikita Orlov; Shiji Song; Humphrey Shi; Gao Huang
Title: AdaFocusV3: On Unified Spatial-Temporal Dynamic Video Recognition
Abstract:
Recent research has revealed that reducing the temporal and spatial redundancy are both effective approaches towards efficient video recognition, e.g., allocating the majority of computation to a task-relevant subset of frames or the most valuable image regions of each frame. However, in most existing works, either type of redundancy is typically modeled with another absent. This paper explores the unified formulation of spatial-temporal dynamic computation on top of the recently proposed AdaFocusV2 algorithm, contributing to an improved AdaFocusV3 framework. Our method reduces the computational cost by activating the expensive high-capacity network only on some small but informative 3D video cubes. These cubes are cropped from the space formed by frame height, width, and video duration, while their locations are adaptively determined with a light-weighted policy network on a per-sample basis. At test time, the number of the cubes corresponding to each video is dynamically configured, i.e., video cubes are processed sequentially until a sufficiently reliable prediction is produced. Notably, AdaFocusV3 can be effectively trained by approximating the non-differentiable cropping operation with the interpolation of deep features. Extensive empirical results on six benchmark datasets (i.e., ActivityNet, FCVID, Mini-Kinetics, Something-Something V1&V2 and Diving48) demonstrate that our model is considerably more efficient than competitive baselines.



Paperid:141
Authors:Ruize Han; Haomin Yan; Jiacheng Li; Songmiao Wang; Wei Feng; Song Wang
Title: Panoramic Human Activity Recognition
Abstract:
To obtain a more comprehensive activity understanding for a crowded scene, in this paper, we propose a new problem of panoramic human activity recognition (PAR), which aims to simultaneously achieve the the recognition of individual actions, social group activities, and global activities. This is a challenging yet practical problem in real-world applications. To track this problem, we develop a novel hierarchical graph neural network to progressively represent and model the multi-granular human activities and mutual social relations for a crowd of people. We further build a benchmark to evaluate the proposed method and other related methods. Experimental results verify the rationality of the proposed PAR problem, the effectiveness of our method and the usefulness of the benchmark. We have released the source code and benchmark to the public for promoting the study on this problem.



Paperid:142
Authors:Shuxian Liang; Xu Shen; Jianqiang Huang; Xian-Sheng Hua
Title: Delving into Details: Synopsis-to-Detail Networks for Video Recognition
Abstract:
In this paper, we explore the details in video recognition with the aim to improve the accuracy. It is observed that most failure cases in recent works fall on the mis-classifications among very similar actions (such as high kick vs. side kick) that need a capturing of fine-grained discriminative details. To solve this problem, we propose synopsis-to-detail networks for video action recognition. Firstly, a synopsis network is introduced to predict the top-k likely actions and generate the synopsis (location & scale of details and contextual features). Secondly, according to the synopsis, a detail network is applied to extract the discriminative details in the input and infer the final action prediction. The proposed synopsis-to-detail networks enable us to train models directly from scratch in an end-to-end manner and to investigate various architectures for synopsis/detail recognition. Extensive experiments on benchmark datasets, including Kinetics-400, Mini-Kinetics and Something-Something V1 & V2, show that our method is more effective and efficient than the competitive baselines. Code is available at: https://github.com/liang4sx/S2DNet.



Paperid:143
Authors:Rahul Rahaman; Dipika Singhania; Alexandre Thiery; Angela Yao
Title: A Generalized \& Robust Framework for Timestamp Supervision in Temporal Action Segmentation
Abstract:
In temporal action segmentation, Timestamp supervision requires only a handful of labeled frames per video sequence. For unlabelled frames, Timestamp works rely on assigning hard labels and performance rapidly collapses under subtle violations of the annotation assumptions. We propose a novel Expectation-Maximization (EM) based approach which leverages label uncertainty of unlabelled frames and is robust enough to accommodate possible annotation errors. With accurate Timestamp annotations, our proposed method produces state-of-the-art results and even exceeds the fully-supervised setup in several metrics and datasets. When applied to timestamp annotations with missed action segments, we show that our method remains stable in terms of performance. To further test the robustness of our formulation, we introduce a new challenging annotation setup of SkipTag supervision. SkipTag is a relaxation on timestamps to allow for annotations of any fixed number of random frames in a video, making it more flexible than Timestamp supervision while remaining competitive.



Paperid:144
Authors:Sipeng Zheng; Shizhe Chen; Qin Jin
Title: Few-Shot Action Recognition with Hierarchical Matching and Contrastive Learning
Abstract:
Few-shot action recognition aims to recognize actions in test videos based on limited annotated data of target action classes. The dominant approaches project videos into a metric space and classify videos via nearest neighboring. They mainly measure video similarities using global or temporal alignment alone, while an optimum matching should be multi-level. However, the complexity of learning coarse-to-fine matching quickly rises as we focus on finer-grained visual cues, and the lack of detailed local supervision is another challenge. In this work, we propose a hierarchical matching model to support comprehensive similarity measure at global, temporal and spatial levels via a zoom-in matching module.We further propose a mixed-supervised hierarchical contrastive learning (HCL) in training, which not only employs supervised contrastive learning to differentiate videos at different levels, but also utilizes cycle consistency as weak supervision to align discriminative temporal clips or spatial patches. Our model achieves state-of-the-art performance on four benchmarks especially under the most challenging 1-shot action recognition setting. Code will be provided in supplementary material.



Paperid:145
Authors:Carlos Hinojosa; Miguel Marquez; Henry Arguello; Ehsan Adeli; Li Fei-Fei; Juan Carlos Niebles
Title: PrivHAR: Recognizing Human Actions from Privacy-Preserving Lens
Abstract:
The accelerated use of digital cameras prompts an increasing concern about privacy and security, particularly in applications such as action recognition. In this paper, we propose an optimizing framework to provide robust visual privacy protection along the human action recognition pipeline. Our framework parameterizes the camera lens to successfully degrade the quality of the videos to inhibit privacy attributes and protect against adversarial attacks while maintaining relevant features for activity recognition. We validate our approach with extensive simulations and hardware experiments.



Paperid:146
Authors:Guoqiu Li; Guanxiong Cai; Xingyu Zeng; Rui Zhao
Title: Scale-Aware Spatio-Temporal Relation Learning for Video Anomaly Detection
Abstract:
Recent progress in video anomaly detection (VAD) has shown that feature discrimination is the key to effectively distinguishing anomalies from normal events. We observe that many anomalous events occur in limited local regions, and the severe background noise increases the difficulty of feature learning. In this paper, we propose a scale-aware weakly supervised learning approach to capture local and salient anomalous patterns from the background, using only coarse video-level labels as supervision. We achieve this by segmenting frames into non-overlapping patches and then capturing inconsistencies among different regions through our patch spatial relation (PSR) module, which consists of self-attention mechanisms and dilated convolutions. To address the scale variation of anomalies and enhance the robustness of our method, a multi-scale patch aggregation method is further introduced to enable local-to-global spatial perception by merging features of patches with different scales. Considering the importance of temporal cues, we extend the relation modeling from the spatial domain to the spatio-temporal domain with the help of the existing video temporal relation network to effectively encode the spatio-temporal dynamics in the video. Experimental results show that our proposed method achieves new state-of-the-art performance on UCF-Crime and ShanghaiTech benchmarks. Code are available at https://github.com/nutuniv/SSRL.



Paperid:147
Authors:Yifei Huang; Lijin Yang; Yoichi Sato
Title: Compound Prototype Matching for Few-Shot Action Recognition
Abstract:
Few-shot action recognition aims to recognize novel action classes using only a small number of labeled training samples. In this work, we propose a novel approach that first summarizes each video into compound prototypes consisting of a group of global prototypes and a group of focused prototypes, and then compares video similarity based on the prototypes. Each global prototype is encouraged to summarize a specific aspect from the entire video, for example, the start of the action or the evolution of the action. Since no clear annotation is provided for the global prototypes, we use a group of focused prototypes and to focus on certain timestamps in the video. We compare similarity by matching the compound prototypes between the support and query videos. The global prototypes are directly matched so that the actions can be compared from the same perspective, for example, whether two actions start similarly. For the focused prototypes, since actions have various temporal shifts in the videos, we apply bipartite matching to allow comparison of the same action on different timestamps. Extensive experiments demonstrate that our proposed method achieves state-of-the-art results on multiple benchmarks by a large margin. A detailed ablation study analyzes the importance of each group of prototypes in capturing different aspects of the video.



Paperid:148
Authors:Lukas Hedegaard; Alexandros Iosifidis
Title: Continual 3D Convolutional Neural Networks for Real-Time Processing of Videos
Abstract:
We introduce Continual 3D Convolutional Neural Networks (Co3D CNNs), a new computational formulation of spatio-temporal 3D CNNs, in which videos are processed frame-by-frame rather than by clip. In online tasks demanding frame-wise predictions, Co3D CNNs dispense with the computational redundancies of regular 3D CNNs, namely the repeated convolutions over frames, which appear in overlapping clips. We show that Continual 3D CNNs can reuse preexisting 3D-CNN weights to reduce the per-prediction floating point operations (FLOPs) in proportion to the temporal receptive field while retaining similar memory requirements and accuracy. This is validated with multiple models on Kinetics-400 and Charades with remarkable results: CoX3D models attain state-of-the-art complexity/accuracy trade-offs on Kinetics-400 with 12.1-15.3x reductions of FLOPs and 2.3-3.8% improvements in accuracy compared to regular X3D models while reducing peak memory consumption by up to 48%. Moreover, we investigate the transient response of Co3D CNNs at start-up and perform extensive benchmarks of on-hardware processing characteristics for publicly available 3D CNNs.



Paperid:149
Authors:Tianjiao Li; Lin Geng Foo; Qiuhong Ke; Hossein Rahmani; Anran Wang; Jinghua Wang; Jun Liu
Title: Dynamic Spatio-Temporal Specialization Learning for Fine-Grained Action Recognition
Abstract:
The goal of fine-grained action recognition is to successfully discriminate between action categories with subtle differences. To tackle this, we derive inspiration from the human visual system which contains specialized regions in the brain that are dedicated towards handling specific tasks. We design a novel Dynamic Spatio-Temporal Specialization (DSTS) module, which consists of specialized neurons that are only activated for a subset of samples that are highly similar. During training, the loss forces the specialized neurons to learn discriminative fine-grained differences to distinguish between these similar samples, improving fine-grained recognition. Moreover, a spatio-temporal specialization method further optimizes the architectures of the specialized neurons to capture either more spatial or temporal fine-grained information, to better tackle the large range of spatio-temporal variations in the videos. Lastly, we design an Upstream-Downstream Learning algorithm to optimize our model’s dynamic decisions, allowing our DSTS module to generalize better. We obtain state-of-the-art performance on two widely-used fine-grained action recognition datasets. We will release our code.



Paperid:150
Authors:Zhiwei Yang; Peng Wu; Jing Liu; Xiaotao Liu
Title: Dynamic Local Aggregation Network with Adaptive Clusterer for Anomaly Detection
Abstract:
Existing methods for anomaly detection based on memory-augmented autoencoder (AE) have the following drawbacks: (1) Establishing a memory bank requires additional memory space. (2) The fixed number of prototypes from subjective assumptions ignores the data feature differences and diversity. To overcome these drawbacks, we introduce DLAN-AC, a Dynamic Local Aggregation Network with Adaptive Clusterer, for anomaly detection. First, The proposed DLAN can automatically learn and aggregate high-level features from the AE to obtain more representative prototypes, while freeing up extra memory space. Second, The proposed AC can adaptively cluster video data to derive initial prototypes with prior information. In addition, we also propose a dynamic redundant clustering strategy (DRCS) to enable DLAN for automatically eliminating feature clusters that do not contribute to the construction of prototypes. Extensive experiments on benchmarks demonstrate that DLAN-AC outperforms most existing methods, validating the effectiveness of our method. Our code is publicly available at https://github.com/Beyond-Zw/DLAN-AC.



Paperid:151
Authors:Yang Bai; Desen Zhou; Songyang Zhang; Jian Wang; Errui Ding; Yu Guan; Yang Long; Jingdong Wang
Title: Action Quality Assessment with Temporal Parsing Transformer
Abstract:
Action Quality Assessment(AQA) is important for action understanding and resolving the task poses unique challenges due to subtle visual differences. Existing state-of-the-art methods typically rely on the holistic video representations for score regression or ranking, which limits the generalization to capture fine-grained intra-class variation. To overcome the above limitation, we propose a temporal parsing transformer to decompose the holistic feature into temporal part-level representations. Specifically, we utilize a set of learnable queries to represent the atomic temporal patterns for a specific action. Our decoding process converts the frame representations to a fixed number of temporally ordered part representations. To obtain the quality score, we adopt the state-of-the-art contrastive regression based on the part representations. Since existing AQA datasets do not provide temporal part-level labels or partitions, we propose two novel loss functions on the cross attention responses of the decoder: a ranking loss to ensure the learnable queries to satisfy the temporal order in cross attention and a sparsity loss to encourage the part representations to be more discriminative. Extensive experiments show that our proposed method outperforms prior work on three public AQA benchmarks by a considerable margin.



Paperid:152
Authors:Bo Hu; Tat-Jen Cham
Title: Entry-Flipped Transformer for Inference and Prediction of Participant Behavior
Abstract:
Some group activities, such as team sports and choreographed dances, involve closely coupled interaction between participants. Here we investigate the tasks of inferring and predicting participant behavior, in terms of motion paths and actions, under such conditions. We narrow the problem to that of estimating how a set target participants react to the behavior of other observed participants. Our key idea is to model the spatio-temporal relations among participants in a manner that is robust to error accumulation during frame-wise inference and prediction. We propose a novel Entry-Flipped Transformer (EF-Transformer), which models the relations of participants by attention mechanisms on both spatial and temporal domains. Unlike typical transformers, we tackle the problem of error accumulation by flipping the order of query, key, and value entries, to increase the importance and fidelity of observed features in the current frame. Comparative experiments show that our EF-Transformer achieves the best performance on a newly-collected tennis doubles dataset, a Ceilidh dance dataset, and two pedestrian datasets. Furthermore, it is also demonstrated that our EF-Transformer is better at limiting accumulated errors and recovering from wrong estimations.



Paperid:153
Authors:Mingzhe Li; Hong-Bo Zhang; Qing Lei; Zongwen Fan; Jinghua Liu; Ji-Xiang Du
Title: Pairwise Contrastive Learning Network for Action Quality Assessment
Abstract:
Considering the complexity of modeling diverse actions of athletes, action quality assessment (AQA) in sports is a challenging task. A common solution is to tackle this problem as a regression task that map the input video to the final score provided by referees. However, it ignores the subtle and critical difference between videos. To address this problem, a new pairwise contrastive learning network (PCLN) is proposed to concern these differences and form an end-to-end AQA model with basic regression network. Specifically, the PCLN encodes video pairs to learn relative scores between videos to improve the performance of basic regression network. Furthermore, a new consistency constraint is defined to guide the training of the proposed AQA model. In the testing phase, only the basic regression network is employed, which makes the proposed method simple but high accuracy. The proposed method is verified on the AQA-7 and MTL-AQA datasets. Several ablation studies are built to verify the effectiveness of each component in the proposed method. The experimental results show that the proposed method achieves the state-of-the-art performance.



Paperid:154
Authors:Tanqiu Qiao; Qianhui Men; Frederick W. B. Li; Yoshiki Kubotani; Shigeo Morishima; Hubert P. H. Shum
Title: Geometric Features Informed Multi-Person Human-Object Interaction Recognition in Videos
Abstract:
Human-Object Interaction (HOI) recognition in videos is important for analyzing human activity. Most existing work focusing on visual features usually suffer from occlusion in the real-world scenarios. Such a problem will be further complicated when multiple people and objects are involved in HOIs. Consider that geometric features such as human pose and object position provide meaningful information to understand HOIs, we argue to combine the benefits of both visual and geometric features in HOI recognition, and propose a novel Two-level Geometric feature-informed Graph Convolutional Network (2G-GCN). The geometric-level graph models the interdependency between geometric features of humans and objects, while the fusion-level graph further fuses them with visual features of humans and objects. To demonstrate the novelty and effectiveness of our method in challenging scenarios, we propose a new multi-person HOI dataset (MPHOI-72). Extensive experiments on MPHOI-72 (multi-person HOI), CAD-120 (single-human HOI) and Bimanual Actions (two-hand HOI) datasets demonstrate our superior performance compared to state-of-the-arts.



Paperid:155
Authors:Chen-Lin Zhang; Jianxin Wu; Yin Li
Title: ActionFormer: Localizing Moments of Actions with Transformers
Abstract:
Self-attention based Transformer models have demonstrated impressive results for image classification and object detection, and more recently for video understanding. Inspired by this success, we investigate the application of Transformer networks for temporal action localization in videos. To this end, we present ActionFormer--a simple yet powerful model to identify actions in time and recognize their categories in a single shot, without using action proposals or relying on pre-defined anchor windows. ActionFormer combines a multiscale feature representation with local self-attention, and uses a light-weighted decoder to classify every moment in time and estimate the corresponding action boundaries. We show that this orchestrated design results in major improvements upon prior works. Without bells and whistles, ActionFormer achieves 71.0% mAP at tIoU=0.5 on THUMOS14, outperforming the best prior model by 14.1 absolute percentage points. Further, ActionFormer demonstrates strong results on ActivityNet 1.3 (36.6% average mAP) and EPIC-Kitchens 100 (+13.5% average mAP over prior works). Our code is available at https://github.com/happyharrycn/actionformer_release.



Paperid:156
Authors:Pei Xu; Jean-Bernard Hayet; Ioannis Karamouzas
Title: SocialVAE: Human Trajectory Prediction Using Timewise Latents
Abstract:
Predicting pedestrian movement is critical for human behavior analysis and also for safe and efficient human-agent interactions. However, despite significant advancements, it is still challenging for existing approaches to capture the uncertainty and multimodality of human navigation decision making. In this paper, we propose SocialVAE, a novel approach for human trajectory prediction. The core of SocialVAE is a timewise variational autoencoder architecture that exploits stochastic recurrent neural networks to perform prediction, combined with a social attention mechanism and a backward posterior approximation to allow for better extraction of pedestrian navigation strategies. We show that SocialVAE improves current state-of-the-art performance on several pedestrian trajectory prediction benchmarks, including the ETH/UCY benchmark, Stanford Drone Dataset, and SportVU NBA movement dataset.



Paperid:157
Authors:Zhaoyu Chen; Bo Li; Shuang Wu; Jianghe Xu; Shouhong Ding; Wenqiang Zhang
Title: Shape Matters: Deformable Patch Attack
Abstract:
Though deep neural networks (DNNs) have demonstrated excellent performance in computer vision, they are susceptible and vulnerable to carefully crafted adversarial examples which can mislead DNNs to incorrect outputs. Patch attack is one of the most threatening forms, which has the potential to threaten the security of real-world systems. Previous work always assumes patches to have fixed shapes, such as circles or rectangles, and it does not consider the shape of patches as a factor in patch attacks. To explore this issue, we propose a novel Deformable Patch Representation (DPR) that can harness the geometric structure of triangles to support the differentiable mapping between contour modeling and masks. Moreover, we introduce a joint optimization algorithm, named Deformable Adversarial Patch (DAPatch), which allows simultaneous and efficient optimization of shape and texture to enhance attack performance. We show that even with a small area, a particular shape can improve attack performance. Therefore, DAPatch achieves state-of-the-art attack performance by deforming shapes on GTSRB and ILSVRC2012 across various network architectures, and the generated patches can be threatening in the real world.



Paperid:158
Authors:Yuyang Long; Qilong Zhang; Boheng Zeng; Lianli Gao; Xianglong Liu; Jian Zhang; Jingkuan Song
Title: Frequency Domain Model Augmentation for Adversarial Attack
Abstract:
For black-box attacks, the gap between the substitute model and the victim model is usually large, which manifests as a weak attack performance. Motivated by the observation that the transferability of adversarial examples can be improved by attacking diverse models simultaneously, model augmentation methods which simulate different models by using transformed images are proposed. However, existing transformations for spatial domain do not translate to significantly diverse augmented models. To tackle this issue, we propose a novel spectrum simulation attack to craft more transferable adversarial examples against both normally trained and defense models. Specifically, we apply a spectrum transformation to the input and thus perform the model augmentation in the frequency domain. We theoretically prove that the transformation derived from frequency domain leads to a diverse spectrum saliency map, an indicator we proposed to reflect the diversity of substitute models. Notably, our method can be generally combined with existing attacks. Extensive experiments on the ImageNet dataset demonstrate the effectiveness of our method, e.g., attacking nine state-of-the-art defense models with an average success rate of 95.4%. Our code is available in https://github.com/yuyang-long/SSA.



Paperid:159
Authors:Xiaojun Jia; Yong Zhang; Xingxing Wei; Baoyuan Wu; Ke Ma; Jue Wang; Xiaochun Cao
Title: Prior-Guided Adversarial Initialization for Fast Adversarial Training
Abstract:
Fast adversarial training (FAT) effectively improves the efficiency of standard adversarial training (SAT). However, initial FAT encounters catastrophic overfitting, i.e., the robust accuracy against adversarial attacks suddenly decreases to 0% during training. Though several FAT variants spare no effort to prevent overfitting, they sacrifice much calculation cost. In this paper, we explore the difference between the training processes of SAT and FAT and observe that the attack success rate of adversarial examples (AEs) of FAT gets worse gradually in the late training stage, resulting in overfitting. The AEs are generated by the fast gradient sign method (FGSM) with a zero or random initialization. Based on the observation, we propose a prior-guided FGSM initialization method to avoid overfitting after investigating several initialization strategies, improving the quality of the AEs during the whole training process. The initialization is formed by leveraging historically generated AEs without additional calculation cost. We further provide a theoretical analysis for the proposed initialization method. Moreover, we also propose a simple yet effective regularizer based on the prior guided initialization, i.e., the currently generated perturbation should not deviate too much from the prior-guided initialization. The regularizer adopts both historical and current adversarial perturbations to guide the model learning. Evaluations on four datasets demonstrate that the proposed method can prevent catastrophic overfitting and outperform state-of-the-art FAT methods at a low computational cost.



Paperid:160
Authors:Shiji Zhao; Jie Yu; Zhenlong Sun; Bo Zhang; Xingxing Wei
Title: Enhanced Accuracy and Robustness via Multi-Teacher Adversarial Distillation
Abstract:
Adversarial training is an effective approach for improving the robustness of deep neural networks against adversarial attacks. Although bringing reliable robustness, adversarial training (AT) will reduce the performance of identifying clean examples. Meanwhile, Adversarial training can bring more robustness for large models than small models. To improve the robust and clean accuracy of small models, we introduce the Multi-Teacher Adversarial Robustness Distillation (MTARD) to guide the adversarial training process of small models. Specifically, MTARD uses multiple large teacher models, including an adversarial teacher and a clean teacher to guide a small student model in the adversarial training by knowledge distillation. In addition, we design a dynamic training algorithm to balance the influence between the adversarial teacher and clean teacher models. A series of experiments demonstrate that our MTARD can outperform the state-of-the-art adversarial training and distillation methods against various adversarial attacks.



Paperid:161
Authors:Martin Gubri; Maxime Cordy; Mike Papadakis; Yves Le Traon; Koushik Sen
Title: LGV: Boosting Adversarial Example Transferability from Large Geometric Vicinity
Abstract:
We propose transferability from Large Geometric Vicinity (LGV), a new technique to increase the transferability of black-box adversarial attacks. LGV starts from a pretrained surrogate model and collects multiple weight sets from a few additional training epochs with a constant and high learning rate. LGV exploits two geometric properties that we relate to transferability. First, models that belong to a wider weight optimum are better surrogates. Second, we identify a subspace able to generate an effective surrogate ensemble among this wider optimum. Through extensive experiments, we show that LGV alone outperforms all (combinations of) four established test-time transformations by 1.8 to 59.9 percentage points. Our findings shed new light on the importance of the geometry of the weight space to explain the transferability of adversarial examples.



Paperid:162
Authors:Siyuan Liang; Longkang Li; Yanbo Fan; Xiaojun Jia; Jingzhi Li; Baoyuan Wu; Xiaochun Cao
Title: A Large-Scale Multiple-Objective Method for Black-Box Attack against Object Detection
Abstract:
Recent studies have shown that detectors based on deep models are vulnerable to adversarial examples, even in the black-box scenario where the attacker cannot access the model information. Most existing attack methods aim to minimize the true positive rate, which often shows poor attack performance, as another sub-optimal bounding box may be detected around the attacked bounding box to be the new true positive one. To settle this challenge, we propose to minimize the true positive rate and maximize the false positive rate, which can encourage more false positive objects to block the generation of new true positive bounding boxes. It is modeled as a multi-objective optimization (MOP) problem, of which the generic algorithm can search the Pareto-optimal. However, our task has more than two million decision variables, leading to low searching efficiency. Thus, we extend the standard \textbf{G}enetic \textbf{A}lgorithm with \textbf{R}andom \textbf{S}ubset selection and \textbf{D}ivide-and-\textbf{C}onquer, called GARSDC, which significantly improves the efficiency. Moreover, to alleviate the sensitivity to population quality in generic algorithms, we generate a gradient-prior initial population, utilizing the transferability between different detectors with similar backbones. Compared with the state-of-art attack methods, GARSDC decreases by an average 12.0 in the mAP and queries by about 1000 times in extensive experiments. Our codes can be found at https://github.com/LiangSiyuan21/GARSDC.



Paperid:163
Authors:Jianhong Pan; Qichen Zheng; Zhipeng Fan; Hossein Rahmani; Qiuhong Ke; Jun Liu
Title: GradAuto: Energy-Oriented Attack on Dynamic Neural Networks
Abstract:
Dynamic neural networks could adapt their structures or parameters based on different inputs. By reducing the computation redundancy for certain samples, it can greatly improve the computational efficiency without compromising the accuracy. In this paper, we investigate the robustness of dynamic neural networks against energy-oriented attacks. We present a novel algorithm, named GradAuto, to attack both dynamic depth and dynamic width models, where dynamic depth networks reduce redundant computation by skipping some intermediate layers while dynamic width networks adaptively activate a subset of neurons in each layer. Our GradAuto carefully adjusts the direction and the magnitude of the gradients to efficiently find an almost imperceptible perturbation for each input, which will activate more computation units during inference. In this way, GradAuto effectively boosts the computational cost of models with dynamic architectures. Compared to previous energy-oriented attack techniques, GradAuto obtains the state-of-the-art result and recovers 100\% dynamic network reduced FLOPs on average for both dynamic depth and dynamic width models. Furthermore, we demonstrate that GradAuto offers us great control over the attacking process and could serve as one of the keys to unlock the potential of the energy-oriented attack.



Paperid:164
Authors:Jiachen Sun; Akshay Mehra; Bhavya Kailkhura; Pin-Yu Chen; Dan Hendrycks; Jihun Hamm; Z. Morley Mao
Title: A Spectral View of Randomized Smoothing under Common Corruptions: Benchmarking and Improving Certified Robustness
Abstract:
Certified robustness guarantee gauges a model’s resistance to test-time attacks and can assess the model’s readiness for deployment in the real world. In this work, we explore a new problem setting to critically examine how the adversarial robustness guarantees change when state-of-the-art randomized smoothing-based certifications encounter common corruptions of the test data. Our analysis demonstrates a previously unknown vulnerability of these certifiably robust models to low-frequency corruptions such as weather changes, rendering these models unfit for deployment in the wild. To alleviate this issue, we propose a novel data augmentation scheme, FourierMix, that produces augmentations to improve the spectral coverage of the training data. Furthermore, we propose a new regularizer that encourages consistent predictions on noise perturbations of the augmented data to improve the quality of the smoothed models. We show that FourierMix helps eliminate the spectral bias of certifiably robust models, enabling them to achieve significantly better certified robustness on a range of corruption benchmarks. Our evaluation also uncovers the inability of current corruption benchmarks to highlight the spectral biases of the models. To this end, we propose a comprehensive benchmarking suite that contains corruptions from different regions in the spectral domain. Evaluation of models trained with popular augmentation methods on the proposed suite unveils their spectral biases. It also establishes the superiority of FourierMix trained models in achieving stronger certified robustness guarantees under corruptions over the entire frequency spectrum.



Paperid:165
Authors:Guanlin Li; Guowen Xu; Han Qiu; Ruan He; Jiwei Li; Tianwei Zhang
Title: Improving Adversarial Robustness of 3D Point Cloud Classification Models
Abstract:
3D point cloud classification models based on deep neural networks were proven to be vulnerable to adversarial examples, with a quantity of novel attack techniques proposed by researchers recently. It is of paramount importance to preserve the robustness of 3D models under adversarial environments, considering their broad application in safety- and security-critical tasks. Unfortunately, existing defenses are not general enough to satisfactorily mitigate all types of attacks. In this paper, we design two innovative methodologies to improve the adversarial robustness of 3D point cloud classification models. (1) We introduce CCN, a novel point cloud architecture which can smooth and disrupt the adversarial perturbations. (2) We propose AMS, a novel data augmentation strategy to adaptively balance the model usability and robustness. Extensive evaluations indicate the integration of the two techniques provides much more robustness than existing defense solutions for 3D classification models. Our code can be found in https://github.com/GuanlinLee/CCNAMS.



Paperid:166
Authors:Xian Wei; Yangyu Xu; Yanhui Huang; Hairong Lv; Hai Lan; Mingsong Chen; Xuan Tang
Title: Learning Extremely Lightweight and Robust Model with Differentiable Constraints on Sparsity and Condition Number
Abstract:
Learning lightweight and robust deep learning models is an enormous challenge for safety-critical devices with limited computing and memory resources, owing to robustness against adversarial attacks being proportional to network capacity. The community has extensively explored the integration of adversarial training and model compression, such as weight pruning. However, lightweight models generated by highly pruned over-parameterized models lead to sharp drops in both robust and natural accuracy. It has been observed that the parameters of these models lie in ill-conditioned weight space, i.e., the condition number of weight matrices tend to be large enough that the model is not robust. In this work, we propose a framework for building extremely lightweight models, which combines tensor product with the differentiable constraints for reducing condition number and promoting sparsity. Moreover, the proposed framework is incorporated into adversarial training with the min-max optimization scheme. We evaluate the proposed approach on VGG-16 and Visual Transformer. Experimental results on datasets such as ImageNet, SVHN, and CIFAR-10 show that we can achieve an overwhelming advantage at a high compression ratio, e.g., 200 times.



Paperid:167
Authors:Huy Phan; Cong Shi; Yi Xie; Tianfang Zhang; Zhuohang Li; Tianming Zhao; Jian Liu; Yan Wang; Yingying Chen; Bo Yuan
Title: RIBAC: Towards Robust and Imperceptible Backdoor Attack against Compact DNN
Abstract:
Recently backdoor attack has become an emerging threat to the security of deep neural network (DNN) models. To date, most of the existing studies focus on backdoor attack against the uncompressed model; while the vulnerability of compressed DNNs, which are widely used in the practical applications, is little exploited yet. In this paper, we propose to study and develop Robust and Imperceptible Backdoor Attack against Compact DNN models (RIBAC). By performing systematic analysis and exploration on the important design knobs, we propose a framework that can learn the proper trigger patterns, model parameters and pruning masks in an efficient way. Thereby achieving high trigger stealthiness, high attack success rate and high model efficiency simultaneously. Extensive evaluations across different datasets, including the test against the state-of-the-art defense mechanisms, demonstrate the high robustness, stealthiness and model efficiency of RIBAC. Code is available at https://github.com/huyvnphan/ECCV2022-RIBAC"



Paperid:168
Authors:Xiao Yang; Yinpeng Dong; Tianyu Pang; Hang Su; Jun Zhu
Title: Boosting Transferability of Targeted Adversarial Examples via Hierarchical Generative Networks
Abstract:
Transfer-based adversarial attacks can evaluate model robustness in the black-box setting. Several methods have demonstrated impressive untargeted transferability, however, it is still challenging to efficiently produce targeted transferability. To this end, we develop a simple yet effective framework to craft targeted transfer-based adversarial examples, applying a hierarchical generative network. In particular, we contribute to amortized designs that well adapt to multi-class targeted attacks. Extensive experiments on ImageNet show that our method improves the success rates of targeted black-box attacks by a significant margin over the existing methods --- it reaches an average success rate of 29.1% against six diverse models based only on one substitute white-box model, which significantly outperforms the state-of-the-art gradient-based attack methods. Moreover, the proposed method is also more efficient beyond an order of magnitude than gradient-based methods.



Paperid:169
Authors:Zheng Yuan; Jie Zhang; Shiguang Shan
Title: Adaptive Image Transformations for Transfer-Based Adversarial Attack
Abstract:
Adversarial attacks provide a good way to study the robustness of deep learning models. One category of methods in transfer-based black-box attack utilizes several image transformation operations to improve the transferability of adversarial examples, which is effective, but fails to take the specific characteristic of the input image into consideration. In this work, we propose a novel architecture, called Adaptive Image Transformation Learner (AITL), which incorporates different image transformation operations into a unified framework to further improve the transferability of adversarial examples. Unlike the fixed combinational transformations used in existing works, our elaborately designed transformation learner adaptively selects the most effective combination of image transformations specific to the input image. Extensive experiments on ImageNet demonstrate that our method significantly improves the attack success rates on both normally trained models and defense models under various settings.



Paperid:170
Authors:Xiaoming Zhao; Fangchang Ma; David Güera; Zhile Ren; Alexander G. Schwing; Alex Colburn
Title: Generative Multiplane Images: Making a 2D GAN 3D-Aware
Abstract:
What is really needed to make an existing 2D GAN 3Daware? To answer this question, we modify a classical GAN, i.e., StyleGANv2, as little as possible. We find that only two modifications are absolutely necessary: 1) a multiplane image style generator branch which produces a set of alpha maps conditioned on their depth; 2) a pose conditioned discriminator. We refer to the generated output as a ‘generative multiplane image’ (GMPI) and emphasize that its renderings are not only high-quality but also guaranteed to be view-consistent, which makes GMPIs different from many prior works. Importantly, the number of alpha maps can be dynamically adjusted and can differ between training and inference, alleviating memory concerns and enabling fast training of GMPIs in less than half a day at a resolution of 1024^2. Our findings are consistent across three challenging and common high-resolution datasets, including FFHQ, AFHQv2 and MetFaces.



Paperid:171
Authors:Yulong Cao; Chaowei Xiao; Anima Anandkumar; Danfei Xu; Marco Pavone
Title: AdvDO: Realistic Adversarial Attacks for Trajectory Prediction
Abstract:
Trajectory prediction is essential for autonomous vehicles (AVs) to plan correct and safe driving behaviors. While many prior works aim to achieve higher prediction accuracy, few studies the adversarial robustness of their methods. To bridge this gap, we propose to study the adversarial robustness of data-driven trajectory prediction systems. We devise an optimization-based adversarial attack framework that leverages a carefully-designed differentiable dynamic model to generate realistic adversarial trajectories. Empirically, we benchmark the adversarial robustness of state-of-the-art prediction models and show that our attack increases the prediction error for both general metrics and planning-aware metrics by more than 50% and 37%. We also show that our attack can lead an AV to drive off-road or collide into other vehicles in simulation. Finally, we demonstrate how to mitigate the adversarial attacks using an adversarial training scheme.



Paperid:172
Authors:Qiying Yu; Jieming Lou; Xianyuan Zhan; Qizhang Li; Wangmeng Zuo; Yang Liu; Jingjing Liu
Title: Adversarial Contrastive Learning via Asymmetric InfoNCE
Abstract:
Contrastive learning (CL) has recently been applied to adversarial learning tasks. Such practice considers adversarial perturbations as additional positive samples of an instance, and by maximizing their agreements with each other, yields better adversarial robustness. However, this mechanism can be potentially flawed, since adversarial perturbations may cause instance-level identity confusion, which can impede CL performance by pulling together different instances with separate identities. To address this issue, we propose to treat adversarial samples unequally when contrasted to positive and negative samples, with an asymmetric InfoNCE objective (A-InfoNCE) that allows discriminating considerations of adversarial samples. Specifically, adversaries are viewed as inferior positives that induce weaker learning signals, or as hard negatives exhibiting higher contrast to other negative samples. In the asymmetric fashion, the adverse impacts of conflicting objectives between CL and adversarial learning can be effectively mitigated. Experiments show that our approach consistently outperforms existing Adversarial CL methods across different finetuning schemes without additional computational cost. The proposed A-InfoNCE is also a generic form that can be readily extended to different CL methods. Code is available at https://github.com/yqy2001/A-InfoNCE.



Paperid:173
Authors:Shuo Yang; Chang Xu
Title: One Size Does NOT Fit All: Data-Adaptive Adversarial Training
Abstract:
Adversarial robustness is critical for deep learning models to defend against adversarial attacks. Although adversarial training is considered to be one of the most effective ways to improve the model’s adversarial robustness, it usually yields models with lower natural accuracy. In this paper, we argue that, for the attackable examples, traditional adversarial training which utilizes a fixed size perturbation ball can create adversarial examples that deviate far away from the original class towards the target class. Thus, the model’s performance on the natural target class will drop drastically, which leads to the decline of natural accuracy. To this end, we propose the Data-Adaptive Adversarial Training (DAAT) which adaptively adjusts the perturbation ball to a proper size for each of the natural examples with the help of a natural trained calibration network. Besides, a dynamic training strategy empowers the DAAT models with impressive robustness while retaining remarkable natural accuracy. Based on a toy example, we theoretically prove the recession of the natural accuracy caused by adversarial training and show how the data-adaptive perturbation size helps the model resist it. Finally, empirical experiments on benchmark datasets demonstrate the significant improvement of DAAT models on natural accuracy compared with strong baselines.



Paperid:174
Authors:Hanbin Hong; Binghui Wang; Yuan Hong
Title: UniCR: Universally Approximated Certified Robustness via Randomized Smoothing
Abstract:
We study certified robustness of machine learning classifiers against adversarial perturbations. In particular, we propose the first universally approximated certified robustness (UniCR) framework, which can approximate the robustness certification of \emph{any} input on \emph{any} classifier against \emph{any} $\ell_p$ perturbations with noise generated by \emph{any} continuous probability distribution. Compared with the state-of-the-art certified defenses, UniCR provides many significant benefits: (1) the first universal robustness certification framework for the above 4 “any”s; (2) automatic robustness certification that avoids case-by-case analysis, (3) tightness validation of certified robustness, and (4) optimality validation of noise distributions used by randomized smoothing. We conduct extensive experiments to validate the above benefits of UniCR and the advantages of UniCR over state-of-the-art certified defenses against $\ell_p$ perturbations.



Paperid:175
Authors:Jiawang Bai; Kuofeng Gao; Dihong Gong; Shu-Tao Xia; Zhifeng Li; Wei Liu
Title: Hardly Perceptible Trojan Attack against Neural Networks with Bit Flips
Abstract:
The security of deep neural networks (DNNs) has attracted increasing attention due to their widespread use in various applications. Recently, the deployed DNNs have been demonstrated to be vulnerable to Trojan attacks, which manipulate model parameters with bit flips to inject a hidden behavior and activate it by a specific trigger pattern. However, all existing Trojan attacks adopt noticeable patch-based triggers (e.g., a square pattern), making them perceptible to humans and easy to be spotted by machines. In this paper, we present a novel attack, namely hardly perceptible Trojan attack (HPT). HPT crafts hardly perceptible Trojan images by utilizing the additive noise and per-pixel flow field to tweak the pixel values and positions of the original images, respectively. To achieve superior attack performance, we propose to jointly optimize bit flips, additive noise, and flow field. Since the weight bits of the DNNs are binary, this problem is very hard to be solved. We handle the binary constraint with equivalent replacement and provide an effective optimization algorithm. Extensive experiments on CIFAR-10, SVHN, and ImageNet datasets show that the proposed HPT can generate hardly perceptible Trojan images, while achieving comparable or better attack performance compared to the state-of-the-art methods. The code is available at: https://github.com/jiawangbai/HPT.



Paperid:176
Authors:Yaguan Qian; Shenghui Huang; Bin Wang; Xiang Ling; Xiaohui Guan; Zhaoquan Gu; Shaoning Zeng; Wujie Zhou; Haijiang Wang
Title: Robust Network Architecture Search via Feature Distortion Restraining
Abstract:
The vulnerability of DNNs severely limits the application of it in the security-sensitive domains. Most of the existing methods improve the robustness of models from weight optimization, such as adversarial training and regularization. However, the architecture is also a key factor to robustness, which is often neglected or underestimated. We propose Robust Network Architecture Search (RNAS) to obtain a robust network against adversarial attacks. We observe that an adversarial perturbation distorting the non-robust features in latent feature space can further aggravate misclassification. Based on this observation, we search the robust architecture through restricting feature distortion in the search process. Specifically, we define a network vulnerability metric based on feature distortion as a constraint in the search process. This process is modeled as a multi-objective bilevel optimization problem and an effective algorithm is proposed to solve this optimization. Extensive experiments conducted on CIFAR-10/100, SVHN and Tiny-ImageNet show that RNAS achieves the best robustness under various adversarial attacks compared with extensive baselines and state-of-the-art methods.



Paperid:177
Authors:Zhuowen Yuan; Fan Wu; Yunhui Long; Chaowei Xiao; Bo Li
Title: SecretGen: Privacy Recovery on Pre-trained Models via Distribution Discrimination
Abstract:
Transfer learning through the use of pre-trained models has become a growing trend for the machine learning community. Consequently, numerous pre-trained models are released online to facilitate further research. However, it raises extensive concerns on whether these pre-trained models would leak privacy-sensitive information of their training data. Thus, in this work, we aim to answer the following questions: ""Can we effectively recover private information from these pre-trained models? What are the sufficient conditions to retrieve such sensitive information?"" We first explore different statistical information which can discriminate the private training distribution from other distributions. Based on our observations, we propose a novel private data reconstruction framework, SecretGen, to effectively recover private information. Compared with previous methods which can recover private data with the ground true prediction of the targeted recovery instance, SecretGen does not require such prior knowledge, making it more practical. We conduct extensive experiments on different datasets under diverse scenarios to compare SecretGen with other baselines and provide a systematic benchmark to better understand the impact of different auxiliary information and optimization operations. We show that without prior knowledge about true class prediction, SecretGen is able to recover private data with similar performance compared with the ones that leverage such prior knowledge. If the prior knowledge is given, SecretGen will significantly outperform baseline methods. We also propose several quantitative metrics to further quantify the privacy vulnerability of pre-trained models, which will help the model selection for privacy-sensitive applications. Our code is available at: https://github.com/AI-secure/SecretGen.



Paperid:178
Authors:Xiaosen Wang; Zeliang Zhang; Kangheng Tong; Dihong Gong; Kun He; Zhifeng Li; Wei Liu
Title: Triangle Attack: A Query-Efficient Decision-Based Adversarial Attack
Abstract:
Decision-based attack poses a severe threat to real-world applications since it regards the target model as a black box and only accesses the hard prediction label. Great efforts have been made recently to decrease the number of queries; however, existing decision-based attacks still require thousands of queries in order to generate good quality adversarial examples. In this work, we find that a benign sample, the current and the next adversarial examples can naturally construct a triangle in a subspace for any iterative attacks. Based on the law of sines, we propose a novel Triangle Attack (TA) to optimize the perturbation by utilizing the geometric information that the longer side is always opposite the larger angle in any triangle. However, directly applying such information on the input image is ineffective because it cannot thoroughly explore the neighborhood of the input sample in the high dimensional space. To address this issue, TA optimizes the perturbation in the low frequency space for effective dimensionality reduction owing to the generality of such geometric property. Extensive evaluations on ImageNet dataset show that TA achieves a much higher attack success rate within 1,000 queries and needs a much less number of queries to achieve the same attack success rate under various perturbation budgets than existing decision-based attacks. With such high efficiency, we further validate the applicability of TA on real-world API, i.e., Tencent Cloud API.



Paperid:179
Authors:Runkai Zheng; Rongjun Tang; Jianze Li; Li Liu
Title: Data-Free Backdoor Removal Based on Channel Lipschitzness
Abstract:
Recent studies have shown that Deep Neural Networks (DNNs) are vulnerable to the backdoor attacks, which leads to malicious behaviors of DNNs when specific triggers are attached to the input images. It was further demonstrated that the infected DNNs possess a collection of channels, which are more sensitive to the backdoor triggers compared with normal channels. Pruning these channels was then shown to be effective in mitigating the backdoor behaviors. To locate those channels, it is natural to consider their Lipschitzness, which measures their sensitivity against worst-case perturbations on the inputs. In this work, we introduce a novel concept called Channel Lipschitz Constant (CLC), which is defined as the Lipschitz constant of the mapping from the input images to the output of each channel. Then we provide empirical evidences to show the strong correlation between an Upper bound of the CLC (UCLC) and the trigger-activated change on the channel activation. Since UCLC can be directly calculated from the weight matrices, we can detect the potential backdoor channels in a data-free manner, and do simple pruning on the infected DNN to repair the model. The proposed Channel Lipschitzness based Pruning (CLP) method is super fast, simple, data-free and robust to the choice of the pruning threshold. Extensive experiments are conducted to evaluate the efficiency and effectiveness of CLP, which achieves state-of-the-art results among the mainstream defense methods even without any data. Source codes are available at https://github.com/rkteddy/channel-Lipschitzness-based-pruning.



Paperid:180
Authors:Yixu Wang; Jie Li; Hong Liu; Yan Wang; Yongjian Wu; Feiyue Huang; Rongrong Ji
Title: Black-Box Dissector: Towards Erasing-Based Hard-Label Model Stealing Attack
Abstract:
Previous studies have verified that the functionality of black-box models can be stolen with full probability outputs. However, under the more practical hard-label setting, we observe that existing methods suffer from catastrophic performance degradation. We argue this is due to the lack of rich information in the probability prediction and the overfitting caused by hard labels. To this end, we propose a novel hard-label model stealing method termed black-box dissector, which consists of two erasing-based modules. One is a CAM-driven erasing strategy that is designed to increase the information capacity hidden in hard labels from the victim model. The other is a random-erasing-based self-knowledge distillation module that utilizes soft labels from the substitute model to mitigate overfitting. Extensive experiments on four widely-used datasets consistently demonstrate that our method outperforms state-of-the-art methods, with an improvement of at most 8.27%. We also validate the effectiveness and practical potential of our method on real-world APIs and defense methods. Furthermore, our method promotes other downstream tasks, i.e., transfer adversarial attacks.



Paperid:181
Authors:Xuwang Yin; Shiying Li; Gustavo K. Rohde
Title: Learning Energy-Based Models with Adversarial Training
Abstract:
We study a new approach to learning energy-based models (EBMs) based on adversarial training (AT). We show that (binary) AT learns a special kind of energy function that models the support of the data distribution, and the learning process is closely related to MCMC-based maximum likelihood learning of EBMs. We further propose improved techniques for generative modeling with AT, and demonstrate that this new approach is capable of generating diverse and realistic images. Aside from having competitive image generation performance to explicit EBMs, the studied approach is stable to train, is well-suited for image translation tasks, and exhibits strong out-of-distribution adversarial robustness. Our results demonstrate the viability of the AT approach to generative modeling, suggesting that AT is a competitive alternative approach to learning EBMs.



Paperid:182
Authors:Ganlin Liu; Xiaowei Huang; Xinping Yi
Title: Adversarial Label Poisoning Attack on Graph Neural Networks via Label Propagation
Abstract:
Graph neural networks (GNNs) have achieved outstanding performance in semi-supervised learning tasks with partially labeled graph structured data. However, labeling graph data for training is a challenging task, and inaccurate labels may mislead the training process to erroneous GNN models for node classification. In this paper, we consider label poisoning attacks on training data, where the labels of input data are modified by an adversary before training, to understand to what extent the state-of-the-art GNN models are resistant/vulnerable to such attacks. Specifically, we propose a label poisoning attack framework for graph convolutional networks (GCNs), inspired by the equivalence between label propagation and decoupled GCNs that separate message passing from neural networks. Instead of attacking the entire GCN models, we propose to attack solely label propagation for message passing. It turns out that a gradient-based attack on label propagation is effective and efficient towards the misleading of GCN training. More remarkably, such label attack can be topology-agnostic in the sense that the labels to be attacked can be efficiently chosen without knowing graph structures. Extensive experimental results demonstrate the effectiveness of the proposed method against state-of-the-art GCN-like models.



Paperid:183
Authors:Ali Dabouei; Fariborz Taherkhani; Sobhan Soleymani; Nasser M. Nasrabadi
Title: Revisiting Outer Optimization in Adversarial Training
Abstract:
Despite the fundamental distinction between adversarial and natural training (AT and NT), AT methods generally adopt momentum SGD (MSGD) for the outer optimization. This paper aims to analyze this choice by investigating the overlooked role of outer optimization in AT. Our exploratory evaluations reveal that AT induces higher gradient norm and variance compared to NT. This phenomenon hinders the outer optimization in AT since the convergence rate of MSGD is highly dependent on the variance of the gradients. To this end, we propose an optimization method called ENGM which regularizes the contribution of each input example to the average mini-batch gradients. We prove that the convergence rate of ENGM is independent of the variance of the gradients, and thus, it is suitable for AT. We introduce a trick to reduce the computational cost of ENGM using empirical observations on the correlation between the norm of gradients w.r.t. the network parameters and input examples. Our extensive evaluations and ablation studies on CIFAR-10, CIFAR-100, and TinyImageNet demonstrate that ENGM and its variants consistently improve the performance of a wide range of AT methods. Furthermore, ENGM alleviates major shortcomings of AT including robust overfitting and high sensitivity to hyperparameter settings.



Paperid:184
Authors:Nasim Shafiee; Ehsan Elhamifar
Title: Zero-Shot Attribute Attacks on Fine-Grained Recognition Models
Abstract:
Zero-shot fine-grained recognition is an important classification task, whose goal is to recognize visually very similar classes, including the ones without training images. Despite recent advances on the development of zero-shot fine-grained recognition methods, the robustness of such models to adversarial attacks is not well understood. On the other hand, adversarial attacks have been widely studied for conventional classification with visually distinct classes. Such attacks, in particular, universal perturbations that are class-agnostic and ideally should generalize to unseen classes, however, cannot leverage or capture small distinctions among fine-grained classes. Therefore, we propose a compositional attribute-based framework for generating adversarial attacks on zero-shot fine-grained recognition models. To generate attacks that capture small differences between fine-grained classes, generalize well to previously unseen classes and can be applied in real-time, we propose to learn and compose multiple attribute-based universal perturbations (AUPs). Each AUP corresponds to an image-agnostic perturbation on a specific attribute. To build our attack, we compose AUPs with weights obtained by learning a class-attribute compatibility function. To learn the AUPs and the parameters of our model, we minimize a loss, consisting of a ranking loss and a novel utility loss, which ensures AUPs are effectively learned and utilized. By extensive experiments on three datasets for zero-shot fine-grained recognition, we show that our attacks outperform conventional universal classification attacks and transfer well between different recognition architectures.



Paperid:185
Authors:Kien Do; Haripriya Harikumar; Hung Le; Dung Nguyen; Truyen Tran; Santu Rana; Dang Nguyen; Willy Susilo; Svetha Venkatesh
Title: Towards Effective and Robust Neural Trojan Defenses via Input Filtering
Abstract:
Trojan attacks on deep neural networks are both dangerous and surreptitious. Over the past few years, Trojan attacks have advanced from using only a single input-agnostic trigger and targeting only one class to using multiple, input-specific triggers and targeting multiple classes. However, Trojan defenses have not caught up with this development. Most defense methods still make out-of-date assumptions about Trojan triggers and target classes, thus, can be easily circumvented by modern Trojan attacks. To deal with this problem, we propose two novel ""filtering"" defenses called Variational Input Filtering (VIF) and Adversarial Input Filtering (AIF) which leverage lossy data compression and adversarial learning respectively to effectively purify all potential Trojan triggers in the input at run time without making assumptions about the number of triggers/target classes or the input dependence property of triggers. In addition, we introduce a new defense mechanism called ""Filtering-then-Contrasting"" (FtC) which helps avoid the drop in classification accuracy on clean data caused by ""filtering"", and combine it with VIF/AIF to derive new defenses of this kind. Extensive experimental results and ablation studies show that our proposed defenses significantly outperform well-known baseline defenses in mitigating five advanced Trojan attacks including two recent state-of-the-art while being quite robust to small amounts of training data and large-norm triggers.



Paperid:186
Authors:Sravanti Addepalli; Samyak Jain; Gaurang Sriramanan; R. Venkatesh Babu
Title: Scaling Adversarial Training to Large Perturbation Bounds
Abstract:
The vulnerability of Deep Neural Networks to Adversarial Attacks has fuelled research towards building robust models. While most Adversarial Training algorithms aim at defending attacks constrained within low magnitude Lp norm bounds, real-world adversaries are not limited by such constraints. In this work, we aim to achieve adversarial robustness within larger bounds, against perturbations that may be perceptible, but do not change human (or Oracle) prediction. The presence of images that flip Oracle predictions and those that do not makes this a challenging setting for adversarial robustness. We discuss the ideal goals of an adversarial defense algorithm beyond perceptual limits, and further highlight the shortcomings of naively extending existing training algorithms to higher perturbation bounds. In order to overcome these shortcomings, we propose a novel defense, Oracle-Aligned Adversarial Training (OA-AT), to align the predictions of the network with that of an Oracle during adversarial training. The proposed approach achieves state-of-the-art performance at large epsilon bounds (such as an L-inf bound of 16/255 on CIFAR-10) while outperforming existing defenses (AWP, TRADES, PGD-AT) at standard bounds (8/255) as well.



Paperid:187
Authors:Hoang Tran; Dan Lu; Guannan Zhang
Title: Exploiting the Local Parabolic Landscapes of Adversarial Losses to Accelerate Black-Box Adversarial Attack
Abstract:
Existing black-box adversarial attacks on image classifiers update the perturbation at each iteration from only a small number of queries of the loss function. Since the queries contain very limited information about the loss, black-box methods usually require much more queries than white-box methods. We propose to improve the query efficiency of black-box methods by exploiting the smoothness of the local loss landscape. However, many adversarial losses are not locally smooth with respect to pixel perturbations. To resolve this issue, our first contribution is to theoretically and experimentally justify that the adversarial losses of many standard and robust image classifiers behave like parabolas with respect to perturbations in the Fourier domain. Our second contribution is to exploit the parabolic landscape to build a quadratic approximation of the loss around the current state, and use this approximation to interpolate the loss value as well as update the perturbation without additional queries. Since the local region is already informed by the quadratic fitting, we use large perturbation steps to explore far areas. We demonstrate the efficiency of our method on MNIST, CIFAR-10 and ImageNet datasets for various standard and robust models, as well as on Google Cloud Vision. The experimental results show that exploiting the loss landscape can help significantly reduce the number of queries and increase the success rate. Our codes are available at https://github.com/HoangATran/BABIES.



Paperid:188
Authors:Qianyu Zhou; Ke-Yue Zhang; Taiping Yao; Ran Yi; Kekai Sheng; Shouhong Ding; Lizhuang Ma
Title: Generative Domain Adaptation for Face Anti-Spoofing
Abstract:
Face anti-spoofing (FAS) approaches based on unsupervised domain adaption (UDA) have drawn growing attention due to promising performances for target scenarios. Most existing UDA FAS methods typically fit the trained models to the target domain via aligning the distribution of semantic high-level features. However, insufficient supervision of unlabeled target domains and neglect of low-level feature alignment degrade the performances of existing methods. To address these issues, we propose a novel perspective of UDA FAS that directly fits the target data to the models, i.e., stylizes the target data to the source-domain style via image translation, and further feeds the stylized data into the well-trained source model for classification. The proposed Generative Domain Adaptation (GDA) framework combines two carefully designed consistency constraints: 1) Inter-domain neural statistic consistency guides the generator in narrowing the inter-domain gap. 2) Dual-level semantic consistency ensures the semantic quality of stylized images. Besides, we propose intra-domain spectrum mixup to further expand target data distributions to ensure generalization and reduce the intra-domain gap. Extensive experiments and visualizations demonstrate the effectiveness of our method against the state-of-the-art methods.



Paperid:189
Authors:Huanzhang Dou; Pengyi Zhang; Wei Su; Yunlong Yu; Xi Li
Title: MetaGait: Learning to Learn an Omni Sample Adaptive Representation for Gait Recognition
Abstract:
Gait recognition, which aims at identifying individuals by their walking patterns, has recently drawn increasing research attention. However, gait recognition still suffers from the conflicts between the limited binary visual clues of the silhouette and numerous covariates with diverse scales, which brings challenges to the model’s adaptiveness. In this paper, we address this conflict by developing a novel MetaGait that learns to learn an omni sample adaptive representation. Towards this goal, MetaGait injects meta-knowledge, which could guide the model to perceive sample-specific properties, into the calibration network of the attention mechanism to improve the adaptiveness from the omni-scale, omni-dimension, and omni-process perspectives. Specifically, we leverage the meta-knowledge across the entire process, where Meta Triple Attention and Meta Temporal Pooling are presented respectively to adaptively capture omni-scale dependency from spatial/channel/temporal dimensions simultaneously and to adaptively aggregate temporal information through integrating the merits of three complementary temporal aggregation methods. Extensive experiments demonstrate the state-of-the-art performance of the proposed MetaGait. On CASIA-B, we achieve rank-1 accuracy of 98.7%, 96.0%, and 89.3% under three conditions, respectively. On OU-MVLP, we achieve rank-1 accuracy of 92.4%.



Paperid:190
Authors:Junhao Liang; Chao Fan; Saihui Hou; Chuanfu Shen; Yongzhen Huang; Shiqi Yu
Title: GaitEdge: Beyond Plain End-to-End Gait Recognition for Better Practicality
Abstract:
Gait is one of the most promising biometrics to identify individuals at a long distance. Although most previous methods have focused on recognizing the silhouettes, several end-to-end methods that extract gait features directly from RGB images perform better. However, we demonstrate that these end-to-end methods may inevitably suffer from the gait-irrelevant noises, i.e. low-level texture and color information. Experimentally, we design the cross-domain evaluation to support this view. In this work, we propose a novel end-to-end framework named GaitEdge which can effectively block gait-irrelevant information and release end-to-end training potential. Specifically, GaitEdge synthesizes the output of the pedestrian segmentation network and then feeds it to the subsequent recognition network, where the synthetic silhouettes consist of trainable edges of bodies and fixed interiors to limit the information that the recognition network receives. Besides, GaitAlign for aligning silhouettes is embedded into the GaitEdge without losing differentiability. Experimental results on CASIA-B and our newly built TTG-200 indicate that GaitEdge significantly outperforms the previous methods and provides a more practical end-to-end paradigm. All the source code are available at https://github.com/ShiqiYu/OpenGait.



Paperid:191
Authors:Wanyi Zhuang; Qi Chu; Zhentao Tan; Qiankun Liu; Haojie Yuan; Changtao Miao; Zixiang Luo; Nenghai Yu
Title: UIA-ViT: Unsupervised Inconsistency-Aware Method Based on Vision Transformer for Face Forgery Detection
Abstract:
Intra-frame inconsistency has been proved to be effective for the generalization of face forgery detection. However, learning to focus on these inconsistency requires extra pixel-level forged location annotations. Acquiring such annotations is non-trivial. Some existing methods generate large-scale synthesized data with location annotations, which is time-consuming. Others generate forgery location labels by subtracting paired real and fake images, yet such paired data is difficult to collected and the generated label is usually discontinuous. To overcome these limitations, we propose a novel Unsupervised Inconsistency-Aware method based on Vision Transformer, called UIA-ViT. Thanks to the self-attention mechanism, the attention map among patch embeddings naturally represents the consistency relation, making the vision Transformer suitable for the consistency representation learning. Specifically, we propose two key components: Unsupervised Patch Consistency Learning (UPCL) and Progressive Consistency Weighted Assemble (PCWA). UPCL is designed for learning the consistency-related representation with progressive optimized pseudo annotations that are derived from multivariate Gaussian estimation. PCWA enhances the final classification embedding with previous patch embeddings optimized by UPCL to further improve the detection performance. Extensive experiments demonstrate the effectiveness of the proposed method.



Paperid:192
Authors:Wentian Zhang; Haozhe Liu; Feng Liu; Raghavendra Ramachandra; Christoph Busch
Title: Effective Presentation Attack Detection Driven by Face Related Task
Abstract:
The robustness and generalization ability of Presentation Attack Detection (PAD) methods is critical to ensure the security of Face Recognition Systems (FRSs). However, in a real scenario, Presentation Attacks (PAs) are various and it is hard to predict the Presentation Attack Instrument (PAI) species that will be used by the attacker. Existing PAD methods are highly dependent on the limited training set and cannot generalize well to unknown PAI species. Unlike this specific PAD task, other face related tasks trained by huge amount of real faces (e.g. face recognition and attribute editing) can be effectively adopted into different application scenarios. Inspired by this, we propose to trade position of PAD and face related work in a face system and apply the free acquired prior knowledge from face related tasks to solve face PAD, so as to improve the generalization ability in detecting PAs. The proposed method, first introduces task specific features from other face related task, then, we design a Cross-Modal Adapter using a Graph Attention Network (GAT) to re-map such features to adapt to PAD task. Finally, face PAD is achieved by using the hierarchical features from a CNN-based PA detector and the re-mapped features. The experimental results show that the proposed method can achieve significant improvements in the complicated and hybrid datasets, when compared with the state-of-the-art methods. In particular, when training on the datasets OULU-NPU, CASIA-FASD, and Idiap Replay-Attack, we obtain HTER (Half Total Error Rate) of 5.48% for the testing dataset MSU-MFSD, outperforming the baseline by 7.39%. The source code is available at https://github.com/WentianZhang-ML/FRT-PAD.



Paperid:193
Authors:Haoyu Ma; Zhe Wang; Yifei Chen; Deying Kong; Liangjian Chen; Xingwei Liu; Xiangyi Yan; Hao Tang; Xiaohui Xie
Title: PPT: Token-Pruned Pose Transformer for Monocular and Multi-View Human Pose Estimation
Abstract:
Recently, the vision transformer and its variants have played an increasingly important role in both monocular and multi-view human pose estimation. Considering image patches as tokens, transformers can model the global dependencies within the entire image or across images from other views. However, global attention is computationally expensive. As a consequence, it is difficult to scale up these transformer-based methods to high-resolution features and many views. In this paper, we propose the token-Pruned Pose Transformer (PPT) for 2D human pose estimation, which can locate a rough human mask and performs self-attention only within selected tokens. Furthermore, we extend our PPT to multi-view human pose estimation. Built upon PPT, we propose a new cross-view fusion strategy, called human area fusion, which considers all human foreground pixels as corresponding candidates. Experimental results on COCO and MPII demonstrate that our PPT can match the accuracy of previous pose transformer methods while reducing the computation. Moreover, experiments on Human 3.6M and Ski-Pose demonstrate that our Multi-view PPT can efficiently fuse cues from multiple views and achieve new state-of-the-art results. Source code and trained model can be found at \url{https://github.com/HowieMa/PPT}.



Paperid:194
Authors:Jiaxi Jiang; Paul Streli; Huajian Qiu; Andreas Fender; Larissa Laich; Patrick Snape; Christian Holz
Title: AvatarPoser: Articulated Full-Body Pose Tracking from Sparse Motion Sensing
Abstract:
Today’s Mixed Reality head-mounted displays track the user’s head pose in world space as well as the user’s hands for interaction in both Augmented Reality and Virtual Reality scenarios. While this is adequate to support user input, it unfortunately limits users’ virtual representations to just their upper bodies. Current systems thus resort to floating avatars, whose limitation is particularly evident in collaborative settings. To estimate full-body poses from the sparse input sources, prior work has incorporated additional trackers and sensors at the pelvis or lower body, which increases setup complexity and limits practical application in mobile settings. In this paper, we present AvatarPoser, the first learning-based method that predicts full-body poses in world coordinates using only motion input from the user’s head and hands. Our method builds on a Transformer encoder to extract deep features from the input signals and decouples global motion from the learned local joint orientations to guide pose estimation. To obtain accurate full-body motions that resemble motion capture animations, we refine the arm joints’ positions using an optimization routine with inverse kinematics to match the original tracking input. In our evaluation, AvatarPoser achieved new state-of-the-art results in evaluations on large motion capture datasets (AMASS). At the same time, our method’s inference speed supports real-time operation, providing a practical interface to support holistic avatar control and representation for Metaverse applications.



Paperid:195
Authors:Wenkang Shan; Zhenhua Liu; Xinfeng Zhang; Shanshe Wang; Siwei Ma; Wen Gao
Title: P-STMO: Pre-trained Spatial Temporal Many-to-One Model for 3D Human Pose Estimation
Abstract:
This paper introduces a novel Pre-trained Spatial Temporal Many-to-One (P-STMO) model for 2D-to-3D human pose estimation task. To reduce the difficulty of capturing spatial and temporal information, we divide this task into two stages: pre-training (Stage I) and fine-tuning (Stage II). In Stage I, a self-supervised pre-training sub-task, termed masked pose modeling, is proposed. The human joints in the input sequence are randomly masked in both spatial and temporal domains. A general form of denoising auto-encoder is exploited to recover the original 2D poses and the encoder is capable of capturing spatial and temporal dependencies in this way. In Stage II, the pre-trained encoder is loaded to STMO model and fine-tuned. The encoder is followed by a many-to-one frame aggregator to predict the 3D pose in the current frame. Especially, an MLP block is utilized as the spatial feature extractor in STMO, which yields better performance than other methods. In addition, a temporal downsampling strategy is proposed to diminish data redundancy. Extensive experiments on two benchmarks show that our method outperforms state-of-the-art methods with fewer parameters and less computational overhead. For example, our P-STMO model achieves 42.1mm MPJPE on Human3.6M dataset when using 2D poses from CPN as inputs. Meanwhile, it brings a 1.5-7.1× speedup to state-of-the-art methods. Code is available at https://github.com/paTRICK-swk/P-STMO.



Paperid:196
Authors:Jiefeng Li; Siyuan Bian; Chao Xu; Gang Liu; Gang Yu; Cewu Lu
Title: D\&D: Learning Human Dynamics from Dynamic Camera
Abstract:
3D human pose estimation from a monocular video has recently seen significant improvements. However, most state-of-the-art methods are kinematics-based, which are prone to physically implausible motions with pronounced artifacts. Current dynamics-based methods can predict physically plausible motion but are restricted to simple scenarios with static camera view. In this work, we present D&D (Learning Human Dynamics from Dynamic Camera), which leverages the laws of physics to reconstruct 3D human motion from the in-the-wild videos with a moving camera. D&D introduces inertial force control (IFC) to explain the 3D human motion in the non-inertial local frame by considering the inertial forces of the dynamic camera. To learn the ground contact with limited annotations, we develop probabilistic contact torque (PCT), which is computed by differentiable sampling from contact probabilities and used to generate motions. The contact state can be weakly supervised by encouraging the model to generate correct motions. Furthermore, we propose an attentive PD controller that adjusts target pose states using temporal information to obtain smooth and accurate pose control. Our approach is entirely neural-based and runs without offline optimization or simulation in physics engines. Experiments on large-scale 3D human motion benchmarks demonstrate the effectiveness of D&D, where we exhibit superior performance against both state-of-the-art kinematics-based and dynamics-based methods. Code is available at https://github.com/Jeff-sjtu/DnD"



Paperid:197
Authors:Qihao Liu; Yi Zhang; Song Bai; Alan Yuille
Title: Explicit Occlusion Reasoning for Multi-Person 3D Human Pose Estimation
Abstract:
Occlusion poses a great threat to monocular multi-person 3D human pose estimation due to large variability in terms of the shape, appearance, and position of occluders. While existing methods try to handle occlusion with pose priors/constraints, data augmentation, or implicit reasoning, they still fail to generalize to unseen poses or occlusion cases and may make large mistakes when multiple people are present. Inspired by the remarkable ability of humans to infer occluded joints from visible cues, we develop a method to explicitly model this process that significantly improves bottom-up multi-person human pose estimation with or without occlusions. First, we split the task into two subtasks: visible keypoints detection and occluded keypoints reasoning, and propose a Deeply Supervised Encoder Distillation (DSED) network to solve the second one. To train our model, we propose a Skeleton-guided human Shape Fitting (SSF) approach to generate pseudo occlusion labels on the existing datasets, enabling explicit occlusion reasoning. Experiments show that explicitly learning from occlusions improves human pose estimation. In addition, exploiting feature-level information of visible joints allows us to reason about occluded joints more accurately. Our method outperforms both the state-of-the-art top-down and bottom-up methods on several benchmarks.



Paperid:198
Authors:Xiaohan Zhang; Bharat Lal Bhatnagar; Sebastian Starke; Vladimir Guzov; Gerard Pons-Moll
Title: COUCH: Towards Controllable Human-Chair Interactions
Abstract:
Humans can interact with an object in the scene in many different ways, which are often associated with different modalities of contacting with the object. This creates a highly complex motion space that can be difficult to learn, particularly when synthesizing such human interactions in a controllable manner. Existing works on synthesizing human scene interaction focus on the high-level control of interacting with a particular object without considering fine-grained control of limb motion variations within one task. In this work, we drive this direction and study the problem of synthesizing scene interactions conditioned on a wide range of contact positions on the object. We pick human-chair interactions as an example. We propose a novel synthesis framework COUCH, which firstly plans ahead the motion by predicting contact-aware control signals of the hands, which are then used to synthesize contact-conditioned interactions. Furthermore, we contribute a large human-chair interaction dataset with clean annotations, the COUCH Dataset. Our method shows consistent quantitative and qualitative improvements over existing methods for learning human-object interactions.



Paperid:199
Authors:Deying Kong; Linguang Zhang; Liangjian Chen; Haoyu Ma; Xiangyi Yan; Shanlin Sun; Xingwei Liu; Kun Han; Xiaohui Xie
Title: Identity-Aware Hand Mesh Estimation and Personalization from RGB Images
Abstract:
Reconstructing 3D hand meshes from monocular RGB images has attracted increasing amount of attention due to its enormous potential applications in the field of AR/VR. Most state-of-the-art methods attempt to tackle this task in an anonymous manner. Specifically, the identity of the subject is ignored even though it is practically available in real applications where the user is unchanged in a continuous recording session. In this paper, we propose an identity-aware hand mesh estimation model, which can incorporate the identity information represented by the intrinsic shape parameters of the subject. We demonstrate the importance of the identity information by comparing the proposed identity-aware model to a baseline which treats subject anonymously. Furthermore, to handle the use case where the test subject is unseen, we propose a novel personalization pipeline to calibrate the intrinsic shape parameters using only a few unlabeled RGB images of the subject. Experiments on two large scale public datasets validate the state-of-the-art performance of our proposed method.



Paperid:200
Authors:Cunlin Wu; Yang Xiao; Boshen Zhang; Mingyang Zhang; Zhiguo Cao; Joey Tianyi Zhou
Title: C3P: Cross-Domain Pose Prior Propagation for Weakly Supervised 3D Human Pose Estimation
Abstract:
This paper first proposes and solves weakly supervised 3D human pose estimation (HPE) problem in point cloud, via propagating the pose prior within unlabelled RGB-point cloud sequence to 3D domain. Our approach termed C3P does not require any labor-consuming 3D keypoint annotation for training. To this end, we propose to transfer 2D HPE annotation information within the existing large-scale RGB datasets (e.g., MS COCO) to 3D task, using unlabelled RGB-point cloud sequence easy to acquire for linking 2D and 3D domains. The self-supervised 3D HPE clues within point cloud sequence are also exploited, concerning spatial-temporal constraints on human body symmetry, skeleton length and joints’ motion. And, a refined point set network structure for weakly supervised 3D HPE is proposed in encoder-decoder manner. The experiments on CMU Panoptic and ITOP datasets demonstrate that, our method can achieve the comparable results to the 3D fully supervised state-of-the-art counterparts. When large-scale unlabelled data (e.g., NTU RGB+D 60) is used, our approach can even outperform them under the more challenging cross-setup test setting. The source code is released at ""https://github.com/wucunlin/C3P"" for research use only.



Paperid:201
Authors:Garvita Tiwari; Dimitrije Antić; Jan Eric Lenssen; Nikolaos Sarafianos; Tony Tung; Gerard Pons-Moll
Title: Pose-NDF: Modeling Human Pose Manifolds with Neural Distance Fields
Abstract:
We present Pose-NDF, a continuous model for plausible human poses based on neural distance fields (NDFs). Pose or motion priors are important for generating realistic new poses and for reconstructing accurate poses from noisy or partial observations. Pose-NDF learns a manifold of plausible poses as the zero level set of a neural implicit function, extending the idea of modelling implicit surfaces in 3D to the high dimensional domain SO(3)K, where a human pose is defined by a single data point, represented by K quaternions. The resulting high dimensional implicit function can be differentiated with respect to the input poses and thus can be used to project arbitrary poses onto the manifold by using gradient descent on the set of 3-dimensional hyperspheres. In contrast to previous VAE-based human pose priors, which transform the pose space into a Gaussian distribution, we model the actual pose manifold, preserving the distances between poses. We demonstrate that this approach and thus, Pose-NDF, outperforms existing state-of-the-art methods as a prior in various downstream tasks, ranging from denoising real world human mocap data, pose recovery from occluded data to 3D pose reconstruction from images. Furthermore, we show that it can be used to generate more diverse poses by random sampling and projection than VAE based methods. We will release our code and pre-trained model for further research.



Paperid:202
Authors:Zhihao Li; Jianzhuang Liu; Zhensong Zhang; Songcen Xu; Youliang Yan
Title: CLIFF: Carrying Location Information in Full Frames into Human Pose and Shape Estimation
Abstract:
Top-down methods dominate the field of 3D human pose and shape estimation, because they are decoupled from human detection and allow researchers to focus on the core problem. However, cropping, their first step, discards the location information from the very beginning, which makes themselves unable to accurately predict the global rotation in the original camera coordinate system. To address this problem, we propose to Carry Location Information in Full Frames (CLIFF) into this task. Specifically, we feed more holistic features to CLIFF by concatenating the cropped image feature with its bounding box information. We calculate the 2D reprojection loss with a broader view of the full frame, taking a projection process similar to that of the person projected in the image. Fed and supervised by global-location-aware information, CLIFF directly predicts the global rotation along with more accurate articulated poses. Besides, we propose a pseudo-ground-truth annotator based on CLIFF, which provides high-quality 3D annotations for in-the-wild 2D datasets and offers crucial full supervision for regression-based methods. Extensive experiments on popular benchmarks show that CLIFF outperforms prior arts by a significant margin, and reaches the first place on the AGORA leaderboard (the SMPL-Algorithms track).



Paperid:203
Authors:Ailing Zeng; Xuan Ju; Lei Yang; Ruiyuan Gao; Xizhou Zhu; Bo Dai; Qiang Xu
Title: DeciWatch: A Simple Baseline for 10× Efficient 2D and 3D Pose Estimation
Abstract:
This paper proposes a simple baseline framework for video-based 2D/3D human pose estimation that can achieve 10 times efficiency improvement over existing works without any performance degradation, named DeciWatch. Unlike current solutions that estimate each frame in a video, DeciWatch introduces a simple yet effective sample-denoise-recover framework that only watches sparsely sampled frames, taking advantage of the continuity of human motions and the lightweight pose representation. Specifically, DeciWatch uniformly samples less than 10% video frames for detailed estimation, denoises the estimated 2D/3D poses with an efficient Transformer architecture, and then accurately recovers the rest of the frames using another Transformer-based network. Comprehensive experimental results on three video-based human pose estimation and body mesh recovery tasks with four datasets validate the efficiency and effectiveness of DeciWatch. Code is available at https://github.com/cure-lab/DeciWatch.



Paperid:204
Authors:Ailing Zeng; Lei Yang; Xuan Ju; Jiefeng Li; Jianyi Wang; Qiang Xu
Title: SmoothNet: A Plug-and-Play Network for Refining Human Poses in Videos
Abstract:
When analyzing human motion videos, the output jitters from existing pose estimators are highly-unbalanced with varied estimation errors across frames. Most frames in a video are relatively easy to estimate and only suffer from slight jitters. In contrast, for rarely seen or occluded actions, the estimated positions of multiple joints largely deviate from the ground truth values for a consecutive sequence of frames, rendering significant jitters on them. To tackle this problem, we propose to attach a dedicated temporal-only refinement network to existing pose estimators for jitter mitigation, named SmoothNet. Unlike existing learning-based solutions that employ spatio-temporal models to co-optimize per-frame precision and temporal smoothness at all the joints, SmoothNet models the natural smoothness characteristics in body movements by learning the long-range temporal relations of every joint without considering the noisy correlations among joints. With a simple yet effective motion-aware fully-connected network, SmoothNet improves the temporal smoothness of existing pose estimators significantly and enhances the estimation accuracy of those challenging frames as a side-effect. Moreover, as a temporal-only model, a unique advantage of SmoothNet is its strong transferability across various types of estimators and datasets. Comprehensive experiments on five datasets with eleven popular backbone networks across 2D and 3D pose estimation and body recovery tasks demonstrate the efficacy of the proposed solution. Code is available at https://github.com/cure-lab/SmoothNet.



Paperid:205
Authors:Wentao Jiang; Sheng Jin; Wentao Liu; Chen Qian; Ping Luo; Si Liu
Title: PoseTrans: A Simple yet Effective Pose Transformation Augmentation for Human Pose Estimation
Abstract:
Human pose estimation aims to accurately estimate a wide variety of human poses. However, existing datasets often follow a long-tailed distribution that unusual poses only occupy a small portion, which further leads to the lack of diversity of rare poses. These issues result in the inferior generalization ability of current pose estimators. In this paper, we present a simple yet effective data augmentation method, termed Pose Transformation (PoseTrans), to alleviate the aforementioned problems. Specifically, we propose Pose Transformation Module (PTM) to create new training samples that have diverse poses and adopt a pose discriminator to ensure the plausibility of the augmented poses. Besides, we propose Pose Clustering Module (PCM) to measure the pose rarity and select the “rarest” poses to help balance the long-tailed distribution. Extensive experiments on three benchmark datasets demonstrate the effectiveness of our method, especially on rare poses. Also, our method is efficient and simple to implement, which can be easily integrated into the training pipeline of existing pose estimation models.



Paperid:206
Authors:Junuk Cha; Muhammad Saqlain; GeonU Kim; Mingyu Shin; Seungryul Baek
Title: Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement
Abstract:
Estimating 3D poses and shapes in the form of meshes from monocular RGB images is challenging. Obviously, it is more difficult than estimating 3D poses only in the form of skeletons or heatmaps. When interacting persons are involved, the 3D mesh reconstruction becomes more challenging due to the ambiguity introduced by person-to-person occlusions. To tackle the challenges, we propose a coarse-to-fine pipeline that benefits from 1) inverse kinematics from the occlusion-robust 3D skeleton estimation and 2) transformer-based relation-aware refinement techniques. In our pipeline, we first obtain occlusion-robust 3D skeletons for multiple persons from an RGB image. Then, we apply inverse kinematics to convert the estimated skeletons to deformable 3D mesh parameters. Finally, we apply the transformer-based mesh refinement that refines the obtained mesh parameters considering intra- and inter-person relations of 3D meshes. Via extensive experiments, we demonstrate the effectiveness of our method, outperforming state-of-the-arts on 3DPW, MuPoTS and AGORA datasets.



Paperid:207
Authors:Xiaoning Sun; Qiongjie Cui; Huaijiang Sun; Bin Li; Weiqing Li; Jianfeng Lu
Title: Overlooked Poses Actually Make Sense: Distilling Privileged Knowledge for Human Motion Prediction
Abstract:
Previous works on human motion prediction follow the pattern of building a mapping relation between the sequence observed and the one to be predicted. However, due to the inherent complexity of multivariate time series data, it still remains a challenge to find the extrapolation relation between motion sequences. In this paper, we present a new prediction pattern, which introduces previously overlooked human poses, to implement the prediction task from the view of interpolation. These poses exist after the predicted sequence, and form the privileged sequence. To be specific, we first propose an InTerPolation learning Network (ITP-Network) that encodes both the observed sequence and the privileged sequence to interpolate the in-between predicted sequence, wherein the embedded Privileged-sequence-Encoder (Priv-Encoder) learns the privileged knowledge (PK) simultaneously. Then, we propose a Final Prediction Network (FP-Network) for which the privileged sequence is not observable, but is equipped with a novel PK-Simulator that distills PK learned from the previous network. This simulator takes as input the observed sequence, but approximates the behavior of Priv-Encoder, enabling FP-Network to imitate the interpolation process. Extensive experimental results demonstrate that our prediction pattern achieves state-of the-art performance on benchmarked H3.6M, CMU-Mocap and 3DPW datasets in both short-term and long-term predictions.



Paperid:208
Authors:Zhuo Chen; Xu Zhao; Xiaoyue Wan
Title: Structural Triangulation: A Closed-Form Solution to Constrained 3D Human Pose Estimation
Abstract:
We propose Structural Triangulation, a closed-form solution for optimal 3D human pose considering multi-view 2D pose estimations, calibrated camera parameters, and bone lengths. To start with, we focus on embedding structural constraints of human body in the process of 2D-to-3D inference using triangulation. Assume bone lengths are known in prior, then the inference process is formulated as a constrained optimization problem. By proper approximation, the closed-form solution to this problem is achieved. Further, we generalize our method with Step Constraint Algorithm to help converge when large error occurs in 2D estimations. In experiment, public datasets (Human3.6M and Total Capture) and synthesized data are used for evaluation. Our method achieves state-of-the-art results on Human3.6M Dataset when bone lengths are known and competitive results when they are not. The generality and efficiency of our method are also demonstrated.



Paperid:209
Authors:Sheng Ye; Yu-Hui Wen; Yanan Sun; Ying He; Ziyang Zhang; Yaoyuan Wang; Weihua He; Yong-Jin Liu
Title: Audio-Driven Stylized Gesture Generation with Flow-Based Model
Abstract:
Generating stylized audio-driven gestures for robots and virtual avatars has attracted increasing considerations recently. Existing methods require style labels (e.g. speaker identities), or complex preprocessing of the data to obtain style control parameters. In this paper, we propose a new end-to-end flow-based model, which can generate audio-driven gestures of arbitrary styles without the preprocessing procedure and style labels. To achieve this goal, we introduce a global encoder and a gesture perceptual loss into the classic generative flow model to capture both the global and local information. We conduct extensive experiments on two benchmark datasets: the TED Dataset and the Trinity Dataset. Both quantitative and qualitative evaluations show that the proposed model outperforms state-of-the-art models.



Paperid:210
Authors:Zhehan Kan; Shuoshuo Chen; Zeng Li; Zhihai He
Title: Self-Constrained Inference Optimization on Structural Groups for Human Pose Estimation
Abstract:
We observe that human poses exhibit strong group-wise structural correlation and spatial coupling between keypoints due to the biological constraints of different body parts. This group-wise structural correlation can be explored to improve the accuracy and robustness of human pose estimation. In this work, we develop a self-constrained prediction-verification network to characterize and learn the structural correlation between keypoints during training. During the inference stage, the feedback information from the verification network allows us to perform further optimization of pose prediction, which significantly improves the performance of human pose estimation. Specifically, we partition the keypoints into groups according to the biological structure of human body. Within each group, the keypoints are further partitioned into two subsets, high-confidence base keypoints and low-confidence terminal keypoints. We develop a self-constrained prediction-verification network to perform forward and backward predictions between these keypoint subsets. One fundamental challenge in pose estimation, as well as in generic prediction tasks, is that there is no mechanism for us to verify if the obtained pose estimation or prediction results are accurate or not, since the ground truth is not available. Once successfully learned, the verification network serves as an accuracy verification module for the forward pose prediction. During the inference stage, it can be used to guide the local optimization of the pose estimation results of low-confidence keypoints with the self-constrained loss on high-confidence keypoints as the objective function. Our extensive experimental results on benchmark MS COCO and CrowdPose datasets demonstrate that the proposed method can significantly improve the pose estimation results.



Paperid:211
Authors:Hiroyasu Akada; Jian Wang; Soshi Shimada; Masaki Takahashi; Christian Theobalt; Vladislav Golyanik
Title: UnrealEgo: A New Dataset for Robust Egocentric 3D Human Motion Capture
Abstract:
We present UnrealEgo, a new large-scale naturalistic dataset for egocentric 3D human pose estimation. UnrealEgo is based on an advanced concept of eyeglasses equipped with two fisheye cameras that can be used in unconstrained environments. We design their virtual prototype and attach them to 3D human models for stereo view capture. We next generate a large corpus of human motions. As a consequence, UnrealEgo is the first dataset to provide in-the-wild stereo images with the largest variety of motions among existing egocentric datasets. Furthermore, we propose a new benchmark method with a simple but effective idea of devising a 2D keypoint estimation module for stereo inputs to improve 3D human pose estimation. The extensive experiments show that our approach outperforms the previous state-of-the-art methods qualitatively and quantitatively. UnrealEgo and our source codes are available on our project web page.



Paperid:212
Authors:Maosen Li; Siheng Chen; Zijing Zhang; Lingxi Xie; Qi Tian; Ya Zhang
Title: Skeleton-Parted Graph Scattering Networks for 3D Human Motion Prediction
Abstract:
Graph convolutional network based methods that model the body joints’ relations, have recently shown great promise in 3D skeleton-based human motion prediction. However, these methods have two critical issues: first, deep graph convolutions filter features within only limited graph spectrum band, losing sufficient information in the full band; second, using a single graph to model the whole body underestimates the diverse patterns in various body-parts. To address the first issue, we propose adaptive graph scattering, which leverages multiple trainable band-pass graph filters to decompose pose features into various graph spectrum bands to provide richer information, promoting more comprehensive feature extraction. To address the second issue, body parts are modeled separately to learn diverse dynamics, which enables finer feature extraction along the spatial dimensions. Integrating the above two designs, we propose a novel skeleton-parted graph scattering network (SPGSN). The cores of the model are cascaded multi-part graph scattering blocks (MPGSBs), which build adaptive graph scattering for large-band graph filtering on diverse body-parts, as well as fuse the decomposed features based on the inferred spectrum importance and body-part interactions. Extensive experiments have shown that SPGSN outperforms state-of-the-art methods by remarkable margins of 13.8%, 9.3% and 2.7% in terms of 3D mean per joint position error (MPJPE) on Human3.6M, CMU Mocap and 3DPW datasets, respectively.



Paperid:213
Authors:William McNally; Kanav Vats; Alexander Wong; John McPhee
Title: Rethinking Keypoint Representations: Modeling Keypoints and Poses as Objects for Multi-Person Human Pose Estimation
Abstract:
In keypoint estimation tasks such as human pose estimation, heatmap-based regression is the dominant approach despite possessing notable drawbacks: heatmaps intrinsically suffer from quantization error and require excessive computation to generate and post-process. Motivated to find a more efficient solution, we propose to model individual keypoints and sets of spatially related keypoints (i.e., poses) as objects within a dense single-stage anchor-based detection framework. Hence, we call our method KAPAO (pronounced ""Ka-Pow""), for Keypoints And Poses As Objects. KAPAO is applied to the problem of single-stage multi-person human pose estimation by simultaneously detecting human pose and keypoint objects and fusing the detections to exploit the strengths of both object representations. In experiments, we observe that KAPAO is faster and more accurate than previous methods, which suffer greatly from heatmap post-processing. The accuracy-speed trade-off is especially favourable in the practical setting when not using test-time augmentation. Source code: https://github.com/wmcnally/kapao.



Paperid:214
Authors:Jiajun Su; Chunyu Wang; Xiaoxuan Ma; Wenjun Zeng; Yizhou Wang
Title: VirtualPose: Learning Generalizable 3D Human Pose Models from Virtual Data
Abstract:
While monocular 3D pose estimation seems to have achieved very accurate results on the public datasets, their generalization ability is largely overlooked. In this work, we perform a systematic evaluation of the existing methods and find that they get notably larger errors when tested on different cameras, human poses and appearance. To address the problem, we introduce VirtualPose, a two-stage learning framework to exploit the hidden ""free lunch"" specific to this task, i.e. generating infinite number of poses and cameras for training models at no cost. To that end, the first stage transforms images to abstract geometry representations (AGR), and then the second maps them to 3D poses. It addresses the generalization issue from two aspects: (1) the first stage can be trained on diverse 2D datasets to reduce the risk of over-fitting to limited appearance; (2) the second stage can be trained on diverse AGR synthesized from a large number of virtual cameras and poses. It outperforms the SOTA methods without using any paired images and 3D poses from the benchmarks, which paves the way for practical applications. Code is available at https://github.com/wkom/VirtualPose.



Paperid:215
Authors:Weian Mao; Yongtao Ge; Chunhua Shen; Zhi Tian; Xinlong Wang; Zhibin Wang; Anton van den Hengel
Title: Poseur: Direct Human Pose Regression with Transformers
Abstract:
We propose a direct, regression-based approach to 2D human pose estimation from single images. We formulate the problem as a sequence prediction task, which we solve using a Transformer network. This network directly learns a regression mapping from images to the keypoint coordinates, without resorting to intermediate representations such as heatmaps. This approach avoids much of the complexity associated with heatmap-based approaches. To overcome the feature misalignment issues of previous regression-based methods, we propose an attention mechanism that adaptively attends to the features that are most relevant to the target keypoints, considerably improving the accuracy. Importantly, our framework is end-to-end differentiable, and naturally learns to exploit the dependencies between keypoints. Experiments on MS-COCO and MPII, two predominant pose-estimation datasets, demonstrate that our method significantly improves upon the state-of-the-art in regression-based pose estimation. More notably, ours is the first regression-based approach to perform favorably compared to the best heatmap-based pose estimation methods. Code is available at: https://github.com/aim-uofa/Poseur"



Paperid:216
Authors:Yanjie Li; Sen Yang; Peidong Liu; Shoukui Zhang; Yunxiao Wang; Zhicheng Wang; Wankou Yang; Shu-Tao Xia
Title: SimCC: A Simple Coordinate Classification Perspective for Human Pose Estimation
Abstract:
The 2D heatmap-based approaches have dominated Human Pose Estimation (HPE) for years due to high performance. However, the long-standing quantization error problem in the 2D heatmap-based methods leads to several well-known drawbacks: 1) The performance for the low-resolution inputs is limited; 2) To improve the feature map resolution for higher localization precision, multiple costly upsampling layers are required; 3) Extra post-processing is adopted to reduce the quantization error. To address these issues, we aim to explore a brand new scheme, called SimCC, which reformulates HPE as two classification tasks for horizontal and vertical coordinates. The proposed SimCC uniformly divides each pixel into several bins, thus achieving sub-pixel localization precision and low quantization error. Benefiting from that, SimCC can omit additional refinement post-processing and exclude upsampling layers under certain settings, resulting in a more simple and effective pipeline for HPE. Extensive experiments conducted over COCO, CrowdPose, and MPII datasets show that SimCC outperforms heatmap-based counterparts, especially in low-resolution settings by a large margin. Code is now publicly available at https://github.com/leeyegy/SimCC.



Paperid:217
Authors:Haixin Wang; Lu Zhou; Yingying Chen; Ming Tang; Jinqiao Wang
Title: Regularizing Vector Embedding in Bottom-Up Human Pose Estimation
Abstract:
The embedding-based method such as Associative Embedding is popular in bottom-up human pose estimation. Methods under this framework group candidate keypoints according to the predicted identity embeddings. However, the identity embeddings of different instances are likely to be linearly inseparable in some complex scenes, such as crowded scene or when the number of instances in the image is large. To reduce the impact of this phenomenon on keypoint grouping, we try to learn a sparse multidimensional embedding for each keypoint. We observe that the different dimensions of embeddings are highly linearly correlated. To address this issue, we impose an additional constraint on the embeddings during training phase. Based on the fact that the scales of instances usually have significant variations, we uilize the scales of instances to regularize the embeddings, which effectively reduces the linear correlation of embeddings and makes embeddings being sparse. We evaluate our model on CrowdPose Test and COCO Test-dev. Compared to vanilla Associative Embedding, our method has an impressive superiority in keypoint grouping, especially in crowded scenes with a large number of instances. Furthermore, our method achieves state-of-the-art results on CrowdPose Test (74.5 AP) and COCO Test-dev (72.8 AP), outperforming other bottom-up methods. Our code and pretrained models are available at https://github.com/CR320/CoupledEmbedding.



Paperid:218
Authors:Jiaxin Guo; Fangxun Zhong; Rong Xiong; Yun-Hui Liu; Yue Wang; Yiyi Liao
Title: A Visual Navigation Perspective for Category-Level Object Pose Estimation
Abstract:
This paper studies category-level object pose estimation based on a single monocular image. Recent advances in pose-aware generative models have paved the way for addressing this challenging task using analysis-by-synthesis. The idea is to sequentially update a set of latent variables,e.g., pose, shape, and appearance, of the generative model until the generated image best agrees with the observation. However, convergence and efficiency are two challenges of this inference procedure. In this paper, we take a deeper look at the inference of analysis-by-synthesis from the perspective of visual navigation, and investigate what is a good navigation policy for this specific task. We evaluate three different strategies, including gradient descent, reinforcement learning and imitation learning, via thorough comparisons in terms of convergence, robustness and efficiency. Moreover, we show that a simple hybrid approach leads to an effective and efficient solution. We further compare these strategies to state-of-the-art methods, and demonstrate superior performance on synthetic and real-world datasets leveraging off-the-shelf pose-aware generative models.



Paperid:219
Authors:Hang Ye; Wentao Zhu; Chunyu Wang; Rujie Wu; Yizhou Wang
Title: Faster VoxelPose: Real-Time 3D Human Pose Estimation by Orthographic Projection
Abstract:
While the voxel-based methods have achieved promising results for multi-person 3D pose estimation from multi-cameras, they suffer from heavy computation burdens, especially for large scenes. We present Faster VoxelPose to address the challenge by re-projecting the feature volume to the three two-dimensional coordinate planes and estimating X, Y, Z coordinates from them separately. To that end, we first localize each person by a 3D bounding box by estimating a 2D box and its height based on the volume features projected to the xy-plane and z-axis, respectively. Then for each person, we estimate partial joint coordinates from the three coordinate planes separately which are then fused to obtain the final 3D pose. The method is free from costly 3D-CNNs and improves the speed of VoxelPose by ten times and meanwhile achieves competitive accuracy as the state-of-the-art methods, proving its potential in real-time applications.



Paperid:220
Authors:Vasileios Choutas; Federica Bogo; Jingjing Shen; Julien Valentin
Title: Learning to Fit Morphable Models
Abstract:
Fitting parametric models of human bodies, hands or faces to sparse input signals in an accurate, robust, and fast manner has the promise of significantly improving immersion in AR and VR scenarios. A common first step in systems that tackle these problems is to regress the parameters of the parametric model directly from the input data. This approach is fast, robust, and is a good starting point for an iterative minimization algorithm. The latter searches for the minimum of an energy function, typically composed of a data term and priors that encode our knowledge about the problem’s structure. While this is undoubtedly a very successful recipe, priors are often hand defined heuristics and finding the right balance between the different terms to achieve high quality results is a non-trivial task. Furthermore, converting and optimizing these systems to run in a performant way requires custom implementations that demand significant time investments from both engineers and domain experts. In this work, we build upon recent advances in learned optimization and propose an update rule inspired by the classic Levenberg-Marquardt algorithm. We show the effectiveness of the proposed neural optimizer on three problems, 3D body estimation from a head-mounted device, 3D body estimation from sparse 2D keypoints and face surface estimation from dense 2D landmarks. Our method can easily be applied to new model fitting problems and offers a competitive alternative to well-tuned ‘traditional’ model fitting pipelines, both in terms of accuracy and speed.



Paperid:221
Authors:Siwei Zhang; Qianli Ma; Yan Zhang; Zhiyin Qian; Taein Kwon; Marc Pollefeys; Federica Bogo; Siyu Tang
Title: EgoBody: Human Body Shape and Motion of Interacting People from Head-Mounted Devices
Abstract:
Understanding social interactions from egocentric views is crucial for many applications, ranging from assistive robotics to AR/VR. Key to reasoning about interactions is to understand the body pose and motion of the interaction partner from the egocentric view. However, research in this area is severely hindered by the lack of datasets. Existing datasets are limited in terms of either size, capture/annotation modalities, ground-truth quality, or interaction diversity. We fill this gap by proposing EgoBody, a novel large-scale dataset for human pose, shape and motion estimation from egocentric views, during interactions in complex 3D scenes. We employ Microsoft HoloLens2 headsets to record rich egocentric data streams (including RGB, depth, eye gaze, head and hand tracking). To obtain accurate 3D ground truth, we calibrate the headset with a multi-Kinect rig and fit expressive SMPL-X body meshes to multi-view RGB-D frames, reconstructing 3D human shapes and poses relative to the scene, over time. We collect 68 sequences, spanning diverse interaction scenarios, and propose the first benchmark for 3D full-body pose and shape estimation of the social partner from egocentric views. We extensively evaluate state-of-the-art methods, highlight their limitations in the egocentric scenario, and address such limitations leveraging our high-quality annotations. Data and code will be available for research purposes.



Paperid:222
Authors:Dylan Turpin; Liquan Wang; Eric Heiden; Yun-Chun Chen; Miles Macklin; Stavros Tsogkas; Sven Dickinson; Animesh Garg
Title: Grasp'D: Differentiable Contact-Rich Grasp Synthesis for Multi-Fingered Hands
Abstract:
The study of hand-object interaction requires generating viable grasp poses for high-dimensional multi-finger models, often relying on analytic grasp synthesis which tends to produce brittle and unnatural results. This paper presents Grasp’D, an approach to grasp synthesis by differentiable contact simulation that can work with both known models and visual inputs. We use gradient-based methods as an alternative to sampling-based grasp synthesis, which fails without simplifying assumptions, such as pre-specified contact locations and eigengrasps. Such assumptions limit grasp discovery and, in particular, exclude high-contact power grasps. In contrast, our simulation-based approach allows for stable, efficient, physically realistic, high-contact grasp synthesis, even for gripper morphologies with high-degrees of freedom. We identify and address challenges in making grasp simulation amenable to gradient-based optimization, such as non-smooth object surface geometry, contact sparsity, and a rugged optimization landscape. Grasp’D compares favorably to analytic grasp synthesis on human and robotic hand models, and resultant grasps achieve over 4× denser contact, leading to significantly higher grasp stability. Video and code available at: graspd-eccv22.github.io.



Paperid:223
Authors:Ziqian Bai; Timur Bagautdinov; Javier Romero; Michael Zollhöfer; Ping Tan; Shunsuke Saito
Title: AutoAvatar: Autoregressive Neural Fields for Dynamic Avatar Modeling
Abstract:
Neural fields such as implicit surfaces have recently enabled avatar modeling from raw scans without explicit temporal correspondences. In this work, we exploit autoregressive modeling to further extend this notion to capture dynamic effects, such as soft-tissue deformations. Although autoregressive models are naturally capable of handling dynamics, it is non-trivial to apply them to implicit representations, as explicit state decoding is infeasible due to prohibitive memory requirements. In this work, for the first time, we enable autoregressive modeling of implicit avatars. To reduce the memory bottleneck and efficiently model dynamic implicit surfaces, we introduce the notion of articulated observer points, which relate implicit states to the explicit surface of a parametric human body model. We demonstrate that encoding implicit surfaces as a set of height fields defined on articulated observer points leads to significantly better generalization compared to a latent representation. The experiments show that our approach outperforms the state of the art, achieving plausible dynamic deformations even for unseen motions. https://zqbai-jeremy.github.io/autoavatar"



Paperid:224
Authors:Yuecong Min; Peiqi Jiao; Yanan Li; Xiaotao Wang; Lei Lei; Xiujuan Chai; Xilin Chen
Title: Deep Radial Embedding for Visual Sequence Learning
Abstract:
Connectionist Temporal Classification (CTC) is a popular objective function in sequence recognition, which provides supervision for unsegmented sequence data through aligning sequence and its corresponding labeling iteratively. The blank class of CTC plays a crucial role in the alignment process and is often considered responsible for the peaky behavior of CTC. In this study, we propose an objective function named RadialCTC that constrains sequence features on a hypersphere while retaining the iterative alignment mechanism of CTC. The learned features of each non-blank class are distributed on a radial arc from the center of the blank class, which provides a clear geometric interpretation and makes the alignment process more efficient. Besides, RadialCTC can control the peaky behavior by simply modifying the logit of the blank class. Experimental results of recognition and localization demonstrate the effectiveness of RadialCTC on two sequence recognition applications.



Paperid:225
Authors:Yan Wu; Jiahao Wang; Yan Zhang; Siwei Zhang; Otmar Hilliges; Fisher Yu; Siyu Tang
Title: SAGA: Stochastic Whole-Body Grasping with Contact
Abstract:
The synthesis of human grasping has numerous applications including AR/VR, video games and robotics. While methods have been proposed to generate realistic hand-object interaction for object grasping and manipulation, these typically only consider interacting hand alone. Our goal is to synthesize whole-body grasping motions. Starting from an arbitrary initial pose, we aim to generate diverse and natural whole-body human motions to approach and grasp a target object in 3D space. This task is challenging as it requires modeling both whole-body dynamics and dexterous finger movements. To this end, we propose SAGA (StochAstic whole-body Grasping with contAct), a framework which consists of two key components: (a) Static whole-body grasping pose generation. Specifically, we propose a multi-task generative model, to jointly learn static whole-body grasping poses and human-object contacts. (b) Grasping motion infilling. Given an initial pose and the generated whole-body grasping pose as the start and end of the motion respectively, we design a novel contact-aware generative motion infilling module to generate a diverse set of grasp-oriented motions. We demonstrate the effectiveness of our method, which is a novel generative framework to synthesize realistic and expressive whole-body motions that approach and grasp randomly placed unseen objects. Code and models are available at https://jiahaoplus.github.io/SAGA/saga.html.



Paperid:226
Authors:Gusi Te; Xiu Li; Xiao Li; Jinglu Wang; Wei Hu; Yan Lu
Title: Neural Capture of Animatable 3D Human from Monocular Video
Abstract:
We present a novel paradigm of building an animatable 3D human representation from a monocular video input, such that it can be rendered in any unseen poses and views. Our method is based on a dynamic Neural Radiance Field (NeRF) rigged by a mesh-based parametric 3D human model serving as a geometry proxy. Previous methods usually rely on multi-view videos or accurate 3D geometry information as additional inputs; besides, most methods suffer from degraded quality when generalized to unseen poses. We identify that the key to generalization is a good input embedding for querying dynamic NeRF: A good input embedding should define an injective mapping in the full volumetric space, guided by surface mesh deformation under pose variation. Based on this observation, we propose to embed the input query with its relationship to local surface regions spanned by a set of geodesic nearest neighbors on mesh vertices. By including both position and relative distance information, our embedding defines a distance-preserved deformation mapping and generalizes well to unseen poses. To reduce the dependency on additional inputs, we first initialize per-frame 3D meshes using off-the-shelf tools and then propose a pipeline to jointly optimize NeRF and refine the initial mesh. Extensive experiments show our method can synthesize plausible human rendering results under unseen poses and views.



Paperid:227
Authors:Yukun Su; Guosheng Lin; Ruizhou Sun; Qingyao Wu
Title: General Object Pose Transformation Network from Unpaired Data
Abstract:
Object pose transformation is a challenging task. Yet, most existing pose transformation networks only focus on synthesizing humans. These methods either rely on the keypoints information or rely on the manual annotations of the paired target pose images for training. However, collecting such paired data is laboring and the cue of keypoints is inapplicable to general objects. In this paper, we address a problem of novel general object pose transformation from unpaired data. Given a source image of an object that provides appearance information and the desired pose image as a reference in the absence of paired examples, we produce a depiction of that object in that pose, retaining the appearance of both the object and background. Specifically, to preserve the source information, we propose an adversarial network with $\textbf{S}$patial-$\textbf{S}$tructural (SS) block and $\textbf{T}$exture-$\textbf{S}$tyle-$\textbf{C}$olor (TSC) block after the correlation matching module that facilitates the output to be semantically corresponding to the target pose image while contextually related to the source image. In addition, we can extend our network to complete multi-object and cross-category pose transformation. Extensive experiments demonstrate the effectiveness of our method which can create more realistic images when compared to those of recent approaches in terms of image quality. Moreover, we show the practicality of our method for several applications.



Paperid:228
Authors:Kaifeng Zhao; Shaofei Wang; Yan Zhang; Thabo Beeler; Siyu Tang
Title: Compositional Human-Scene Interaction Synthesis with Semantic Control
Abstract:
Synthesizing natural interactions between virtual humans and their 3D environments is critical for numerous applications, such as computer games and AR/VR experiences. Recent methods mainly focus on modeling geometric relations between 3D environments and humans, where the high-level semantics of the human-scene interaction has frequently been ignored. Our goal is to synthesize humans interacting with a given 3D scene controlled by high-level semantic specifications as pairs of action categories and object instances, e.g., “sit on the chair”. The key challenge of incorporating interaction semantics into the generation framework is to learn a joint representation that effectively captures heterogeneous information, including human body articulation, 3D object geometry, and the intent of the interaction. To address this challenge, we design a novel transformer-based generative model, in which the articulated 3D human body surface points and 3D objects are jointly encoded in a unified latent space, and the semantics of the interaction between the human and objects are embedded via positional encoding. Furthermore, inspired by the compositional nature of interactions that humans can simultaneously interact with multiple objects, we define interaction semantics as the composition of varying numbers of atomic action-object pairs. Our proposed generative model can naturally incorporate varying numbers of atomic interactions, which enables synthesizing compositional human-scene interactions without requiring composite interaction data. We extend the PROX dataset with interaction semantic labels and scene instance segmentation to evaluate our method and demonstrate that our method can generate realistic human-scene interactions with semantic control. Our perceptual study shows that our synthesized virtual humans can naturally interact with 3D scenes, considerably outperforming existing methods. We name our method COINS, for COmpositional INteraction Synthesis with Semantic Control. Code and data are available at https://github.com/zkf1997/COINS.



Paperid:229
Authors:Patrick Grady; Chengcheng Tang; Samarth Brahmbhatt; Christopher D. Twigg; Chengde Wan; James Hays; Charles C. Kemp
Title: PressureVision: Estimating Hand Pressure from a Single RGB Image
Abstract:
People often interact with their surroundings by applying pressure with their hands. While hand pressure can be measured by placing pressure sensors between the hand and the environment, doing so can alter contact mechanics, interfere with human tactile perception, require costly sensors, and scale poorly to large environments. We explore the possibility of using a conventional RGB camera to infer hand pressure, enabling machine perception of hand pressure from uninstrumented hands and surfaces. The central insight is that the application of pressure by a hand results in informative appearance changes. Hands share biomechanical properties that result in similar observable phenomena, such as soft-tissue deformation, blood distribution, hand pose, and cast shadows. We collected videos of 36 participants with diverse skin tone applying pressure to an instrumented planar surface. We then trained a deep model (PressureVisionNet) to infer a pressure image from a single RGB image. Our model infers pressure for participants outside of the training data and outperforms baselines. We also show that the output of our model depends on the appearance of the hand and cast shadows near contact regions. Overall, our results suggest the appearance of a previously unobserved human hand can be used to accurately infer applied pressure. Data, code, and models are available online.



Paperid:230
Authors:Ginger Delmas; Philippe Weinzaepfel; Thomas Lucas; Francesc Moreno-Noguer; Grégory Rogez
Title: PoseScript: 3D Human Poses from Natural Language
Abstract:
Natural language is leveraged in many computer vision tasks such as image captioning, cross-modal retrieval or visual question answering, to provide fine-grained semantic information. While human pose is key to human understanding, current 3D human pose datasets lack detailed language descriptions. In this work, we introduce the PoseScript dataset, which pairs a few thousand 3D human poses from AMASS with rich human-annotated descriptions of the body parts and their spatial relationships. To increase the size of this dataset to a scale compatible with typical data hungry learning algorithms, we propose an elaborate captioning process that generates automatic synthetic descriptions in natural language from given 3D keypoints. This process extracts low-level pose information -- the posecodes -- using a set of simple but generic rules on the 3D keypoints. The posecodes are then combined into higher level textual descriptions using syntactic rules. Automatic annotations substantially increase the amount of available data, and make it possible to effectively pretrain deep models for finetuning on human captions. To demonstrate the potential of annotated poses, we show applications of the PoseScript dataset to retrieval of relevant poses from large-scale datasets and to synthetic pose generation, both based on a textual pose description. Code and dataset are available at https://europe.naverlabs.com/research/computer-vision/posescript/.



Paperid:231
Authors:Jaewoo Park; Nam Ik Cho
Title: DProST: Dynamic Projective Spatial Transformer Network for 6D Pose Estimation
Abstract:
Predicting the object’s 6D pose from a single RGB image is a fundamental computer vision task. Generally, the distance between transformed object vertices is employed as an objective function for pose estimation methods. However, projective geometry in the camera space is not considered in those methods and causes performance degradation. In this regard, we propose a new pose estimation system based on a projective grid instead of object vertices. Our pose estimation method, dynamic projective spatial transformer network (DProST), localizes the region of interest grid on the rays in camera space and transforms the grid to object space by estimated pose. The transformed grid is used as both a sampling grid and a new criterion of the estimated pose. Additionally, because DProST does not require object vertices, our method can be used in a mesh-less setting by replacing the mesh with a reconstructed feature. Experimental results show that mesh-less DProST outperforms the state-of-the-art mesh-based methods on the LINEMOD and LINEMOD-OCCLUSION dataset, and shows competitive performance on the YCBV dataset with mesh data. The source code is available at https://github.com/parkjaewoo0611/DProST.



Paperid:232
Authors:Hao Meng; Sheng Jin; Wentao Liu; Chen Qian; Mengxiang Lin; Wanli Ouyang; Ping Luo
Title: 3D Interacting Hand Pose Estimation by Hand De-Occlusion and Removal
Abstract:
Estimating 3D interacting hand pose from a single RGB image is essential for understanding human actions. Unlike most previous works that directly predict the 3D poses of two interacting hands simultaneously, we propose to decompose the challenging interacting hand pose estimation task and estimate the pose of each hand separately. In this way, it is straightforward to take advantage of the latest research progress on the single-hand pose estimation system. However, hand pose estimation in interacting scenarios is very challenging, due to (1) severe hand-hand occlusion and (2) ambiguity caused by the homogeneous appearance of hands. To tackle these two challenges, we propose a novel Hand De-occlusion and Removal (HDR) framework to perform hand de-occlusion and distractor removal. We also propose the first large-scale synthetic amodal hand dataset, termed Amodal InterHand Dataset (AIH), to facilitate model training and promote the development of the related research. Experiments show that the proposed method significantly outperforms previous state-of-the-art interacting hand pose estimation approaches. Codes and data are available at https://github.com/MengHao666/HDR.



Paperid:233
Authors:Lumin Xu; Sheng Jin; Wang Zeng; Wentao Liu; Chen Qian; Wanli Ouyang; Ping Luo; Xiaogang Wang
Title: Pose for Everything: Towards Category-Agnostic Pose Estimation
Abstract:
Existing works on 2D pose estimation mainly focus on a certain category, e.g. human, animal, and vehicle. However, there are lots of application scenarios that require detecting the poses/keypoints of the unseen class of objects. In this paper, we introduce the task of Category-Agnostic Pose Estimation (CAPE), which aims to create a pose estimation model capable of detecting the pose of any class of object given only a few samples with keypoint definition. To achieve this goal, we formulate the pose estimation problem as a keypoint matching problem and design a novel CAPE framework, termed POse Matching Network (POMNet). A transformer-based Keypoint Interaction Module (KIM) is proposed to capture both the interactions among different keypoints and the relationship between the support and query images. We also introduce Multi-category Pose (MP-100) dataset, which is a 2D pose dataset of 100 object categories containing over 20K instances and is well-designed for developing CAPE algorithms. Experiments show that our method outperforms other baseline approaches by a large margin. Codes and data are available at https://github.com/luminxu/Pose-for-Everything.



Paperid:234
Authors:Thomas Lucas; Fabien Baradel; Philippe Weinzaepfel; Grégory Rogez
Title: PoseGPT: Quantization-Based 3D Human Motion Generation and Forecasting
Abstract:
We address the problem of action-conditioned generation of human motion sequences. Existing work falls into two categories: forecast models conditioned on observed past motions, or generative models conditioned action labels and duration only. In contrast, we generate motion conditioned on observations of arbitrary length, including none. To solve this generalized problem, we propose PoseGPT, an auto-regressive transformer-based approach which internally compresses human motion into quantized latent sequences. An auto-encoder first maps human motion to latent index sequences in a discrete space, and vice-versa. Inspired by the Generative Pretrained Transformer (GPT), we propose to train a GPT-like model for next-index prediction in that space; this allows PoseGPT to output distributions on possible futures, with or without conditioning on past motion. The discrete and compressed nature of the latent space allows the GPT- like model to focus on long-range signal, as it removes low-level redundancy in the input signal. Predicting discrete indices also alleviates the common pitfall of predicting averaged poses, a typical failure case when regressing continuous values, as the average of discrete targets is not a target itself. Our experimental results show that our proposed approach achieves state-of-the-art results on Hu- manAct12 - a standard but small scale dataset, on BABEL - a recent large scale MoCap dataset and on GRAB - a human-object interactions dataset.



Paperid:235
Authors:Linzhi Huang; Jiahao Liang; Weihong Deng
Title: DH-AUG: DH Forward Kinematics Model Driven Augmentation for 3D Human Pose Estimation
Abstract:
Due to the lack of diversity of datasets, the generalization ability of the pose estimator is poor. To solve this problem, we propose a pose augmentation solution via DH forward kinematics model, which we call DH-AUG. We observe that the previous work is all based on single-frame pose augmentation, if it is directly applied to video pose estimator, there will be several previously ignored problems: (i) angle ambiguity in bone rotation (multiple solutions); (ii) the generated skeleton video lacks movement continuity. To solve these problems, we propose a special generator based on DH forward kinematics model, which is called DH-generator. Extensive experiments demonstrate that DH-AUG can greatly increase the generalization ability of the video pose estimator. In addition, when applied to a single-frame 3D pose estimator, our method outperforms the previous best pose augmentation method. The source code has been released at https://github.com/hlz0606/DH-AUG-DH-Forward-Kinematics-Model-Driven-Augmentation-for-3D-Human-Pose-Estimation.



Paperid:236
Authors:Jiajun Tang; Yongjie Zhu; Haoyu Wang; Jun Hoong Chan; Si Li; Boxin Shi
Title: Estimating Spatially-Varying Lighting in Urban Scenes with Disentangled Representation
Abstract:
We present an end-to-end network for spatially-varying outdoor lighting estimation in urban scenes given a single limited field-of-view LDR image and any assigned 2D pixel position. We use three disentangled latent spaces learned by our network to represent sky light, sun light, and lighting-independent local contents respectively. At inference time, our lighting estimation network can run efficiently in an end-to-end manner by merging the global lighting and the local appearance rendered by the local appearance renderer with the predicted local silhouette. We enhance an existing synthetic dataset with more realistic material models and diverse lighting conditions for more effective training. We also capture the first real dataset with HDR labels for evaluating spatially-varying outdoor lighting estimation. Experiments on both synthetic and real datasets show that our method achieves state-of-the-art performance with more flexible editability.



Paperid:237
Authors:Wenming Weng; Yueyi Zhang; Zhiwei Xiong
Title: Boosting Event Stream Super-Resolution with a Recurrent Neural Network
Abstract:
Existing methods for event stream super-resolution (SR) either require high-quality and high-resolution frames or underperform for large factor SR. To address these problems, we propose a recurrent neural network for event SR without frames. First, we design a temporal propagation net for incorporating neighboring and long-range event-aware contexts that facilitates event SR. Second, we build a spatiotemporal fusion net for reliably aggregating the spatiotemporal clues of event stream. These two elaborate components are tightly synergized for achieving satisfying event SR results even for 16X SR. Synthetic and real-world experimental results demonstrate the clear superiority of our method. Furthermore, we evaluate our method on two downstream event-driven applications, i.e., object recognition and video reconstruction, achieving remarkable performance boost over existing methods.



Paperid:238
Authors:Yuxi Li; Huijie Zhao; Hongzhi Jiang; Xudong Li
Title: Projective Parallel Single-Pixel Imaging to Overcome Global Illumination in 3D Structure Light Scanning
Abstract:
We consider robust and efficient 3D structure light scanning method in situations dominated by global illumination. One typical way of solving this problem is via the analysis of 4D light transport coefficients (LTCs), which contains complete information for a projector-camera pair, and is a 4D data set. However, the process of capturing LTCs generally takes long time. We present projective parallel single-pixel imaging (pPSI), wherein the 4D LTCs are reduced to multiple projection functions to facilitate a highly efficient data capture process. We introduce local maximum constraint, which provides necessary condition for the location of correspondence matching points when projection functions are captured. Local slice extension method is introduced to further accelerate the capture of projection functions. We study the influence of scan ratio in local slice extension method on the accuracy of the correspondence matching points, and conclude that partial scanning is enough for satisfactory results. Our discussions and experiments include three typical kinds of global illuminations: inter-reflections, subsurface scattering, and step edge fringe aliasing. The proposed method is validated in several challenging scenarios.



Paperid:239
Authors:Yunpeng Bai; Chao Dong; Zenghao Chai; Andong Wang; Zhengzhuo Xu; Chun Yuan
Title: Semantic-Sparse Colorization Network for Deep Exemplar-Based Colorization
Abstract:
Exemplar-based colorization approaches rely on reference image to provide plausible colors for target gray-scale image. The key and difficulty of exemplar-based colorization is to establish an accurate correspondence between these two images. Previous approaches have attempted to construct such a correspondence but are faced with two obstacles. First, using luminance channels for the calculation of correspondence is inaccurate. Second, the dense correspondence they built introduces wrong matching results and increases the computation burden. To address these two problems, we propose Semantic-Sparse Colorization Network (SSCN) to transfer both the global image style and detailed semantic-related colors to the gray-scale image in a coarse-to-fine manner. Our network can perfectly balance the global and local colors while alleviating the ambiguous matching problem. Experiments show that our method outperforms existing methods in both quantitative and qualitative evaluation and achieves state-of-the-art performance.



Paperid:240
Authors:Alexandros Lattas; Yiming Lin; Jayanth Kannan; Ekin Ozturk; Luca Filipi; Giuseppe Claudio Guarnera; Gaurav Chawla; Abhijeet Ghosh
Title: Practical and Scalable Desktop-Based High-Quality Facial Capture
Abstract:
We present a novel desktop-based system for high-quality facial capture including geometry and facial appearance. The proposed acquisition system is highly practical and scalable, consisting purely of commodity components. The setup consists of a set of displays for controlled illumination for reflectance capture, in conjunction with multiview acquisition of facial geometry. We additionally present a novel set of binary illumination patterns for efficient acquisition of reflectance and photometric normals using our setup, with diffuse-specular separation. We demonstrate high-quality results with two different variants of the capture setup - one entirely consisting of portable mobile devices targeting static facial capture, and the other consisting of desktop LCD displays targeting both static and dynamic facial capture.



Paperid:241
Authors:Haoning Wu; Chaofeng Chen; Jingwen Hou; Liang Liao; Annan Wang; Wenxiu Sun; Qiong Yan; Weisi Lin
Title: FAST-VQA: Efficient End-to-End Video Quality Assessment with Fragment Sampling
Abstract:
Current deep video quality assessment (VQA) methods are usually with high computational costs when evaluating high-resolution videos. This cost hinders them from learning better video-quality-related representations via end-to-end training. Existing approaches typically consider naive sampling to reduce the computational cost, such as resizing and cropping. However, they obviously corrupt quality-related information in videos and are thus not optimal to learn good representations for VQA. Therefore, there is an eager need to design a new quality-retained sampling scheme for VQA. In this paper, we propose Grid Mini-patch Sampling (GMS), which allows consideration of local quality by sampling patches at their raw resolution and covers global quality with contextual relations via mini-patches sampled in uniform grids. These mini-patches are spliced and aligned temporally, named as fragments. We further build the Fragment Attention Network (FANet) specially designed to accommodate fragments as inputs. Consisting of fragments and FANet, the proposed FrAgment Sample Transformer for VQA (FAST-VQA) enables efficient end-to-end deep VQA and learns effective video-quality-related representations. It improves state-of-the-art accuracy by around 10% while reducing 99.5% FLOPs on 1080P high-resolution videos. The newly learned video-quality-related representations can also be transferred into smaller VQA datasets, boosting the performance on these scenarios. Extensive experiments show that FAST-VQA has good performance on inputs of various resolutions while retaining high efficiency. We publish our code at https://github.com/timothyhtimothy/FAST-VQA.



Paperid:242
Authors:Zhengqin Li; Jia Shi; Sai Bi; Rui Zhu; Kalyan Sunkavalli; Miloš Hašan; Zexiang Xu; Ravi Ramamoorthi; Manmohan Chandraker
Title: Physically-Based Editing of Indoor Scene Lighting from a Single Image
Abstract:
We present a method to edit complex indoor lighting from a single image with its predicted depth and light source segmentation masks. This is an extremely challenging problem that requires modeling complex light transport, and disentangling HDR lighting from material and geometry with only a partial LDR observation of the scene. We tackle this problem using two novel components: 1) a holistic scene reconstruction method that estimates scene reflectance and parametric 3D lighting, and 2) a neural rendering framework that re-renders the scene from our predictions. We use physically-based indoor light representations that allow for intuitive editing, and infer both visible and invisible light sources. Our neural rendering framework combines physically-based direct illumination and shadow rendering with deep networks to approximate global illumination. It can capture challenging lighting effects, such as soft shadows, directional lighting, specular materials, and interreflections. Previous single image inverse rendering methods usually entangle scene lighting and geometry and only support applications like object insertion. Instead, by combining parametric 3D lighting estimation with neural scene rendering, we demonstrate the first automatic method to achieve full scene relighting, including light source insertion, removal, and replacement, from a single image. All source code and data will be publicly released.



Paperid:243
Authors:Shangchen Zhou; Chongyi Li; Chen Change Loy
Title: LEDNet: Joint Low-Light Enhancement and Deblurring in the Dark
Abstract:
Night photography typically suffers from both low light and blurring issues due to the dim environment and the common use of long exposure. While existing light enhancement and deblurring methods could deal with each problem individually, a cascade of such methods cannot work harmoniously to cope well with joint degradation of visibility and sharpness. Training an end-to-end network is also infeasible as no paired data is available to characterize the coexistence of low light and blurs. We address the problem by introducing a novel data synthesis pipeline that models realistic low-light blurring degradations, especially for blurs in saturated regions, e.g., light streaks, that often appear in the night images. With the pipeline, we present the first large-scale dataset for joint low-light enhancement and deblurring. The dataset, LOL-Blur, contains 12,000 low-blur/normal-sharp pairs with diverse darkness and blurs in different scenarios. We further present an effective network, named LEDNet, to perform joint low-light enhancement and deblurring. Our network is unique as it is specially designed to consider the synergy between the two inter-connected tasks. Both the proposed dataset and network provide a foundation for this challenging joint task. Extensive experiments demonstrate the effectiveness of our method on both synthetic and real-world datasets.



Paperid:244
Authors:Juewen Peng; Jianming Zhang; Xianrui Luo; Hao Lu; Ke Xian; Zhiguo Cao
Title: MPIB: An MPI-Based Bokeh Rendering Framework for Realistic Partial Occlusion Effects
Abstract:
Partial occlusion effects are a phenomenon that blurry objects near a camera are semi-transparent, resulting in partial appearance of occluded background. However, it is challenging for existing bokeh rendering methods to simulate realistic partial occlusion effects due to the missing information of the occluded area in an all-in-focus image. Inspired by the learnable 3D scene representation, Multiplane Image (MPI), we attempt to address the partial occlusion by introducing a novel MPI-based high-resolution bokeh rendering framework, termed MPIB. To this end, we first present an analysis on how to apply the MPI representation to bokeh rendering. Based on this analysis, we propose an MPI representation module combined with a background inpainting module to implement high-resolution scene representation. This representation can then be reused to render various bokeh effects according to the controlling parameters. To train and test our model, we also design a ray-tracing-based bokeh generator for data generation. Extensive experiments on synthesized and real-world images validate the effectiveness and flexibility of this framework.



Paperid:245
Authors:Huanjing Yue; Zhiming Zhang; Jingyu Yang
Title: Real-RawVSR: Real-World Raw Video Super-Resolution with a Benchmark Dataset
Abstract:
In recent years, real image super-resolution (SR) has achieved promising results due to the development of SR datasets and corresponding real SR methods. In contrast, the field of real video SR is lagging behind, especially for real raw videos. Considering the superiority of raw image SR over sRGB image SR, we construct a real-world raw video SR (Real-RawVSR) dataset and propose a corresponding SR method. We utilize two DSLR cameras and a beam-splitter to simultaneously capture low-resolution (LR) and high-resolution (HR) raw videos with 2x, 3x, and 4x magnifications. There are 450 video pairs in our dataset, with scenes varying from indoor to outdoor, and motions including camera and object movements. To our knowledge, this is the first real-world raw VSR dataset. Since the raw video is characterized by the Bayer pattern, we propose a two-branch network, which deals with both the packed RGGB sequence and the original Bayer pattern sequence, and the two branches are complementary to each other. After going through the proposed co-alignment, interaction, fusion, and reconstruction modules, we generate the corresponding HR sRGB sequence. Experimental results demonstrate that the proposed method outperforms benchmark real and synthetic video SR methods with either raw or sRGB inputs. Our code and dataset are available at https://github.com/zmzhang1998/Real-RawVSR.



Paperid:246
Authors:Ardhendu Shekhar Tripathi; Martin Danelljan; Samarth Shukla; Radu Timofte; Luc Van Gool
Title: Transform Your Smartphone into a DSLR Camera: Learning the ISP in the Wild
Abstract:
We propose a trainable Image Signal Processing (ISP) framework that produces DSLR quality images given RAW images captured by a smartphone. To address the color misalignments between training image pairs, we employ a color-conditional ISP network and optimize a novel parametric color mapping between each input RAW and reference DSLR image. During inference, we predict the target color image by designing a color prediction network with efficient Global Context Transformer modules. The latter effectively leverage global information to learn consistent color and tone mappings. We further propose a robust masked aligned loss to identify and discard regions with inaccurate motion estimation during training. Lastly, we introduce the ISP in the Wild (ISPW) dataset, consisting of weakly paired phone RAW and DSLR sRGB images. We extensively evaluate our method, setting a new state-of-the-art on two datasets. The code is available at https://github.com/4rdhendu/TransformPhone2DSLR.



Paperid:247
Authors:Yuhui Quan; Zhuojie Chen; Huan Zheng; Hui Ji
Title: Learning Deep Non-Blind Image Deconvolution without Ground Truths
Abstract:
Non-blind image deconvolution (NBID) is about restoring a latent sharp image from a blurred one, given an associated blur kernel. Most existing deep neural networks for NBID are trained over many ground truth (GT) images, which limits their applicability in practical applications such as microscopic imaging and medical imaging. This paper proposes an unsupervised deep learning approach for NBID which avoids accessing GT images. The challenge raised from the absence of GT images is tackled by a self-supervised reconstruction loss that approximates its supervised counterpart well. The possible errors of blur kernels are addressed by a self-supervised prediction loss based on intermediate samples as well as an ensemble inference scheme based on kernel perturbation. The experiments show that the proposed approach provides very competitive performance to existing supervised learning-based methods, no matter under accurate kernels or erroneous kernels.



Paperid:248
Authors:Minggui Teng; Chu Zhou; Hanyue Lou; Boxin Shi
Title: NEST: Neural Event Stack for Event-Based Image Enhancement
Abstract:
Event cameras demonstrate unique characteristics such as high temporal resolution, low latency, and high dynamic range to improve performance for various image enhancement tasks. However, event streams cannot be applied to neural networks directly due to their sparse nature. To integrate events into traditional computer vision algorithms, an appropriate event representation is desirable, while existing voxel grid and event stack representations are less effective in encoding motion and temporal information. This paper presents a novel event representation named Neural Event STack (NEST), which satisfies physical constraints and encodes comprehensive motion and temporal information sufficient for image enhancement. We apply our representation on multiple tasks, which achieves superior performance on image deblurring and image super-resolution than state-of-the-art methods on both synthetic and real datasets. And we further demonstrate the possibility to generate high frame rate videos with our novel event representation.



Paperid:249
Authors:Henrique Weber; Mathieu Garon; Jean-François Lalonde
Title: Editable Indoor Lighting Estimation
Abstract:
We present a method for estimating lighting from a single perspective image of an indoor scene. Previous methods for predicting indoor illumination usually focus on either simple, parametric lighting that lack realism, or on richer representations that are difficult or even impossible to understand or modify after prediction. We propose a pipeline that estimates a parametric light that is easy to edit and allows renderings with strong shadows, alongside with a non-parametric texture with high-frequency information necessary for realistic rendering of specular objects. Once estimated, the predictions obtained with our model are interpretable and can easily be modified by an artist/user with a few mouse clicks. Quantitative and qualitative results show that our approach makes indoor lighting estimation easier to handle by a casual user, while still producing competitive results.



Paperid:250
Authors:Thomas Eboli; Jean-Michel Morel; Gabriele Facciolo
Title: Fast Two-Step Blind Optical Aberration Correction
Abstract:
The optics of any camera degrades the sharpness of photographs, which is a key visual quality criterion. This degradation is characterized by the point-spread function (PSF), which depends on the wavelengths of light and is variable across the imaging field. In this paper, we propose a two-step scheme to correct optical aberrations in a single raw or JPEG image, i.e., without any prior information on the camera or lens. First, we estimate local Gaussian blur kernels for overlapping patches and sharpen them with a non-blind deblurring technique. Based on the measurements of the PSFs of dozens of lenses, these blur kernels are modeled as RGB Gaussians defined by seven parameters. Second, we remove the remaining lateral chromatic aberrations (not contemplated in the first step) with a convolutional neural network, trained to minimize the red/green and blue/green residual images. Experiments on both synthetic and real images show that the combination of these two stages yields a fast state-of-the-art blind optical aberration compensation technique that competes with commercial non-blind algorithms.



Paperid:251
Authors:Zhanghao Sun; Jian Wang; Yicheng Wu; Shree Nayar
Title: Seeing Far in the Dark with Patterned Flash
Abstract:
Flash illumination is widely used in imaging under low-light environments. However, illumination intensity falls off with propagation distance quadratically, which poses significant challenges for flash imaging at a long distance. We propose a new flash technique, named “patterned flash”, for flash imaging at a long distance. Patterned flash concentrates optical power into a dot array. Compared with the conventional uniform flash where the signal is overwhelmed by the noise everywhere, patterned flash provides stronger signals at sparsely distributed points across the field of view to ensure the signals at those points stand out from the sensor noise. This enables post-processing to resolve important objects and details. Additionally, the patterned flash projects texture onto the scene, which can be treated as a structured light system for depth perception. Given the novel system, we develop a joint image reconstruction and depth estimation algorithm with a convolutional neural network. We build a hardware prototype and test the proposed flash technique on various scenes. The experimental results demonstrate that our patterned flash has significantly better performance at long distances in low-light environments. Our code and data are publicly available.



Paperid:252
Authors:Qin Liu; Meng Zheng; Benjamin Planche; Srikrishna Karanam; Terrence Chen; Marc Niethammer; Ziyan Wu
Title: PseudoClick: Interactive Image Segmentation with Click Imitation
Abstract:
The goal of click-based interactive image segmentation is to obtain precise object segmentation masks with limited user interaction, i.e., by a minimal number of user clicks. Existing methods require users to provide all the clicks: by first inspecting the segmentation mask and then providing points on mislabeled regions, iteratively. We ask the question: can our model directly predict where to click, so as to further reduce the user interaction cost? To this end, we propose PseudoClick, a generic framework that enables existing segmentation networks to propose candidate next clicks. These automatically generated clicks, termed pseudo clicks in this work, serve as an imitation of human clicks to refine the segmentation mask. We build PseudoClick on existing segmentation backbones and show how our click prediction mechanism leads to improved performance. We evaluate PseudoClick on 10 public datasets from different domains and modalities, showing that our model not only outperforms existing approaches but also demonstrates strong generalization capability in cross-domain evaluation. We obtain new state-of-the-art results on several popular benchmarks, e.g., on the PASCAL dataset, our model significantly outperforms existing state-of-the-art by reducing 12.4% and 11.4% number of clicks to achieve 85% and 90% IoU, respectively.



Paperid:253
Authors:Shuchen Weng; Jimeng Sun; Yu Li; Si Li; Boxin Shi
Title: CT$^2$: Colorization Transformer via Color Tokens
Abstract:
Automatic image colorization is an ill-posed problem with multi-modal uncertainty, and there remains two main challenges with previous methods: incorrect semantic colors and under-saturation. In this paper, we propose an end-to-end transformer-based model to overcome these challenges. Benefited from the long-range context extraction of transformer and our holistic architecture, our method could colorize images with more diverse colors. Besides, we introduce color tokens into our approach and treat the colorization task as a classification problem, which increases the saturation of results. We also propose a series of modules to make image features interact with color tokens, and restrict the range of possible color candidates, which makes our results visually pleasing and reasonable. In addition, our method does not require any additional external priors, which ensures its well generalization capability. Extensive experiments and user studies demonstrate that our method achieves superior performance than previous works.



Paperid:254
Authors:Liangyu Chen; Xiaojie Chu; Xiangyu Zhang; Jian Sun
Title: Simple Baselines for Image Restoration
Abstract:
Although there have been significant advances in the field of image restoration recently, the system complexity of the state-of-the-art (SOTA) methods is increasing as well, which may hinder the convenient analysis and comparison of methods. In this paper, we propose a simple baseline that exceeds the SOTA methods and is computationally efficient. To further simplify the baseline, we reveal that the nonlinear activation functions, e.g. Sigmoid, ReLU, GELU, Softmax, etc. are not necessary: they could be replaced by multiplication or removed. Thus, we derive a Nonlinear Activation Free Network, namely NAFNet, from the baseline. SOTA results are achieved on various challenging benchmarks, e.g. 33.69 dB PSNR on GoPro (for image deblurring), exceeding the previous SOTA 0.38 dB with only 8.4% of its computational costs; 40.30 dB PSNR on SIDD (for image denoising), exceeding the previous SOTA 0.28 dB with less than half of its computational costs. The code and the pre-trained models are released at https://github.com/megvii-research/NAFNet.



Paperid:255
Authors:Jiyuan Zhang; Lulu Tang; Zhaofei Yu; Jiwen Lu; Tiejun Huang
Title: Spike Transformer: Monocular Depth Estimation for Spiking Camera
Abstract:
Spiking camera is a bio-inspired vision sensor that mimics the sampling mechanism of the primate fovea, which has shown great potential for capturing high-speed dynamic scenes with a sampling rate of 40,000 Hz. Unlike conventional digital cameras, the spiking camera continuously captures photons and outputs asynchronous binary spikes that encode time, location, and light intensity. Because of the different sampling mechanisms, the off-the-shelf image-based algorithms for digital cameras are unsuitable for spike streams generated by the spiking camera. Therefore, it is of particular interest to develop novel, spike-aware algorithms for common computer vision tasks. In this paper, we focus on the depth estimation task, which is challenging due to the natural properties of spike streams, such as irregularity, continuity, and spatial-temporal correlation, and has not been explored for the spiking camera. We present Spike Transformer (Spike-T), a novel paradigm for learning spike data and estimating monocular depth from continuous spike streams. To fit spike data to Transformer, we present an input spike embedding equipped with a spatio-temporal patch partition module to maintain features from both spatial and temporal domains. Furthermore, we build two spike-based depth datasets. One is synthetic, and the other is captured by a real spiking camera. Experimental results demonstrate that the proposed Spike-T can favorably predict the scene’s depth and consistently outperform its direct competitors. More importantly, the representation learned by Spike-T transfers well to the unseen real data, indicating the generalization of Spike-T to real-world scenarios. To our best knowledge, this is the first time that directly depth estimation from spike streams becomes possible. Code and Datasets are available at https://github.com/Leozhangjiyuan/MDE-SpikingCamera.



Paperid:256
Authors:Xiaojie Chu; Liangyu Chen; Chengpeng Chen; Xin Lu
Title: Improving Image Restoration by Revisiting Global Information Aggregation
Abstract:
Global operations, such as global average pooling, are widely used in top-performance image restorers. They aggregate global information from input features along entire spatial dimensions but behave differently during training and inference in image restoration tasks: they are based on different regions, namely the cropped patches (from images) and the full-resolution images. This paper revisits global information aggregation and finds that the image-based features during inference have a different distribution than the patch-based features during training. This train-test inconsistency negatively impacts the performance of models, which is severely overlooked by previous works. To reduce the inconsistency and improve test-time performance, we propose a simple method called Test-time Local Converter (TLC). Our TLC converts global operations to local ones only during inference so that they aggregate features within local spatial regions rather than the entire large images. The proposed method can be applied to various global modules (e.g., normalization, channel and spatial attention) with negligible costs. Without the need for any fine-tuning, TLC improves state-of-the-art results on several image restoration tasks, including single-image motion deblurring, video deblurring, defocus deblurring, and image denoising. In particular, with TLC, our Restormer-Local improves the state-of-the-art result in single image deblurring from 32.92 dB to 33.57 dB on GoPro dataset. The code is available at https://github.com/megvii-research/tlc.



Paperid:257
Authors:Dehao Zhang; Qiankun Ding; Peiqi Duan; Chu Zhou; Boxin Shi
Title: Data Association between Event Streams and Intensity Frames under Diverse Baselines
Abstract:
This paper proposes a learning-based framework to associate event streams and intensity frames under diverse camera baselines, to simultaneously benefit to camera pose estimation under large baseline and depth estimation under small baseline. Based on the observation that event streams are globally sparse (a small percentage of pixels in global frames are triggered with events) and locally dense (a large percentage of pixels in local patches are triggered with events) in the spatial domain, we put forward a two-stage architecture for matching feature maps. LSparse-Net uses a large receptive field to find sparse matches while SDense-Net uses a small receptive field to find dense matches. Both two stages apply transformer modules with self-attention layers and cross-attention layers to effectively process multi-resolution features from the feature pyramid network backbone. Experimental results on public datasets show systematic performance improvement for both tasks compared to state-of-the-art methods.



Paperid:258
Authors:Yuzhi Zhao; Yongzhe Xu; Qiong Yan; Dingdong Yang; Xuehui Wang; Lai-Man Po
Title: D2HNet: Joint Denoising and Deblurring with Hierarchical Network for Robust Night Image Restoration
Abstract:
Night imaging with modern smartphone cameras is troublesome due to low photon count and unavoidable noise in the imaging system. Directly adjusting exposure time and ISO ratings cannot obtain sharp and noise-free images at the same time in low-light conditions. Though many methods have been proposed to enhance noisy or blurry night images, their performances on real-world night photos are still unsatisfactory due to two main reasons: 1) Limited information in a single image and 2) Domain gap between synthetic training images and real-world photos (e.g., differences in blur area and resolution). To exploit the information from successive long- and short-exposure images, we propose a learning-based pipeline to fuse them. A D2HNet framework is developed to recover a high-quality image by deblurring and enhancing a long-exposure image under the guidance of a short-exposure image. To shrink the domain gap, we leverage a two-phase DeblurNet-EnhanceNet architecture, which performs accurate blur removal on a fixed low resolution so that it is able to handle large ranges of blur in different resolution inputs. In addition, we synthesize a D2-Dataset from HD videos and experiment on it. The results on the validation set and real photos demonstrate our methods achieve better visual quality and state-of-the-art quantitative scores. The D2HNet codes and D2-Dataset can be found at https://github.com/zhaoyuzhi/D2HNet.



Paperid:259
Authors:Yongcheng Jing; Yining Mao; Yiding Yang; Yibing Zhan; Mingli Song; Xinchao Wang; Dacheng Tao
Title: Learning Graph Neural Networks for Image Style Transfer
Abstract:
State-of-the-art parametric and non-parametric style transfer approaches are prone to either distorted local style patterns due to global statistics alignment, or unpleasing artifacts resulting from patch mismatching. In this paper, we study a novel semi-parametric neural style transfer framework that alleviates the deficiency of both parametric and non-parametric stylization. The core idea of our approach is to establish accurate and fine-grained content-style correspondences using graph neural networks (GNNs). To this end, we develop an elaborated GNN model with content and style local patches as the graph vertices. The style transfer procedure is then modeled as the attention-based heterogeneous message passing between the style and content nodes in a learnable manner, leading to adaptive many-to-one style-content correlations at the local patch level. In addition, an elaborated deformable graph convolutional operation is introduced for cross-scale style-content matching. Experimental results demonstrate that the proposed semi-parametric image stylization approach yields encouraging results on the challenging style patterns, preserving both global appearance and exquisite details. Furthermore, by controlling the number of edges at the inference stage, the proposed method also triggers novel functionalities like diversified patch-based stylization with a single model.



Paperid:260
Authors:Ashish Tiwari; Shanmuganathan Raman
Title: DeepPS2: Revisiting Photometric Stereo Using Two Differently Illuminated Images
Abstract:
Estimating 3D surface normals through photometric stereo has been of great interest in computer vision research. Despite the success of existing traditional and deep learning-based methods, it is still challenging due to: (i) the requirement of three or more differently illuminated images, (ii) the inability to model unknown general reflectance, and (iii) the requirement of accurate 3D ground truth surface normals and known lighting information for training. In this work, we attempt to address an under-explored problem of photometric stereo using just two differently illuminated images, referred to as the PS2 problem. It is an intermediate case between a single image-based reconstruction method like Shape from Shading (SfS) and the traditional Photometric Stereo (PS), which requires three or more images. We propose an inverse rendering-based deep learning framework, called DeepPS2, that jointly performs surface normal, albedo, lighting estimation, and image relighting in a completely self-supervised manner with no requirement of ground truth data. We demonstrate how image relighting in conjunction with image reconstruction enhances the lighting estimation in a self-supervised setting.



Paperid:261
Authors:Shuchen Weng; Yi Wei; Ming-Ching Chang; Boxin Shi
Title: Instance Contour Adjustment via Structure-Driven CNN
Abstract:
Instance contour adjustment is desirable in image editing, which allows the contour of an instance in a photo to be either dilated or eroded via user sketching. This imposes several requirements for a favorable method in order to generate meaningful textures while preserving clear user-desired contours. Due to the ignorance of these requirements, the off-the-shelf image editing methods herein are unsuited. Therefore, we propose a specialized two-stage method. The first stage extracts the structural cues from the input image, and completes the missing structural cues for the adjusted area. The second stage is a structure-driven CNN which generates image textures following the guidance of the completed structural cues. In the structure-driven CNN, we redesign the context sampling strategy of the convolution operation and attention mechanism such that they can estimate and rank the relevance of the contexts based on the structural cues, and sample the top-ranked contexts regardless of their distribution on the image plane. Thus, the meaningfulness of image textures with clear and user-desired contours are guaranteed by the structure-driven CNN. In addition, our method does not require any semantic label as input, which thus ensures its well generalization capability. We evaluate our method against several baselines adapted from the related tasks, and the experimental results demonstrate its effectiveness.



Paperid:262
Authors:Shrisudhan Govindarajan; Prasan Shedligeri; Sarah; Kaushik Mitra
Title: Synthesizing Light Field Video from Monocular Video
Abstract:
The hardware challenges associated with light-field(LF) imaging has made it difficult for consumers to access its benefits like applications in post-capture focus and aperture control. Learning-based techniques which solve the ill-posed problem of LF reconstruction from sparse (1, 2 or 4) views have significantly reduced the requirement for complex hardware. LF video reconstruction from sparse views poses a special challenge as acquiring ground-truth for training these models is hard. Hence, we propose a self-supervised learning-based algorithm for LF video reconstruction from monocular videos. We use self-supervised geometric, photometric and temporal consistency constraints inspired from a recent self-supervised technique for LF video reconstruction from stereo video. Additionally, we propose three key techniques that are relevant to our monocular video input. We propose an explicit disocclusion handling technique that encourages the network to inpaint disoccluded regions in a LF frame, using information from adjacent input temporal frames. This is crucial for a self-supervised technique as a single input frame does not contain any information about the disoccluded regions. We also propose an adaptive low-rank representation that provides a significant boost in performance by tailoring the representation to each input scene. Finally, we also propose a novel refinement block that is able to exploit the available LF image data using supervised learning to further refine the reconstruction quality. Our qualitative and quantitative analysis demonstrates the significance of each of the proposed building blocks and also the superior results compared to previous state-of-the-art monocular LF reconstruction techniques. We further validate our algorithm by reconstructing LF videos from monocular videos acquired using a commercial GoPro camera.



Paperid:263
Authors:Bo Zhang; Li Niu; Xing Zhao; Liqing Zhang
Title: Human-Centric Image Cropping with Partition-Aware and Content-Preserving Features
Abstract:
Image cropping aims to find visually appealing crops in an image, which is an important yet challenging task. In this paper, we consider a specific and practical application: human-centric image cropping, which focuses on the depiction of a person. To this end, we propose a human-centric image cropping method with two novel feature designs for the candidate crop: partition-aware feature and content-preserving feature. For partition-aware feature, we divide the whole image into nine partitions based on the human bounding box and treat different partitions in a candidate crop differently conditioned on the human information. For content-preserving feature, we predict a heatmap indicating the important content to be included in a good crop, and extract the geometric relation between the heatmap and a candidate crop. Extensive experiments demonstrate that our method can perform favorably against state-of-the-art image cropping methods on human-centric image cropping task. Code is available at https://github.com/bcmi/Human-Centric-Image-Cropping.



Paperid:264
Authors:Jihyong Oh; Munchurl Kim
Title: DeMFI: Deep Joint Deblurring and Multi-Frame Interpolation with Flow-Guided Attentive Correlation and Recursive Boosting
Abstract:
We propose a novel joint deblurring and multi-frame interpolation (DeMFI) framework in a two-stage manner, called DeMFINet, which converts blurry videos of lower-frame-rate to sharp videos at higher-frame-rate based on flow-guided attentive-correlation-based feature bolstering (FAC-FB) module and recursive boosting (RB), in terms of multi-frame interpolation (MFI). Its baseline version performs featureflow-based warping with FAC-FB module to obtain a sharp-interpolated frame as well to deblur two center-input frames. Its extended version further improves the joint performance based on pixel-flow-based warping with GRU-based RB. Our FAC-FB module effectively gathers the distributed blurry pixel information over blurry input frames in featuredomain to improve the joint performances. RB trained with recursive boosting loss enables DeMFI-Net to adequately select smaller RB iterations for a faster runtime during inference, even after the training is finished. As a result, our DeMFI-Net achieves state-of-the-art (SOTA) performances for diverse datasets with significant margins compared to recent joint methods. All source codes, including pretrained DeMFI-Net, are publicly available at https://github.com/JihyongOh/DeMFI.



Paperid:265
Authors:Seonghyeon Nam; Marcus A. Brubaker; Michael S. Brown
Title: Neural Image Representations for Multi-Image Fusion and Layer Separation
Abstract:
We propose a framework for aligning and fusing multiple images into a single view using neural image representations (NIRs), also known as implicit or coordinate-based neural representations. Our framework targets burst images that exhibit camera ego motion and potential changes in the scene. We describe different strategies for alignment depending on the nature of the scene motion---namely, perspective planar (i.e., homography), optical flow with minimal scene change, and optical flow with notable occlusion and disocclusion. With the neural image representation, our framework effectively combines multiple inputs into a single canonical view without the need for selecting one of the images as a reference frame. We demonstrate how to use this multi-frame fusion framework for various layer separation tasks.



Paperid:266
Authors:Zhihang Zhong; Mingdeng Cao; Xiao Sun; Zhirong Wu; Zhongyi Zhou; Yinqiang Zheng; Stephen Lin; Imari Sato
Title: Bringing Rolling Shutter Images Alive with Dual Reversed Distortion
Abstract:
Rolling shutter (RS) distortion can be interpreted as the result of picking a row of pixels from instant global shutter (GS) frames over time during the exposure of the RS camera. This means that the information of each instant GS frame is partially, yet sequentially, embedded into the row-dependent distortion. Inspired by this fact, we address the challenging task of reversing this process, i.e., extracting undistorted GS frames from images suffering from RS distortion. However, since RS distortion is coupled with other factors such as readout settings and the relative velocity of scene elements to the camera, models that only exploit the geometric correlation between temporally adjacent images suffer from poor generality in processing data with different readout settings and dynamic scenes with both camera motion and object motion. In this paper, instead of two consecutive frames, we propose to exploit a pair of images captured by dual RS cameras with reversed RS directions for this highly challenging task. Grounded on the symmetric and complementary nature of dual reversed distortion, we develop a novel end-to-end model, IFED, to generate dual optical flow sequence through iterative learning of the velocity field during the RS time. Extensive experimental results demonstrate that IFED is superior to naive cascade schemes, as well as the state-of-the-art which utilizes adjacent RS images. Most importantly, although it is trained on a synthetic dataset, IFED is shown to be effective at retrieving GS frame sequences from real-world RS distorted images of dynamic scenes. Code is available at https://github.com/zzh-tech/Dual-Reversed-RS.



Paperid:267
Authors:Fitsum Reda; Janne Kontkanen; Eric Tabellion; Deqing Sun; Caroline Pantofaru; Brian Curless
Title: FILM: Frame Interpolation for Large Motion
Abstract:
We present a frame interpolation algorithm that synthesizes an engaging slow-motion video from near-duplicate photos which often exhibit large scene motion. Near-duplicates interpolation is an interesting new application, but large motion poses challenges to existing methods. To address this issue, we adapt a feature extractor that shares weights across the scales, and present a ""scale-agnostic"" motion estimator. It relies on the intuition that large motion at finer scales should be similar to small motion at coarser scales, which boosts the number of available pixels for large motion supervision. To inpaint wide disocclusions caused by large motion and synthesize crisp frames, we propose to optimize our network with the Gram matrix loss that measures the correlation difference between features. To simplify the training process, we further propose a unified single-network approach that removes the reliance on additional optical-flow or depth network and is trainable from frame triplets alone. Our approach outperforms state-of-the-art methods on the Xiph large motion benchmark while performing favorably on Vimeo-90K, Middlebury and UCF101. Codes and pre-trained models are available at https://film-net.github.io.



Paperid:268
Authors:Song Wu; Kaichao You; Weihua He; Chen Yang; Yang Tian; Yaoyuan Wang; Ziyang Zhang; Jianxing Liao
Title: Video Interpolation by Event-Driven Anisotropic Adjustment of Optical Flow
Abstract:
Video frame interpolation is a challenging task due to the ever-changing real-world scene. Previous methods often calculate the bi-directional optical flows and then predict the intermediate optical flows under the linear motion assumptions, leading to isotropic intermediate flow generation. Follow-up research obtained anisotropic adjustment through estimated higher-order motion information with extra frames. Based on the motion assumptions, their methods are hard to model the complicated motion in real scenes. In this paper, we propose an end-to-end training method A^2OF for video frame interpolation with event-driven Anisotropic Adjustment of Optical Flows. Specifically, we use events to generate optical flow distribution masks for the intermediate optical flow, which can model the complicated motion between two frames. Our proposed method outperforms the previous methods in video frame interpolation, taking supervised event-based video interpolation to a higher stage.



Paperid:269
Authors:Ziyun Wang; Kenneth Chaney; Kostas Daniilidis
Title: EvAC3D: From Event-Based Apparent Contours to 3D Models via Continuous Visual Hulls
Abstract:
3D reconstruction from multiple views is a successful computer vision field with multiple deployments in applications. State of the art is based on traditional RGB frames that enable optimization of photo-consistency cross views. In this paper, we study the problem of 3D reconstruction from event-cameras, motivated by the advantages of event-based cameras in terms of low power and latency as well as by the biological evidence that eyes in nature capture the same data and still perceive well 3D shape. The foundation of our hypothesis that 3D-reconstruction is feasible using events lies in the information contained in the occluding contours and in the continuous scene acquisition with events. We propose Apparent Contour Events (ACE), a novel event-based representation that defines the geometry of the apparent contour of an object. We represent ACE by a spatially and temporally continuous implicit function defined in the event x-y-t space. Furthermore, we design a novel continuous Voxel Carving algorithm enabled by the high temporal resolution of the Apparent Contour Events. To evaluate the performance of the method, we collect MOEC-3D, a 3D event dataset of a set of common real-world objects. We demonstrate EvAC3D’s ability to reconstruct high-fidelity mesh surfaces from real event sequences while allowing the refinement of the 3D reconstruction for each individual event.



Paperid:270
Authors:Ben Xue; Shenghui Ran; Quan Chen; Rongfei Jia; Binqiang Zhao; Xing Tang
Title: DCCF: Deep Comprehensible Color Filter Learning Framework for High-Resolution Image Harmonization
Abstract:
Image color harmonization algorithm aims to automatically match the color distribution of foreground and background images captured in different conditions. Previous deep learning based models neglect two issues that are critical for practical applications, namely high resolution (HR) image processing and model comprehensibility. In this paper, we propose a novel Deep Comprehensible Color Filter (DCCF) learning framework for high-resolution image harmonization. Specifically, DCCF first downsamples the original input image to its low-resolution (LR) counter-part, then learns four human comprehensible neural filters (i.e. hue, saturation, value and attentive rendering filters) in an end-to-end manner, finally applies these filters to the original input image to get the harmonized result. Benefiting from the comprehensible neural filters, we could provide a simple yet efficient handler for users to cooperate with deep model to get the desired results with very little effort when necessary. Extensive experiments demonstrate the effectiveness of DCCF learning framework and it outperforms state-of-the-art post-processing method on iHarmony4 dataset on images’ full-resolutions by achieving 7.63% and 1.69% relative improvements on MSE and PSNR respectively. Our code is available at https://github.com/rockeyben/DCCF.



Paperid:271
Authors:David Hart; Michael Whitney; Bryan Morse
Title: SelectionConv: Convolutional Neural Networks for Non-Rectilinear Image Data
Abstract:
Convolutional Neural Networks have revolutionized vision applications. There are image domains and representations, however, that cannot be handled by standard CNNs (e.g., spherical images, superpixels). Such data are usually processed using networks and algorithms specialized for each type. In this work, we show that it may not always be necessary to use specialized neural networks to operate on such spaces. Instead, we introduce a new structured graph convolution operator that can copy 2D convolution weights, transferring the capabilities of already trained traditional CNNs to our new graph network. This network can then operate on any data that can be represented as a positional graph. By converting non-rectilinear data to a graph, we can apply these convolutions on these irregular image domains without requiring training on large domain-specific datasets.



Paperid:272
Authors:Jingtang Liang; Xiaodong Cun; Chi-Man Pun; Jue Wang
Title: Spatial-Separated Curve Rendering Network for Efficient and High-Resolution Image Harmonization
Abstract:
Image harmonization aims to modify the color of the composited region according to the specific background. Previous works model this task as a pixel-wise image translation using UNet family structures. However, the model size and computational cost limit the ability of their models on edge devices and higher-resolution images. In this paper, we propose a spatial-separated curve rendering network (S2CRNet), a novel framework to prove that the simple global editing can effectively address this task as well as the challenge of high-resolution image harmonization for the first time. In S2CRNet, we design a curve rendering module (CRM) using spatial-specific knowledge to generate the parameters of the piece-wise curve mapping in the foreground region and we can directly render the original high-resolution images using the learned color curve. Besides, we also make two extensions of the proposed framework via cascaded refinement and semantic guidance. Experiments show that the proposed method reduces more than 90% of parameters compared with previous methods but still achieves the state-of-the-art performance on 3 benchmark datasets. Moreover, our method can work smoothly on higher resolution images with much lower GPU computational resources. The source codes are available at: \url{http://github.com/stefanLeong/S2CRNet}.



Paperid:273
Authors:Geonung Kim; Kyoungkook Kang; Seongtae Kim; Hwayoon Lee; Sehoon Kim; Jonghyun Kim; Seung-Hwan Baek; Sunghyun Cho
Title: BigColor: Colorization Using a Generative Color Prior for Natural Images
Abstract:
For realistic and vivid colorization, generative priors have recently been exploited. However, such generative priors often fail for in-the-wild complex images due to their limited representation space. In this paper, we propose BigColor, a novel colorization approach that provides vivid colorization for diverse in-the-wild images with complex structures. While previous generative priors are trained to synthesize both image structures and colors, we learn a generative color prior to focus on color synthesis given the spatial structure of an image. In this way, we reduce the burden of synthesizing image structures from the generative prior and expand its representation space to cover diverse images. To this end, we propose a BigGAN-inspired encoder-generator network that uses a spatial feature map instead of a spatially-flattened BigGAN latent code, resulting in an enlarged representation space. Our method enables robust colorization for diverse inputs in a single forward pass, supports arbitrary input resolutions, and provides multi-modal colorization results. We demonstrate that BigColor significantly outperforms existing methods especially on in-the-wild images with complex structures.



Paperid:274
Authors:Cheeun Hong; Sungyong Baik; Heewon Kim; Seungjun Nah; Kyoung Mu Lee
Title: CADyQ: Content-Aware Dynamic Quantization for Image Super-Resolution
Abstract:
Despite breakthrough advances in image super-resolution (SR) with convolutional neural networks (CNNs), SR has yet to enjoy ubiquitous applications due to the high computational complexity of SR networks. Quantization is one of the promising approaches to solve this problem. However, existing methods fail to quantize SR models with a bit-width lower than 8 bits, suffering from severe accuracy loss due to fixed bit-width quantization applied everywhere. In this work, to achieve high average bit-reduction with less accuracy loss, we propose a novel Content-Aware Dynamic Quantization (CADyQ) method for SR networks that allocates optimal bits to local regions and layers adaptively based on the local contents of an input image. To this end, a trainable bit selector module is introduced to determine the proper bit-width and quantization level for each layer and a given local image patch. This module is governed by the quantization sensitivity that is estimated by using both the average magnitude of image gradient of the patch and the standard deviation of the input feature of the layer. The proposed quantization pipeline has been tested on various SR networks and evaluated on several standard benchmarks extensively. Significant reduction in computational complexity and the elevated restoration accuracy clearly demonstrate the effectiveness of the proposed CADyQ framework for SR. Codes are available at https://github.com/Cheeun/CADyQ.



Paperid:275
Authors:Kangfu Mei; Vishal M. Patel; Rui Huang
Title: Deep Semantic Statistics Matching (D2SM) Denoising Network
Abstract:
The ultimate aim of image restoration like denoising is to find an exact correlation between the noisy and clear image domains. But the optimization of end-to-end denoising learning like pixel-wise losses is performed in a sample-to-sample manner, which ignores the intrinsic correlation of images, especially semantics. In this paper, we introduce the Deep Semantic Statistics Matching (D2SM) Denoising Network. It exploits semantic features of pretrained classification networks, then it implicitly matches the probabilistic distribution of clear images at the semantic feature space. By learning to preserve the semantic distribution of denoised images, we empirically find our method significantly improves the denoising capabilities of networks, and the denoised results can be better understood by high-level vision tasks. Comprehensive experiments conducted on the noisy Cityscapes dataset demonstrate the superiority of our method on both the denoising performance and semantic segmentation accuracy. Moreover, the performance improvement observed on our extended tasks including super-resolution and dehazing experiments shows its potentiality as a new general plug-and-play component.



Paperid:276
Authors:Sacha Jungerman; Atul Ingle; Yin Li; Mohit Gupta
Title: 3D Scene Inference from Transient Histograms
Abstract:
Time-resolved image sensors that capture light at pico-to-nanosecond timescales were once limited to niche applications but are now rapidly becoming mainstream in consumer devices. We propose low-cost and low-power imaging modalities that capture scene information from minimal time-resolved image sensors with as few as one pixel. The key idea is to flood illuminate large scene patches (or the entire scene) with a pulsed light source and measure the time-resolved reflected light by integrating over the entire illuminated area. The one-dimensional measured temporal waveform, called transient, encodes both distances and albedoes at all visible scene points and as such is an aggregate proxy for the scene’s 3D geometry. We explore the viability and limitations of the transient waveforms for recovering scene information by itself, and also when combined with traditional RGB cameras. We show that plane estimation can be performed from a single transient and that using only a few more it is possible to recover a depth map of the whole scene. We also show two proof-of-concept hardware prototypes that demonstrate the feasibility of our approach for compact, mobile, and budget-limited applications.



Paperid:277
Authors:Hanyu Wang; Kamal Gupta; Larry Davis; Abhinav Shrivastava
Title: Neural Space-Filling Curves
Abstract:
We present Neural Space-filling Curves (SFCs), a data-driven approach to infer a context-based scan order for a set of images. Linear ordering of pixels forms the basis for many applications such as video scrambling, compression, and auto-regressive models that are used in generative modeling for images. Existing algorithms resort to a fixed scanning algorithm such as Raster scan or Hilbert scan. Instead, our work learns a spatially coherent linear ordering of pixels from the dataset of images using a graph-based neural network. The resulting Neural SFC is optimized for an objective suitable for the downstream task when the image is traversed along with the scan line order. We show the advantage of using Neural SFCs in downstream applications such as image compression. Code available in the supplementary material.



Paperid:278
Authors:An Gia Vien; Chul Lee
Title: Exposure-Aware Dynamic Weighted Learning for Single-Shot HDR Imaging
Abstract:
We propose a novel single-shot high dynamic range (HDR) imaging algorithm based on exposure-aware dynamic weighted learning, which reconstructs an HDR image from a spatially varying exposure (SVE) raw image. First, we recover poorly exposed pixels by developing a network that learns local dynamic filters to exploit local neighboring pixels across color channels. Second, we develop another network that combines only valid features in well-exposed regions by learning exposure-aware feature fusion. Third, we synthesize the raw radiance map by adaptively combining the outputs of the two networks that have different characteristics with complementary information. Finally, a full-color HDR image is obtained by interpolating missing color information. Experimental results show that the proposed algorithm outperforms conventional algorithms significantly on various datasets. The source codes and pretrained models are available at https://github.com/viengiaan/EDWL.



Paperid:279
Authors:Weng-Tai Su; Yi-Chun Hung; Po-Jen Yu; Shang-Hua Yang; Chia-Wen Lin
Title: Seeing through a Black Box: Toward High-Quality Terahertz Imaging via Subspace-and-Attention Guided Restoration
Abstract:
Terahertz (THz) imaging has recently attracted significant attention thanks to its non-invasive, non-destructive, non-ionizing, material-classification, and ultra-fast nature for object exploration and inspection. However, its strong water absorption nature and low noise tolerance lead to undesired blurs and distortions of reconstructed THz images. The performances of existing restoration methods are highly constrained by the diffraction-limited THz signals. To address the problem, we propose a novel Subspace-and-Attention-guided Restoration Network (SARNet) that fuses multi-spectral features of a THz image for effective restoration. To this end, SARNet uses multi-scale branches to extract spatio-spectral features of amplitude and phase which are then fused via shared subspace projection and attention guidance. Here, we experimentally construct a THz time-domain spectroscopy system covering a broad frequency range from 0.1 THz to 4 THz for building up temporal/spectral/spatial/phase/material THz database of hidden 3D objects. Complementary to a quantitative evaluation, we demonstrate the effectiveness of SARNet on 3D THz tomographic reconstruction applications.



Paperid:280
Authors:Nir Shaul; Yoav Y. Schechner
Title: Tomography of Turbulence Strength Based on Scintillation Imaging
Abstract:
Developed areas have plenty of artificial light sources. As the stars, they appear to twinkle, i.e., scintillate. This effect is caused by random turbulence. We leverage this phenomenon in order to reconstruct the spatial distribution the turbulence strength (TS). Sensing is passive, using a multi-view camera setup in a city scale. The cameras sense the scintillation of light sources in the scene. The scintillation signal has a linear model of a line integral over the field of TS. Thus, the TS is recovered by linear tomography analysis. Scintillation offers measurements and TS recovery, which are more informative than tomography based on angle-of-arrival (projection distortion) statistics. We present the background and theory of the method. Then, we describe a large field experiment to demonstrate this idea, using distributed imagers. As far as we know, this work is the first to propose reconstruction of a TS horizontal field, using passive optical scintillation measurements.



Paperid:281
Authors:Jaesung Rim; Geonung Kim; Jungeon Kim; Junyong Lee; Seungyong Lee; Sunghyun Cho
Title: Realistic Blur Synthesis for Learning Image Deblurring
Abstract:
Training learning-based deblurring methods demands a tremendous amount of blurred and sharp image pairs. Unfortunately, existing synthetic datasets are not realistic enough, and deblurring models trained on them cannot handle real blurred images effectively. While real datasets have recently been proposed, they provide limited diversity of scenes and camera settings, and capturing real datasets for diverse settings is still challenging. To resolve this, this paper analyzes various factors that introduce differences between real and synthetic blurred images. To this end, we present RSBlur, a novel dataset with real blurred images and the corresponding sharp image sequences to enable a detailed analysis of the difference between real and synthetic blur. With the dataset, we reveal the effects of different factors in the blur generation process. Based on the analysis, we also present a novel blur synthesis pipeline to synthesize more realistic blur. We show that our synthesis pipeline can improve the deblurring performance on real blurred images.



Paperid:282
Authors:Zaid Tasneem; Giovanni Milione; Yi-Hsuan Tsai; Xiang Yu; Ashok Veeraraghavan; Manmohan Chandraker; Francesco Pittaluga
Title: Learning Phase Mask for Privacy-Preserving Passive Depth Estimation
Abstract:
With over a billion sold each year, cameras are not only becoming ubiquitous, but are driving progress in a wide range of domains such as mixed reality, robotics, and more. However, severe concerns regarding the privacy implications of camera-based solutions currently limit the range of environments where cameras can be deployed. The key question we address is: Can cameras be enhanced with a scalable solution to preserve users’ privacy without degrading their machine intelligence capabilities? Our solution is a novel end-to-end adversarial learning pipeline in which a phase mask placed at the aperture plane of a camera is jointly optimized with respect to privacy and utility objectives. We conduct an extensive design space analysis to determine operating points with desirable privacy-utility tradeoffs that are also amenable to sensor fabrication and real-world constraints. We demonstrate the first working prototype that enables passive depth estimation while inhibiting face identification.



Paperid:283
Authors:Atreyee Saha; Salman S. Khan; Sagar Sehrawat; Sanjana S. Prabhu; Shanti Bhattacharya; Kaushik Mitra
Title: LWGNet – Learned Wirtinger Gradients for Fourier Ptychographic Phase Retrieval
Abstract:
Fourier Ptychographic Microscopy (FPM) is an imaging procedure that overcomes the traditional limit on Space-Bandwidth Product (SBP) of conventional microscopes through computational means. It utilizes multiple images captured using a low numerical aperture (NA) objective and enables high-resolution phase imaging through frequency domain stitching. Existing FPM reconstruction methods can be broadly categorized into two approaches: iterative optimization based methods, which are based on the physics of the forward imaging model, and data-driven methods which commonly employ a feed-forward deep learning framework. We propose a hybrid model-driven residual network that combines the knowledge of the forward imaging system with a deep data-driven network. Our proposed architecture, LWGNet, unrolls traditional Wirtinger flow optimization algorithm into a novel neural network design that enhances the gradient images through complex convolutional blocks. Unlike other conventional unrolling techniques, LWGNet uses fewer stages while performing at par or even better than existing traditional and deep learning techniques, particularly, for low-cost and low dynamic range CMOS sensors. This improvement in performance for low-bit depth and low-cost sensors has the potential to bring down the cost of FPM imaging setup significantly. Finally, we show consistently improved performance on our collected real data.



Paperid:284
Authors:Akshat Dave; Yongyi Zhao; Ashok Veeraraghavan
Title: PANDORA: Polarization-Aided Neural Decomposition of Radiance
Abstract:
Reconstructing an object’s geometry and appearance from multiple images, also known as inverse rendering, is a fundamental problem in computer graphics and vision. Inverse rendering is inherently ill-posed because the captured image is an intricate function of unknown lighting, material properties and scene geometry. Recent progress in representing scene through coordinate-based neural networks has facilitated inverse rendering resulting in impressive geometry reconstruction and novel-view synthesis. Our key insight is that polarization is a useful cue for neural inverse rendering as polarization strongly depends on surface normals and is distinct for diffuse and specular reflectance. With the advent of commodity on-chip polarization sensors, capturing polarization has become practical. We propose PANDORA, a polarimetric inverse rendering approach based on implicit neural representations. From multi-view polarization images of an object, PANDORA jointly extracts the object’s 3D geometry, separates the outgoing radiance into diffuse and specular and estimates the incident illumination. We show that PANDORA outperforms state-of-the-art radiance decomposition techniques. PANDORA outputs clean surface reconstructions free from texture artefacts, models strong specularities accurately and estimates illumination under practical unstructured scenarios.



Paperid:285
Authors:Zhongang Cai; Daxuan Ren; Ailing Zeng; Zhengyu Lin; Tao Yu; Wenjia Wang; Xiangyu Fan; Yang Gao; Yifan Yu; Liang Pan; Fangzhou Hong; Mingyuan Zhang; Chen Change Loy; Lei Yang; Ziwei Liu
Title: HuMMan: Multi-modal 4D Human Dataset for Versatile Sensing and Modeling
Abstract:
4D human sensing and modeling are fundamental tasks in vision and graphics with numerous applications. With the advances of new sensors and algorithms, there is an increasing demand for more versatile datasets. In this work, we contribute HuMMan, a large-scale multi-modal 4D human dataset with 1000 human subjects, 400k sequences and 60M frames. HuMMan has several appealing properties: 1) multi-modal data and annotations including color images, point clouds, keypoints, SMPL parameters, and textured meshes; 2) popular mobile device is included in the sensor suite; 3) a set of 500 actions, designed to cover fundamental movements; 4) multiple tasks such as action recognition, pose estimation, parametric human recovery, and textured mesh reconstruction are supported and evaluated. Extensive experiments on HuMMan voice the need for further study on challenges such as fine-grained action recognition, dynamic human mesh reconstruction, point cloud-based parametric human recovery, and cross-device domain gaps.



Paperid:286
Authors:Songnan Lin; Ye Ma; Zhenhua Guo; Bihan Wen
Title: DVS-Voltmeter: Stochastic Process-Based Event Simulator for Dynamic Vision Sensors
Abstract:
Recent advances in deep learning for event-driven applications with dynamic vision sensors (DVS) primarily rely on training over simulated data. However, most simulators ignore various physics-based characteristics of real DVS, such as the fidelity of event timestamps and comprehensive noise effects. We propose an event simulator, dubbed DVS-Voltmeter, to enable high-performance deep networks for DVS applications. DVS-Voltmeter incorporates the fundamental principle of physics - (1) voltage variations in a DVS circuit, (2) randomness caused by photon reception, and (3) noise effects caused by temperature and parasitic photocurrent - into a stochastic process. With the novel insight into the sensor design and physics, DVS-Voltmeter generates more realistic events, given high frame-rate videos. Qualitative and quantitative experiments show that the simulated events resemble real data. The evaluation on two tasks, i.e., semantic segmentation and intensity-image reconstruction, indicates that neural networks trained with DVS-Voltmeter generalize favorably on real events against state-of- the-art simulators.



Paperid:287
Authors:Yuanhan Zhang; Zhenfei Yin; Jing Shao; Ziwei Liu
Title: Benchmarking Omni-Vision Representation through the Lens of Visual Realms
Abstract:
Though impressive performance has been achieved in specific visual realms (\eg faces, dogs, and places), an omni-vision representation that can generalize to many natural visual domains is highly desirable. Nonetheless, the existing benchmark for evaluating visual representations, such as ImageNet, VTAB-natural, and CLIP benchmark suite, is either limited in the spectrum of realms or built by arbitrarily integrating the current datasets. In this paper, we propose Omni-Realm Benchmark (OmniBenchmark) that enables systematically measuring the generalization ability across a wide range of visual realms. OmniBenchmark firstly integrates the concepts from Wikidata to enlarge the storage of concepts of each sub-tree of WordNet. Then, it leverages expert knowledge from WordNet to define a comprehensive spectrum of 21 semantic realms in the natural domain, which is twice of ImageNet’s. Finally, we manually annotate all 7,372 valid concepts, forming a 21-realm dataset with 1,074,346 images. With OmniBenchmark, we propose a hierarchical instance contrastive learning framework for learning better omni-vision representation, \ie Relational Contrastive learning (ReCo), boosting the performance of representation learning across omni-realms. As the hierarchical semantic relation naturally emerges in the label system of visual datasets, ReCo attracts the representations within the same semantic realm during pre-training, facilitating the model converges faster than conventional contrastive learning when ReCo is further fine-tuned to the specific realm. Extensive experiments demonstrate the superior performance of ReCo over state-of-the-art contrastive learning methods on both ImageNet and OmniBenchmark. Beyond that, We conduct a systematic investigation of recent advances in both architectures (from CNNs to transformers) and learning paradigms (from supervised learning to self-supervised learning) on our benchmark. Multiple practical observations are revealed to facilitate future research.



Paperid:288
Authors:Haiyang Liu; Zihao Zhu; Naoya Iwamoto; Yichen Peng; Zhengqing Li; You Zhou; Elif Bozkurt; Bo Zheng
Title: BEAT: A Large-Scale Semantic and Emotional Multi-modal Dataset for Conversational Gestures Synthesis
Abstract:
Achieving realistic, vivid, and human-like synthesized conversational gestures conditioned on multi-modal data is still an unsolved problem due to the lack of available datasets, models and standard evaluation metrics. To address this, we build Body-Expression-Audio-Text dataset, BEAT, which has i) 76 hours, high-quality, multi-modal data captured from 30 speakers talking with eight different emotions and in four different languages, ii) 32 millions frame-level emotion and semantic relevance annotations. Our statistical analysis on BEAT demonstrates the correlation of conversational gestures with facial expressions, emotions, and semantics, in addition to the known correlation with audio, text, and speaker identity. Based on this observation, we propose a baseline model, Cascaded Motion Network (CaMN), which consists of above six modalities modeled in a cascaded architecture for gesture synthesis. To evaluate the semantic relevancy, we introduce a metric, Semantic Relevance Gesture Recall (SRGR). Qualitative and quantitative experiments demonstrate metrics’ validness, ground truth data quality, and baseline’s state-of-the-art performance. To the best of our knowledge, BEAT is the largest motion capture dataset for investigating human gestures, which may contribute to a number of different research fields, including controllable gesture synthesis, cross-modality analysis, and emotional gesture recognition. The data, code and model are available on https://pantomatrix.github.io/BEAT/"



Paperid:289
Authors:Yuhang Li; Youngeun Kim; Hyoungseob Park; Tamar Geller; Priyadarshini Panda
Title: Neuromorphic Data Augmentation for Training Spiking Neural Networks
Abstract:
Developing neuromorphic intelligence on event-based datasets with Spiking Neural Networks (SNNs) has recently attracted much research attention. However, the limited size of event-based datasets makes SNNs prone to overfitting and unstable convergence. This issue remains unexplored by previous academic works. In an effort to minimize this generalization gap, we propose Neuromorphic Data Augmentation (NDA), a family of geometric augmentations specifically designed for event-based datasets with the goal of significantly stabilizing the SNN training and reducing the generalization gap between training and test performance. The proposed method is simple and compatible with existing SNN training pipelines. Using the proposed augmentation, for the first time, we demonstrate the feasibility of unsupervised contrastive learning for SNNs. We conduct comprehensive experiments on prevailing neuromorphic vision benchmarks and show that NDA yields substantial improvements over previous state-of-the-art results. For example, the NDA-based SNN achieves accuracy gain on CIFAR10-DVS and N-Caltech 101 by 10.1% and 13.7%, respectively. Code is available on GitHub (URL).



Paperid:290
Authors:Hao Zhu; Wayne Wu; Wentao Zhu; Liming Jiang; Siwei Tang; Li Zhang; Ziwei Liu; Chen Change Loy
Title: CelebV-HQ: A Large-Scale Video Facial Attributes Dataset
Abstract:
Large-scale datasets played an indispensable role in the recent success of face generation/editing and significantly facilitate the advances of emerging research fields. However, the academic community still lacks a video dataset with diverse facial attribute annotations, which is crucial for face-related video research. In this paper, we propose a large-scale, high-quality, and diverse video dataset, named the High-Quality Celebrity Video Dataset (CelebV-HQ), with rich facial attribute annotations. CelebV-HQ contains 35,666 video clips involving 15,653 identities and 83 manually labeled facial attributes covering appearance, action, and emotion. We conduct a comprehensive analysis in terms of ethnicity, age, brightness, motion smoothness, head pose diversity, and data quality to demonstrate the diversity and temporal coherence of CelebV-HQ. Besides, its versatility and potential are validated on unconditional video generation and video facial attribute editing tasks. Furthermore, we envision the future potential of CelebV-HQ, as well as the new opportunities and challenges it would bring to related research directions.



Paperid:291
Authors:Alejandro Pardo; Fabian Caba; Juan León Alcázar; Ali Thabet; Bernard Ghanem
Title: MovieCuts: A New Dataset and Benchmark for Cut Type Recognition
Abstract:
Understanding movies and their structural patterns is a crucial task in decoding the craft of video editing. While previous works have developed tools for general analysis, such as detecting characters or recognizing cinematography properties at the shot level, less effort has been devoted to understanding the most basic video edit, the Cut. This paper introduces the Cut type recognition task, which requires modeling multi-modal information. To ignite research in this new task, we construct a large-scale dataset called MovieCuts, which contains 173,967 video clips labeled with ten cut types defined by professionals in the movie industry. We benchmark a set of audio-visual approaches, including some dealing with the problem’s multi-modal nature. Our best model achieves 47.7% mAP, which suggests that the task is challenging and that attaining highly accurate Cut type recognition is an open research problem. Advances in automatic Cut-type recognition can unleash new experiences in the video editing industry, such as movie analysis for education, video re-editing, virtual cinematography, machine-assisted trailer generation, and machine-assisted video editing, among others. Our data and code are publicly available: https://github.com/PardoAlejo/MovieCuts.



Paperid:292
Authors:Paul-Edouard Sarlin; Mihai Dusmanu; Johannes L. Schönberger; Pablo Speciale; Lukas Gruber; Viktor Larsson; Ondrej Miksik; Marc Pollefeys
Title: LaMAR: Benchmarking Localization and Mapping for Augmented Reality
Abstract:
Localization and mapping is the foundational technology for augmented reality (AR) that enables sharing and persistence of digital content in the real world. While significant progress has been made, researchers are still mostly driven by unrealistic benchmarks not representative of real-world AR scenarios. In particular, benchmarks are often based on small-scale datasets with low scene diversity, captured from stationary cameras, and lacking other sensor inputs like inertial, radio, or depth data. Furthermore, ground-truth (GT) accuracy is mostly insufficient to satisfy AR requirements. To close this gap, we introduce a new benchmark with a comprehensive capture and GT pipeline, which allow us to co-register realistic AR trajectories in diverse scenes and from heterogeneous devices at scale. To establish accurate GT, our pipeline robustly aligns the captured trajectories against laser scans in a fully automatic manner. Based on this pipeline, we publish a benchmark dataset of diverse and large-scale scenes recorded with head-mounted and hand-held AR devices. We extend several state-of-the-art methods to take advantage of the AR specific setup and evaluate them on our benchmark. Based on the results, we present novel insights on current research gaps to provide avenues for future work in the community.



Paperid:293
Authors:Fangyi Chen; Han Zhang; Zaiwang Li; Jiachen Dou; Shentong Mo; Hao Chen; Yongxin Zhang; Uzair Ahmed; Chenchen Zhu; Marios Savvides
Title: "Unitail: Detecting, Reading, and Matching in Retail Scene"
Abstract:
To make full use of computer vision technology in stores, it is required to consider the actual needs that fit the characteristics of the retail scene. Pursuing this goal, we introduce the United Retail Datasets (Unitail), a large-scale benchmark of basic visual tasks on products that challenges algorithms for detecting, reading, and matching. With 1.8M quadrilateral-shaped instances annotated, the Unitail offers a detection dataset to align product appearance better. Furthermore, it provides a gallery-style OCR dataset containing 1454 product categories, 30k text regions, and 21k transcriptions to enable robust reading on products and motivate enhanced product matching. Besides benchmarking the datasets using various start-of-the-arts, we customize a new detector for product detection and provide a simple OCR-based matching solution that verifies its effectiveness. The Unitail and its evaluation server are publicly available at https://unitedretail.github.io/"



Paperid:294
Authors:Yunhao Ba; Howard Zhang; Ethan Yang; Akira Suzuki; Arnold Pfahnl; Chethan Chinder Chandrappa; Celso M. de Melo; Suya You; Stefano Soatto; Alex Wong; Achuta Kadambi
Title: Not Just Streaks: Towards Ground Truth for Single Image Deraining
Abstract:
We propose a large-scale dataset of real-world rainy and clean image pairs and a method to remove degradations, induced by rain streaks and rain accumulation, from the image. As there exists no real-world dataset for deraining, current state-of-the-art methods rely on synthetic data and thus are limited by the sim2real domain gap; moreover, rigorous evaluation remains a challenge due to the absence of a real paired dataset. We fill this gap by collecting a real paired deraining dataset through meticulous control of non-rain variations. Our dataset enables paired training and quantitative evaluation for diverse real-world rain phenomena (e.g. rain streaks and rain accumulation). To learn a representation robust to rain phenomena, we propose a deep neural network that reconstructs the underlying scene by minimizing a rain-robust loss between rainy and clean images. Extensive experiments demonstrate that our model outperforms the state-of-the-art deraining methods on real rainy images under various conditions. Project website: https://visual.ee.ucla.edu/gt_rain.htm/.



Paperid:295
Authors:Sanghyuk Chun; Wonjae Kim; Song Park; Minsuk Chang; Seong Joon Oh
Title: ECCV Caption: Correcting False Negatives by Collecting Machine-and-Human-Verified Image-Caption Associations for MS-COCO
Abstract:
Image-Text matching (ITM) is a common task for evaluating the quality of Vision and Language (VL) models. However, existing ITM benchmarks have a significant limitation. They have many missing correspondences, originating from the data construction process itself. For example, a caption is only matched with one image although the caption can be matched with other similar images and vice versa. To correct the massive false negatives, we construct the Extended COCO Validation (ECCV) Caption dataset by supplying the missing associations with machine and human annotators. We employ five state-of-the-art ITM models with diverse properties for our annotation process. Our dataset provides x3.6 positive image-to-caption associations and x8.5 caption-to-image associations compared to the original MS-COCO. We also propose to use an informative ranking-based metric mAP@R, rather than the popular Recall@K (R@K). We re-evaluate the existing 25 VL models on existing and proposed benchmarks. Our findings are that the existing benchmarks, such as COCO 1K R@K, COCO 5K R@K, CxC R@1 are highly correlated with each other, while the rankings change when we shift to the ECCV mAP@R. Lastly, we delve into the effect of the bias introduced by the choice of machine annotator. Source code and dataset are available at https://github.com/naver-ai/eccv-caption"



Paperid:296
Authors:Malte Pedersen; Joakim Bruslund Haurum; Patrick Dendorfer; Thomas B. Moeslund
Title: MOTCOM: The Multi-Object Tracking Dataset Complexity Metric
Abstract:
There exists no comprehensive metric for describing the complexity of Multi-Object Tracking (MOT) sequences. This lack of metrics decreases explainability, complicates comparison of datasets, and reduces the conversation on tracker performance to a matter of leader board position. As a remedy, we present the novel MOT dataset complexity metric (MOTCOM), which is a combination of three sub-metrics inspired by key problems in MOT: occlusion, erratic motion, and visual similarity. The insights of MOTCOM can open nuanced discussions on tracker performance and may lead to a wider acknowledgement of novel contributions developed for either less known datasets or those aimed at solving sub-problems. We evaluate MOTCOM on the comprehensive MOT17, MOT20, and MOTSynth datasets and show that MOTCOM is far better at describing the complexity of MOT sequences compared to the conventional density and number of tracks. Project page at https://vap.aau.dk/motcom.



Paperid:297
Authors:Yuchi Liu; Zhongdao Wang; Tom Gedeon; Liang Zheng
Title: How to Synthesize a Large-Scale and Trainable Micro-Expression Dataset?
Abstract:
This paper does not contain technical novelty but introduces our key discoveries in a data generation protocol, a database and insights. We aim to address the lack of large-scale datasets in micro-expression (MiE) recognition due to the prohibitive cost of data collection, which renders large-scale training less feasible. To this end, we develop a protocol to automatically synthesize large scale MiE training data that allow us to train improved recognition models for real-world test data. Specifically, we discover three types of Action Units (AUs) that can constitute trainable MiEs. These AUs come from real-world MiEs, early frames of macro-expression videos, and the relationship between AUs and expression categories defined by human expert knowledge. With these AUs, our protocol then employs large numbers of face images of various identities and an off-the-shelf face generator for MiE synthesis, yielding the MiE-X dataset. MiE recognition models are trained or pre-trained on MiE-X and evaluated on real-world test sets, where very competitive accuracy is obtained. Experimental results not only validate the effectiveness of the discovered AUs and MiE-X dataset but also reveal some interesting properties of MiEs: they generalize across faces, are close to early-stage macro-expressions, and can be manually defined.



Paperid:298
Authors:Rakesh Shrestha; Siqi Hu; Minghao Gou; Ziyuan Liu; Ping Tan
Title: A Real World Dataset for Multi-View 3D Reconstruction
Abstract:
We present a dataset of 371 3D models of everyday tabletop objects along with their 320,000 real world RGB and depth images. Accurate annotations of camera poses and object poses for each image are performed in a semi-automated fashion to facilitate the use of the dataset for myriad 3D applications like shape reconstruction, object pose estimation, shape retrieval etc. We primarily focus on learned multi-view 3D reconstruction due to the lack of appropriate real world benchmark for the task and demonstrate that our dataset can fill that gap. The entire annotated dataset along with the source code for the annotation tools and evaluation baselines will be made publicly available. Keywords: Dataset, Multi-view 3D reconstruction"



Paperid:299
Authors:Zenghao Chai; Haoxian Zhang; Jing Ren; Di Kang; Zhengzhuo Xu; Xuefei Zhe; Chun Yuan; Linchao Bao
Title: REALY: Rethinking the Evaluation of 3D Face Reconstruction
Abstract:
The evaluation of 3D face reconstruction results typically relies on a rigid shape alignment between the estimated 3D model and the ground-truth scan. We observe that aligning two shapes with different reference points can largely affect the evaluation results. This poses difficulties for precisely diagnosing and improving a 3D face reconstruction method. In this paper, we propose a novel evaluation approach with a new benchmark REALY, consists of 100 globally aligned face scans with accurate facial keypoints, high-quality region masks, and topology-consistent meshes. Our approach performs region-wise shape alignment and leads to more accurate, bidirectional correspondences during computing the shape errors. The fine-grained, region-wise evaluation results provide us detailed understandings about the performance of state-of-the-art 3D face reconstruction methods. For example, our experiments on single-image based reconstruction methods reveal that DECA performs the best on nose regions, while GANFit performs better on cheek regions. Besides, a new and high-quality 3DMM basis, HIFI3D++, is further derived using the same procedure as we construct REALY to align and retopologize several 3D face datasets. We will release REALY, HIFI3D++, and our new evaluation pipeline at https://realy3dface.com.



Paperid:300
Authors:Liqiang Lin; Yilin Liu; Yue Hu; Xingguang Yan; Ke Xie; Hui Huang
Title: "Capturing, Reconstructing, and Simulating: The UrbanScene3D Dataset"
Abstract:
We present UrbanScene3D, a large-scale data platform for research of urban scene perception and reconstruction. UrbanScene3D contains over 128k high-resolution images covering 16 scenes including large-scale real urban regions and synthetic cities with 136 km2 area in total. The dataset also contains high-precision LiDAR scans and hundreds of image sets with different observation patterns, which provide a comprehensive benchmark to design and evaluate aerial path planning and 3D reconstruction algorithms. In addition, the dataset, which is built on Unreal Engine and Airsim simulator together with the manually annotated unique instance label for each building in the dataset, enables the generation of all kinds of data, e.g., 2D depth maps, 2D/3D bounding boxes, and 3D point cloud/mesh segmentations, etc. The simulator with physical engine and lighting system not only produce variety of data but also enable users to simulate cars or drones in the proposed urban environment for future research. The dataset with aerial path planning and 3D reconstruction benchmark is available at: https://vcc.tech/UrbanScene3"



Paperid:301
Authors:Yuchen Li; Ujjwal Upadhyay; Habib Slim; Tezuesh Varshney; Ahmed Abdelreheem; Arpit Prajapati; Suhail Pothigara; Peter Wonka; Mohamed Elhoseiny
Title: 3D CoMPaT: Composition of Materials on Parts of 3D Things
Abstract:
We present 3D CoMPaT, a richly annotated large-scale dataset of more than 7.19 million rendered compositions of Materials on Parts of 7262 unique 3D Models; 990 compositions per model on average. 3D CoMPaT covers 43 shape categories, 235 unique part names, and 167 unique material classes that can be applied to parts of 3D objects. Each object with the applied part-material compositions is rendered from four equally spaced views as well as four randomized views, leading to a total of 58 million renderings (7.19 million compositions ×8 views). This dataset primarily focuses on stylizing 3D shapes at part-level with compatible materials. We introduce a new task, called Grounded CoMPaT Recognition (GCR), to collectively recognize and ground compositions of materials on parts of 3D objects. We present two variations of this task and adapt state-of-art 2D/3D deep learning methods to solve the problem as baselines for future research. We hope our work will help ease future research on compositional 3D Vision. The dataset and code are publicly available at https://www.3dcompat-dataset.org/"



Paperid:302
Authors:Ju He; Shuo Yang; Shaokang Yang; Adam Kortylewski; Xiaoding Yuan; Jie-Neng Chen; Shuai Liu; Cheng Yang; Qihang Yu; Alan Yuille
Title: "PartImageNet: A Large, High-Quality Dataset of Parts"
Abstract:
It is natural to represent objects in terms of their parts. This has the potential to improve the performance of algorithms for object recognition and segmentation but can also help for downstream tasks like activity recognition. Research on part-based models, however, is hindered by the lack of datasets with per-pixel part annotations. This is partly due to the difficulty and high cost of annotating object parts so it has rarely been done except for humans (where there exists a big literature on part-based models). To help address this problem, we propose PartImageNet, a large, high-quality dataset with part segmentation annotations. It consists of $158$ classes from ImageNet with approximately 24,000 images. PartImageNet is unique because it offers part-level annotations on a general set of classes including non-rigid, articulated objects, while having an order of magnitude larger size compared to existing part datasets (excluding datasets of humans). It can be utilized for many vision tasks including Object Segmentation, Semantic Part Segmentation, Few-shot Learning and Part Discovery. We conduct comprehensive experiments which study these tasks and set up a set of baselines.



Paperid:303
Authors:Dustin Schwenk; Apoorv Khandelwal; Christopher Clark; Kenneth Marino; Roozbeh Mottaghi
Title: A-OKVQA: A Benchmark for Visual Question Answering Using World Knowledge
Abstract:
The Visual Question Answering (VQA) task aspires to provide a meaningful testbed for the development of AI models that can jointly reason over visual and natural language inputs. Despite a proliferation of VQA datasets, this goal is hindered by a set of common limitations. These include a reliance on relatively simplistic questions that are repetitive in both concepts and linguistic structure, little world knowledge needed outside of the paired image, and limited reasoning required to arrive at the correct answer. We introduce A-OKVQA, a crowdsourced dataset composed of a diverse set of about 25K questions requiring a broad base of commonsense and world knowledge to answer. In contrast to the existing knowledge-based VQA datasets, the questions cannot be answered by simply querying a knowledge base, and instead primarily require some form of commonsense reasoning about the scene depicted in the image. We demonstrate the potential of this new dataset through a detailed analysis of its contents and baseline performance measurements over a variety of state-of-the-art vision-language models.



Paperid:304
Authors:Bingchen Zhao; Shaozuo Yu; Wufei Ma; Mingxin Yu; Shenxiao Mei; Angtian Wang; Ju He; Alan Yuille; Adam Kortylewski
Title: OOD-CV: A Benchmark for Robustness to Out-of-Distribution Shifts of Individual Nuisances in Natural Images
Abstract:
Enhancing the robustness of vision algorithms in real-world scenarios is challenging. One reason is that existing robustness benchmarks are limited, as they either rely on synthetic data or ignore the effects of individual nuisance factors. We introduce ROBIN, a benchmark dataset that includes out-of-distribution examples of 10 object categories in terms of pose, shape, texture, context and the weather conditions, and enables benchmarking models for image classification, object detection, and 3D pose estimation. Our experiments using popular baseline methods reveal that: 1) Some nuisance factors have a much stronger negative effect on the performance compared to others, also depending on the vision task. 2) Current approaches to enhance robustness have only marginal effects, and can even reduce robustness. 3) We do not observe significant differences between convolutional and transformer architectures. We believe our dataset provides a rich testbed to study robustness and will help push forward research in this area.



Paperid:305
Authors:Minjun Kang; Jaesung Choe; Hyowon Ha; Hae-Gon Jeon; Sunghoon Im; In So Kweon; Kuk-Jin Yoon
Title: Facial Depth and Normal Estimation Using Single Dual-Pixel Camera
Abstract:
Recently, Dual-Pixel (DP) sensors have been adopted in many imaging devices. However, despite their various advantages, DP sensors are used just for faster auto-focus and aesthetic image captures, and research on their usage for 3D facial understanding has been limited due to the lack of datasets and algorithmic designs that exploit parallax in DP images. It is also because the baseline of sub-aperture images is extremely narrow, and parallax exists in the defocus blur region. In this paper, we introduce a DP-oriented Depth/Normal estimation network that reconstructs the 3D facial geometry. In addition, to train the network, we collect DP facial data with more than 135K images for 101 persons captured with our multi-camera structured light systems. It contains ground-truth 3D facial models including depth map and surface normal in metric scale. Our dataset allows the proposed network to be generalized for 3D facial depth/normal estimation. The proposed network consists of two novel modules: Adaptive Sampling Module (ASM) and Adaptive Normal Module (ANM), which are specialized in handling the defocus blur in DP images. Finally, we demonstrate that the proposed method achieves state-of-the-art performances over recent DP-based depth/normal estimation methods.



Paperid:306
Authors:Dawit Mureja Argaw; Fabian Caba; Joon-Young Lee; Markus Woodson; In So Kweon
Title: The Anatomy of Video Editing: A Dataset and Benchmark Suite for AI-Assisted Video Editing
Abstract:
Machine learning is transforming the video editing industry. Recent advances in computer vision have leveled-up video editing tasks such as intelligent reframing, rotoscoping, color grading, or applying digital makeups. However, most of the solutions have focused on video manipulation and VFX. This work introduces the Anatomy of Video Editing, a dataset, and benchmark, to foster research in AI-assisted video editing. Our benchmark suite focuses on video editing tasks, beyond visual effects, such as automatic footage organization and assisted video assembling. To enable research on these fronts, we annotate more than 1.5M tags, with relevant concepts to cinematography, from 196176 shots sampled from movie scenes. We establish competitive baseline methods and detailed analyses for each of the tasks. We hope our work sparks innovative research towards underexplored areas of AI-assisted video editing.



Paperid:307
Authors:Dan Ruta; Andrew Gilbert; Pranav Aggarwal; Naveen Marri; Ajinkya Kale; Jo Briggs; Chris Speed; Hailin Jin; Baldo Faieta; Alex Filipkowski; Zhe Lin; John Collomosse
Title: StyleBabel: Artistic Style Tagging and Captioning
Abstract:
We present StyleBabel, a unique open access dataset of natural language captions and free-form tags describing the artistic style of over 135K digital artworks, collected via a novel participatory method from experts studying at specialist art and design schools. StyleBabel was collected via an iterative method, inspired by ‘Grounded Theory’: a qualitative approach that enables annotation while co-evolving a shared language for fine-grained artistic style attribute description. We demonstrate several downstream tasks for StyleBabel, adapting the recent ALADIN architecture for fine-grained style similarity, to train cross-modal embeddings for: 1) free-form tag generation; 2) natural language description of artistic style; 3) fine-grained text search of style. To do so, we extend ALADIN with recent advances in Visual Transformer (ViT) and cross-modal representation learning, achieving a state of the art accuracy in fine-grained style retrieval.



Paperid:308
Authors:Hang Xu; Qiang Zhao; Yike Ma; Xiaodong Li; Peng Yuan; Bailan Feng; Chenggang Yan; Feng Dai
Title: PANDORA: A Panoramic Detection Dataset for Object with Orientation
Abstract:
Panoramic images have become increasingly popular as omnidirectional panoramic technology has advanced. Many datasets and works resort to object detection to better understand the content of the panoramic image. These datasets and detectors use a Bounding Field of View (BFoV) as a bounding box in panoramic images. However, we observe that the object instances in panoramic images often appear with arbitrary orientations. It indicates that BFoV as a bounding box is inappropriate, limiting the performance of detectors. This paper proposes a new bounding box representation, Rotated Bounding Field of View (RBFoV), for the panoramic image object detection task. Then, based on the RBFoV, we present a PANoramic Detection dataset for Object with oRientAtion (PANDORA). Finally, based on PANDORA, we evaluate the current state-of-the-art panoramic image object detection methods and design an anchor-free object detector called R-CenterNet for panoramic images. Compared with these baselines, our R-CenterNet shows its advantages in terms of detection performance. Our PANDORA dataset and source code are available at https://github.com/tdsuper/SphericalObjectDetection.



Paperid:309
Authors:Pinaki Nath Chowdhury; Aneeshan Sain; Ayan Kumar Bhunia; Tao Xiang; Yulia Gryaditskaya; Yi-Zhe Song
Title: FS-COCO: Towards Understanding of Freehand Sketches of Common Objects in Context
Abstract:
We advance sketch research to scenes with the first dataset of freehand scene sketches, FSCOCO. With practical applications in mind, we collect sketches that convey well scene content but can be sketched within a few minutes by a person with any sketching skills. Our dataset comprises 10,000 freehand scene vector sketches with per point space-time information by 100 non-expert individuals, offering both object- and scene-level abstraction. Each sketch is augmented with its text description. Using our dataset, we study for the first time the problem of fine-grained image retrieval from freehand scene sketches and sketch captions. We draw insights on: (i) Scene salience encoded in sketches using the strokes temporal order; (ii) Performance comparison of image retrieval from a scene sketch and an image caption; (iii) Complementarity of information in sketches and image captions, as well as the potential benefit of combining the two modalities. In addition, we extend a popular vector sketch LSTM-based encoder to handle sketches with larger complexity than was supported by previous work. Namely, we propose a hierarchical sketch decoder, which we leverage at a sketch-specific “pretext” task. Our dataset enables for the first time research on freehand scene sketch understanding and its practical applications. We will release the dataset upon acceptance.



Paperid:310
Authors:Grant Van Horn; Rui Qian; Kimberly Wilber; Hartwig Adam; Oisin Mac Aodha; Serge Belongie
Title: Exploring Fine-Grained Audiovisual Categorization with the SSW60 Dataset
Abstract:
We present a new benchmark dataset, Sapsucker Woods 60 (SSW60), for advancing research on audiovisual fine-grained categorization. While our community has made great strides in fine-grained visual categorization on images, the counterparts in audio and video fine-grained categorization are relatively unexplored. To encourage advancements in this space, we have carefully constructed the SSW60 dataset to enable researchers to experiment with classifying the same set of categories in three different modalities: images, audio, and video. The dataset covers 60 species of birds and is comprised of images from existing datasets, and brand new, expert curated audio and video datasets. We thoroughly benchmark audiovisual classification performance and modality fusion experiments through the use of state-of-the-art transformer methods. Our findings show that performance of audiovisual fusion methods is better than using exclusively image or audio based methods for the task of video classification. We also present interesting modality transfer experiments, enabled by the unique construction of SSW60 to encompass three different modalities. We hope the SSW60 dataset and accompanying baselines spur research in this fascinating area.



Paperid:311
Authors:Justin Kay; Peter Kulits; Suzanne Stathatos; Siqi Deng; Erik Young; Sara Beery; Grant Van Horn; Pietro Perona
Title: The Caltech Fish Counting Dataset: A Benchmark for Multiple-Object Tracking and Counting
Abstract:
We present the Caltech Fish Counting Dataset (CFC), a large-scale dataset for detecting, tracking, and counting fish in sonar videos. We identify sonar videos as a rich source of data for advancing low signal-to-noise computer vision applications and tackling domain generalization in multiple-object tracking (MOT) and counting. In comparison to existing MOT and counting datasets, which are largely restricted to videos of people and vehicles in cities, CFC is sourced from a natural-world domain where targets are not easily resolvable and appearance features cannot be easily leveraged for target re-identification. With over half a million annotations in over 1,500 videos sourced from seven different sonar cameras, CFC allows researchers to train MOT and counting algorithms and evaluate generalization performance at unseen test locations. We perform extensive baseline experiments and identify key challenges and opportunities for advancing the state of the art in generalization in MOT and counting.



Paperid:312
Authors:Andrea Burns; Deniz Arsan; Sanjna Agrawal; Ranjitha Kumar; Kate Saenko; Bryan A. Plummer
Title: A Dataset for Interactive Vision-Language Navigation with Unknown Command Feasibility
Abstract:
Vision-language navigation (VLN), in which an agent follows language instruction in a visual environment, has been studied under the premise that the input command is fully feasible in the environment. Yet in practice, a request may not be possible due to language ambiguity or environment changes. To study VLN with unknown command feasibility, we introduce a new dataset Mobile app Tasks with Iterative Feedback (MoTIF), where the goal is to complete a natural language command in a mobile app. Mobile apps provide a scalable domain to study real downstream uses of VLN methods. Moreover, mobile app commands provide instruction for interactive navigation, as they result in action sequences with state changes via clicking, typing, or swiping. MoTIF is the first to include feasibility annotations, containing both binary feasibility labels and fine-grained labels for why tasks are unsatisfiable. We further collect follow-up questions for ambiguous queries to enable research on task uncertainty resolution. Equipped with our dataset, we propose the new problem of feasibility prediction, in which a natural language instruction and multimodal app environment are used to predict instruction feasibility. MoTIF provides a more realistic app dataset as it contains many diverse environments, high-level goals, and longer action sequences than prior work. We evaluate interactive VLN methods using MoTIF, quantify the generalization ability of current approaches to new app environments, and measure the effect of task feasibility on navigation performance.



Paperid:313
Authors:Davide Moltisanti; Jinyi Wu; Bo Dai; Chen Change Loy
Title: BRACE: The Breakdancing Competition Dataset for Dance Motion Synthesis
Abstract:
Generative models for audio-conditioned dance motion synthesis map music features to dance movements. Models are trained to associate motion patterns to audio patterns, usually without an explicit knowledge of the human body. This approach relies on a few assumptions: strong music-dance correlation, controlled motion data and relatively simple poses and movements. These characteristics are found in all existing datasets for dance motion synthesis, and indeed recent methods can achieve good results. We introduce a new dataset aiming to challenge these common assumptions, compiling a set of dynamic dance sequences displaying complex human poses. We focus on breakdancing which features acrobatic moves and tangled postures. We source our data from the Red Bull BC One competition videos. Estimating human keypoints from these videos is difficult due to the complexity of the dance, as well as the multiple moving cameras recording setup. We adopt a hybrid labelling pipeline leveraging deep estimation models as well as manual annotations to obtain good quality keypoint sequences at a reduced cost. Our efforts produced the BRACE dataset, which contains over 3 hours and 30 minutes of densely annotated poses. We test state-of-the-art methods on BRACE, showing their limitations when evaluated on complex sequences. Our dataset can readily foster advance in dance motion synthesis. With intricate poses and swift movements, models are forced to go beyond learning a mapping between modalities and reason more effectively about body structure and movements.



Paperid:314
Authors:Davide Morelli; Matteo Fincato; Marcella Cornia; Federico Landi; Fabio Cesari; Rita Cucchiara
Title: Dress Code: High-Resolution Multi-Category Virtual Try-On
Abstract:
Image-based virtual try-on strives to transfer the appearance of a clothing item onto the image of a target person. Prior work focuses mainly on upper-body clothes (e.g. t-shirts, shirts, and tops) and neglects full-body or lower-body items. This shortcoming arises from a main factor: current publicly available datasets for image-based virtual try-on do not account for this variety, thus limiting progress in the field. To address this deficiency, we introduce Dress Code, which contains images of multi-category clothes. Dress Code is more than 3x larger than publicly available datasets for image-based virtual try-on and features high-resolution paired images (1024x768) with front-view, full-body reference models. To generate HD try-on images with high visual quality and rich in details, we propose to learn fine-grained discriminating features. Specifically, we leverage a semantic-aware discriminator that makes predictions at pixel-level instead of image- or patch-level. Extensive experimental evaluation demonstrates that the proposed approach surpasses the baselines and state-of-the-art competitors in terms of visual quality and quantitative results. The Dress Code dataset is publicly available at https://github.com/aimagelab/dress-code.



Paperid:315
Authors:Lars Schmarje; Monty Santarossa; Simon-Martin Schröder; Claudius Zelenka; Rainer Kiko; Jenny Stracke; Nina Volkmann; Reinhard Koch
Title: A Data-Centric Approach for Improving Ambiguous Labels with Combined Semi-Supervised Classification and Clustering
Abstract:
Consistently high data quality is essential for the development of novel loss functions and architectures in the field of deep learning. The existence of such data and labels is usually presumed, while acquiring high-quality datasets is still a major issue in many cases. Subjective annotations by annotators often lead to ambiguous labels in real-world datasets. We propose a data-centric approach to relabel such ambiguous labels instead of implementing the handling of this issue in a neural network. A hard classification is by definition not enough to capture the real-world ambiguity of the data. Therefore, we propose our method ""Data-Centric Classification & Clustering (DC3)"" which combines semi-supervised classification and clustering. It automatically estimates the ambiguity of an image and performs a classification or clustering depending on that ambiguity. DC3 is general in nature so that it can be used in addition to many Semi-Supervised Learning (SSL) algorithms. On average, our approach yields a 7.6% better F1-Score for classifications and a 7.9% lower inner distance of clusters across multiple evaluated SSL algorithms and datasets. Most importantly, we give a proof-of-concept that the classifications and clusterings from DC3 are beneficial as proposals for the manual refinement of such ambiguous labels. Overall, a combination of SSL with our method DC3 can lead to better handling of ambiguous labels during the annotation process.



Paperid:316
Authors:Xiaotong Chen; Huijie Zhang; Zeren Yu; Anthony Opipari; Odest Chadwicke Jenkins
Title: ClearPose: Large-Scale Transparent Object Dataset and Benchmark
Abstract:
Transparent objects are ubiquitous in household settings and pose distinct challenges for visual sensing and perception systems. The optical properties of transparent objects leaves conventional 3D sensors alone unreliable for object depth and pose estimation. These challenges are highlighted by the shortage of large-scale RGB-Depth datasets focusing on transparent objects in real-world settings. In this work, we contribute a large-scale real-world RGB-Depth transparent object dataset named ClearPose to serve as a benchmark dataset for segmentation, scene-level depth completion, and object-centric pose estimation tasks. The ClearPose dataset contains over 350K labeled real-world RGB-Depth frames and 4M instance annotations covering 63 household objects. The dataset includes object categories commonly used in daily life under various lighting and occluding conditions as well as challenging test scenarios such as cases of occlusion by opaque or translucent objects, non-planar orientations, presence of liquids, etc. We benchmark several state-of-the-art depth completion and object pose estimation deep neural networks on ClearPose.



Paperid:317
Authors:Iuliia Pliushch; Martin Mundt; Nicolas Lupp; Visvanathan Ramesh
Title: When Deep Classifiers Agree: Analyzing Correlations between Learning Order and Image Statistics
Abstract:
Although a plethora of architectural variants for deep classification has been introduced over time, recent works have found empirical evidence towards similarities in their training process. It has been hypothesized that neural networks converge not only to similar representations, but also exhibit a notion of empirical agreement on which data instances are learned first. Following in the latter works’ footsteps, we define a metric to quantify the relationship between such classification agreement over time, and posit that the agreement phenomenon can be mapped to core statistics of the investigated dataset. We empirically corroborate this hypothesis across the CIFAR10, Pascal, ImageNet and KTH-TIPS2 datasets. Our findings indicate that agreement seems to be independent of specific architectures, training hyper-parameters or labels, albeit follows an ordering according to image statistics.



Paperid:318
Authors:Kangyeol Kim; Sunghyun Park; Jaeseong Lee; Sunghyo Chung; Junsoo Lee; Jaegul Choo
Title: AnimeCeleb: Large-Scale Animation CelebHeads Dataset for Head Reenactment
Abstract:
We present a novel Animation CelebHeads dataset (AnimeCeleb) to address an animation head reenactment. Different from previous animation head datasets, we utilize a 3D animation models as the controllable image samplers, which can provide a large amount of head images with their corresponding detailed pose annotations. To facilitate a data creation process, we build a semi-automatic pipeline leveraging an open 3D computer graphics software with a developed annotation system. After training with the AnimeCeleb, recent head reenactment models produce high-quality animation head reenactment results, which are not achievable with existing datasets. Furthermore, motivated by metaverse application, we propose a novel pose mapping method and architecture to tackle a cross-domain head reenactment task. During inference, a user can easily transfer one’s motion to an arbitrary animation head. Experiments demonstrate an usefulness of the AnimeCeleb to train animation head reenactment models, and the superiority of our cross-domain head reenactment model compared to state-of-the-art methods. Our dataset and code are available at \href{https://github.com/kangyeolk/AnimeCeleb}{\textit{this url}}.



Paperid:319
Authors:Thomas Hayes; Songyang Zhang; Xi Yin; Guan Pang; Sasha Sheng; Harry Yang; Songwei Ge; Qiyuan Hu; Devi Parikh
Title: MUGEN: A Playground for Video-Audio-Text Multimodal Understanding and GENeration
Abstract:
Multimodal video-audio-text understanding and generation can benefit from datasets that are narrow but rich. The narrowness allows bite-sized challenges that the research community can make progress on. The richness ensures we are making progress along the core challenges. To this end, we present a large-scale video-audio-text dataset MUGEN, collected using the open-sourced platform game CoinRun. We made substantial modifications to make the game richer by introducing audio and enabling new interactions. We trained RL agents with different objectives to navigate the game and interact with 13 objects and characters. This allows us to automatically extract a large collection of diverse videos and associated audio. We sample 375K video clips (3.2s each) and collect text descriptions from human annotators. Each video has additional annotations that are extracted automatically from the game engine, such as accurate semantic maps for each frame and templated textual descriptions. Altogether, MUGEN can help progress research in many tasks in multimodal understanding and generation. We benchmark representative approaches on tasks involving video-audio-text retrieval and generation. Our dataset and code are released at: https://mugen-org.github.io/.



Paperid:320
Authors:Paul Upchurch; Ransen Niu
Title: A Dense Material Segmentation Dataset for Indoor and Outdoor Scene Parsing
Abstract:
A key algorithm for understanding the world is material segmentation, which assigns a label (metal, glass, etc.) to each pixel. We find that a model trained on existing data underperforms in some settings and propose to address this with a large-scale dataset of 3.2 million dense segments on 44,560 indoor and outdoor images, which is 23x more segments than existing data. Our data covers a more diverse set of scenes, objects, viewpoints and materials, and contains a more fair distribution of skin types. We show that a model trained on our data outperforms a state-of-the-art model across datasets and viewpoints. We propose a large-scale scene parsing benchmark and baseline of 0.729 per-pixel accuracy, 0.585 mean class accuracy and 0.420 mean IoU across 46 materials.



Paperid:321
Authors:Athanasios Papaioannou; Baris Gecer; Shiyang Cheng; Grigorios G. Chrysos; Jiankang Deng; Eftychia Fotiadou; Christos Kampouris; Dimitrios Kollias; Stylianos Moschoglou; Kritaphat Songsri-In; Stylianos Ploumpis; George Trigeorgis; Panagiotis Tzirakis; Evangelos Ververas; Yuxiang Zhou; Allan Ponniah; Anastasios Roussos; Stefanos Zafeiriou
Title: MimicME: A Large Scale Diverse 4D Database for Facial Expression Analysis
Abstract:
Recently, Deep Neural Networks (DNNs) have been shown to outperform traditional methods in many disciplines such as computer vision, speech recognition and natural language processing. A prerequisite for the successful application of DNNs is the big number of data. Even though various facial datasets exist for the case of 2D images, there is a remarkable absence of datasets when we have to deal with 3D faces. The available facial datasets are limited either in terms of expressions or in the number of subjects. This lack of large datasets hinders the exploitation of the great advances that DNNs can provide. In this paper, we overcome these limitations by introducing MimicMe, a novel large-scale database of dynamic high-resolution 3D faces. MimicMe contains recordings of 4,700 subjects with a great diversity on age, gender and ethnicity. The recordings are in the form of 4D videos of subjects displaying a multitude of facial behaviours, resulting to over 280,000 3D meshes in total. We have also manually annotated a big portion of these meshes with 3D facial landmarks and they have been categorized in the corresponding expressions. We have also built very powerful blendshapes for parametrising facial behaviour. MimicMe will be made publicly available upon publication and we envision that it will be extremely valuable to researchers working in many problems of face modelling and analysis, including 3D/4D face and facial expression recognition. We conduct several experiments and demonstrate the usefulness of the database for various applications.



Paperid:322
Authors:Yu Qiu; Jing Xu
Title: "Delving into Universal Lesion Segmentation: Method, Dataset, and Benchmark"
Abstract:
Most efforts on lesion segmentation from CT slices focus on one specific lesion type. However, universal and multi-category lesion segmentation is more important because the diagnoses of different body parts are usually correlated and carried out simultaneously. The existing universal lesion segmentation methods are weakly-supervised due to the lack of pixel-level annotation data. To bring this field into the fully-supervised era, we establish a large-scale universal lesion segmentation dataset, SegLesion. We also propose a baseline method for this task. Considering that it is easy to encode CT slices owing to the limited CT scenarios, we propose a Knowledge Embedding Module (KEM) to adapt the concept of dictionary learning for this task. Specifically, KEM first learns the knowledge encoding of CT slices and then embeds the learned knowledge encoding into the deep features of a CT slice to increase the distinguishability. With KEM incorporated, a Knowledge Embedding Network (KEN) is designed for universal lesion segmentation. To extensively compare KEN to previous segmentation methods, we build a large benchmark for SegLesion. KEN achieves state-of-the-art performance and can thus serve as a strong baseline for future research. Data and code will be released.



Paperid:323
Authors:Bing Shuai; Alessandro Bergamo; Uta Büchler; Andrew Berneshawi; Alyssa Boden; Joseph Tighe
Title: Large Scale Real-World Multi-person Tracking
Abstract:
This paper presents a new large scale multi-person tracking dataset. Our dataset is over an order of magnitude larger than currently available high quality multi-object tracking datasets such as MOT17, HiEve, and MOT20 datasets. The lack of large scale training and test data for this task has limited the community’s ability to understand the performance of their tracking systems on a wide range of scenarios and conditions such as variations in person density, actions being performed, weather, and time of day. Our dataset was specifically sourced to provide a wide variety of these conditions and our annotations include rich meta-data such that the performance of a tracker can be evaluated along these different dimensions. The lack of training data has also limited the ability to perform end-to-end training of tracking systems. As such, the highest performing tracking systems all rely on strong detectors trained on external image datasets. We hope that the release of this dataset will enable new lines of research that take advantage of large scale video based training data.



Paperid:324
Authors:Yuzhen Zhang; Wentong Wang; Weizhi Guo; Pei Lv; Mingliang Xu; Wei Chen; Dinesh Manocha
Title: D2-TPred: Discontinuous Dependency for Trajectory Prediction under Traffic Lights
Abstract:
A profound understanding of inter-agent relationships and motion behaviors is important to achieve high-quality planning when navigating in complex scenarios, especially at urban traffic intersections. We present a trajectory prediction approach with respect to traffic lights, D2-TPred, which uses a spatial dynamic interaction graph (SDG) and a behavior dependency graph (BDG) to handle the problem of discontinuous dependency in the spatial-temporal space. Specifically, the SDG is used to capture spatial interactions by reconstructing sub-graphs for different agents with dynamic and changeable characteristics during each frame. The BDG is used to infer motion tendency by modeling the implicit dependency of the current state on priors behaviors, especially the discontinuous motions corresponding to acceleration, deceleration, or turning direction. Moreover, we present a new dataset for vehicle trajectory prediction under traffic lights called VTP-TL. Our experimental results show that our model achieves more than {20.45\% and 20.78\% }improvement in terms of ADE and FDE, respectively, on VTP-TL as compared to other trajectory prediction algorithms. We will release all of the source code, dataset, and the trained model at the time of publication.



Paperid:325
Authors:Jasper Uijlings; Thomas Mensink; Vittorio Ferrari
Title: The Missing Link: Finding Label Relations across Datasets
Abstract:
Computer Vision is driven by the many datasets which can be used for training or evaluating novel methods. Each of these dataset, however, has its own design principles resulting in a different set of labels,different appearance domains and different annotation instructions. In this paper we explore the automatic discovery of visual-semantic relations between labels across datasets. We want to understand how the instances with label a in dataset A relate to the instances with label b in dataset B,are they in an identity, parent/child, or overlap relation? Or is there no visual link between these two? To find relations between labels across datasets,we propose methods based on language, on vision, and on a combination of both. In order to evaluate these we establish ground-truth relations between three datasets: COCO, ADE20k, and Berkeley Deep Drive. Our methods can effectively discover label relations across datasets and the type of the relations. We use these results for a deeper inspection on why instances relate, find missing aspects, and use our relations to create finer-grained annotations. We conclude that label relations cannot be established by looking at the label-name semantics alone, the relations depend highly on how each of the individual datasets was constructed.



Paperid:326
Authors:Keshav Bhandari; Bin Duan; Gaowen Liu; Hugo Latapie; Ziliang Zong; Yan Yan
Title: Learning Omnidirectional Flow in 360° Video via Siamese Representation
Abstract:
Optical flow estimation in omnidirectional videos faces two significant issues: the lack of benchmark datasets and the challenge of adapting perspective video-based methods to accommodate the omnidirectional nature. This paper proposes the first perceptually natural-synthetic omnidirectional benchmark dataset with a 360° field of view, FLOW360, with 40 different videos and 4,000 video frames. We conduct comprehensive characteristic analysis and comparisons between our dataset and existing optical flow datasets, which manifest perceptual realism, uniqueness, and diversity. To accommodate the omnidirectional nature, we present a novel Siamese representation Learning framework for Omnidirectional Flow (SLOF). We train our network in a contrastive manner with a hybrid loss function that combines contrastive loss and optical flow loss. Extensive experiments verify the proposed framework’s effectiveness and show up to 40% performance improvement over the state-of-the-art approaches. Our FLOW360 dataset and code are available at https://siamlof.github.io/.



Paperid:327
Authors:Yu-Yun Tseng; Alexander Bell; Danna Gurari
Title: VizWiz-FewShot: Locating Objects in Images Taken by People with Visual Impairments
Abstract:
We introduce a few-shot localization dataset originating from photographers who authentically were trying to learn about the visual content in the images they took. It includes over 8,000 segmentations of 100 categories in over 4,000 images that were taken by people with visual impairments. Compared to existing few-shot object detection and instance segmentation datasets, our dataset is the first to locate holes in objects (e.g., found in 12.4% of our segmentations), it shows objects that occupy a much larger range of sizes relative to the images, and text is over five times more common in our objects (e.g., found in 24.7% of our segmentations). Analysis of two modern few-shot localization algorithms demonstrates that they generalize poorly to our new dataset. The algorithms commonly struggle to locate objects with holes, very small and very large objects, and objects lacking text. To encourage a larger community to work on these unsolved challenges, we publicly share our annotated few-shot dataset at http://anonymous.



Paperid:328
Authors:Shubham Dokania; Anbumani Subramanian; Manmohan Chandraker; C.V. Jawahar
Title: TRoVE: Transforming Road Scene Datasets into Photorealistic Virtual Environments
Abstract:
High-quality structured data with rich annotations are critical components in intelligent vehicle systems dealing with road scenes. However, data curation and annotation require intensive investments and yield low-diversity scenarios. The recently growing interest in synthetic data raises questions about the scope of improvement in such systems and the amount of manual work still required to produce high volumes and variations of simulated data. This work proposes a synthetic data generation pipeline that utilizes existing datasets, like nuScenes, to address the difficulties and domain-gaps present in simulated datasets. We show that using annotations and visual cues from existing datasets, we can facilitate automated multi-modal data generation, mimicking real scene properties with high-fidelity, along with mechanisms to diversify samples in a physically meaningful way. We demonstrate improvements in mIoU metrics by presenting qualitative and quantitative experiments with real and synthetic data for semantic segmentation on the Cityscapes and KITTI-STEP datasets. All relevant code and data is released on github.



Paperid:329
Authors:Johannes Theodoridis; Jessica Hofmann; Johannes Maucher; Andreas Schilling
Title: Trapped in Texture Bias? A Large Scale Comparison of Deep Instance Segmentation
Abstract:
Do deep learning models for instance segmentation generalize to novel objects in a systematic way? For classification, such behavior has been questioned. In this study, we aim to understand if certain design decisions such as framework, architecture or pre-training contribute to the semantic understanding of instance segmentation. To answer this question, we consider a special case of robustness and compare pre-trained models on a challenging benchmark for object-centric out-of-distribution texture. We do not introduce another method in this work. Instead, we take a step back and evaluate a broad range of existing literature. This includes Cascade and Mask R-CNN, Swin Transformer, BMask, YOLACT(++), DETR, BCNet, SOTR and SOLOv2. We find that YOLACT++, SOTR and SOLOv2 are significantly more robust to out-of-distribution texture than other frameworks. In addition, we show that deeper and dynamic architectures improve robustness whereas training schedules, data augmentation and pre-training have only a minor impact. In summary we evaluate 68 models on 61 versions of MS COCO for a total of 4148 evaluations.



Paperid:330
Authors:Zehui Chen; Zhenyu Li; Shiquan Zhang; Liangji Fang; Qinhong Jiang; Feng Zhao
Title: Deformable Feature Aggregation for Dynamic Multi-modal 3D Object Detection
Abstract:
Point clouds and RGB images are two general perceptional sources in autonomous driving. The former can provide accurate localization of objects, and the latter is denser and richer in semantic information. Recently, AutoAlign presents a learnable paradigm in combining these two modalities for 3D object detection. However, it suffers from high computational cost introduced by the global-wise attention. To solve the problem, we propose Cross-Domain DeformCAFA module in this work. It attends to sparse learnable sampling points for cross-modal relational modeling, which enhances the tolerance to calibration error and greatly speeds up the feature aggregation across different modalities. To overcome the complex GT-AUG under multi-modal settings, we design a simple yet effective cross-modal augmentation strategy on convex combination of image patches given their depth information. Moreover, by carrying out a novel image-level dropout training scheme, our model is able to infer in a dynamic manner. To this end, we propose AutoAlignV2, a faster and stronger multi-modal 3D detection framework, built on top of AutoAlign. Extensive experiments on nuScenes benchmark demonstrate the effectiveness and efficiency of AutoAlignV2. Notably, our best model reaches 72.4 NDS on nuScenes test leaderboard, achieving new state-of-the-art results among all published multi-modal 3D object detectors.



Paperid:331
Authors:Shishir Reddy Vutukur; Ivan Shugurov; Benjamin Busam; Andreas Hutter; Slobodan Ilic
Title: WeLSA: Learning to Predict 6D Pose from Weakly Labeled Data Using Shape Alignment
Abstract:
Object pose estimation is a crucial task in computer vision and augmented reality. One of its key challenges is the difficulty of annotation of real training data and the lack of textured CAD models. Therefore, pipelines which do not require CAD models and which can be trained with few labeled images are desirable. We propose a weakly-supervised approach for object pose estimation from RGB-D data using training sets composed of very few labeled images with pose annotations along with weakly-labeled images with ground truth segmentation masks without pose labels. We achieve this by learning to annotate weakly-labeled training data through shape alignment while simultaneously training a pose prediction network. Point cloud alignment is performed using structure and rotation-invariant feature-based losses. We further learn an implicit shape representation, which allows the method to work without the known CAD model and also contributes to pose alignment and pose refinement during training on weakly labeled images. The experimental evaluation shows that our method achieves state-of-the-art results on LineMOD, Occlusion-LineMOD and TLess despite being trained using relative poses and on only a fraction of labeled data used by the other methods. We also achieve comparable results to state-of-the-art RGB-D based pose estimation approaches even when further reducing the amount of unlabeled training data. In addition, our method works even if relative camera poses are given instead of object pose annotations which are typically easier to obtain.



Paperid:332
Authors:Honghui Yang; Zili Liu; Xiaopei Wu; Wenxiao Wang; Wei Qian; Xiaofei He; Deng Cai
Title: Graph R-CNN: Towards Accurate 3D Object Detection with Semantic-Decorated Local Graph
Abstract:
Two-stage detectors have gained much popularity in 3D object detection. Most two-stage 3D detectors utilize grid points, voxel grids, or sampled keypoints for RoI feature extraction in the second stage. Such methods, however, are inefficient in handling unevenly distributed and sparse outdoor points. This paper solves this problem in three aspects. 1) Dynamic Point Aggregation. We propose the patch search to quickly search points in a local region for each 3D proposal. The dynamic farthest voxel sampling is then applied to evenly sample the points. Especially, the voxel size varies along the distance to accommodate the uneven distribution of points. 2) RoI-graph Pooling. We build local graphs on the sampled points to better model contextual information and mine point relations through iterative message passing. 3) Visual Features Augmentation. We introduce a simple yet effective fusion strategy to compensate for sparse LiDAR points with limited semantic cues. Based on these modules, we construct our Graph R-CNN as the second stage, which can be applied to existing one-stage detectors to consistently improve the detection performance. Extensive experiments show that Graph R-CNN outperforms the state-of-the-art 3D detection models by a large margin on both the KITTI and Waymo Open Dataset. And we rank first place on the KITTI BEV car detection leaderboard.



Paperid:333
Authors:Xuesong Chen; Shaoshuai Shi; Benjin Zhu; Ka Chun Cheung; Hang Xu; Hongsheng Li
Title: MPPNet: Multi-Frame Feature Intertwining with Proxy Points for 3D Temporal Object Detection
Abstract:
Accurate and reliable 3D detection is vital for many applications including autonomous driving vehicles and service robots. In this paper, we present a flexible and high-performance 3D detection frame-work, named MPPNet, for 3D temporal object detection with point cloud sequences. We propose a novel three-hierarchy framework with proxy points for multi-frame feature encoding and interactions to achieve better detection. The three hierarchies conduct per-frame feature encoding, short-clip feature fusion, and whole-sequence feature aggregation, respectively. To enable processing long-sequence point clouds with reasonable computational resources, intra-group feature mixing and inter-group feature attention are proposed to form the second and third feature encoding hierarchies, which are recurrently applied for aggregating multi-frame trajectory features. The proxy points not only act as consistent object representations for each frame, but also serves as the courier to facilitate feature interaction between frames. The experiments on large Waymo Open dataset show that our approach outperforms state-of-the-art methods with large margins when applied to both short (e.g., 4-frame) and long (e.g., 16-frame) point cloud sequences. Code will be publicly available at https://github.com/open-mmlab/OpenPCDet.



Paperid:334
Authors:Jang Hyun Cho; Philipp Krähenbühl
Title: Long-Tail Detection with Effective Class-Margins
Abstract:
Large-scale object detection and instance segmentation faces a severe data imbalance. The finer-grained object classes become, the less frequent they appear in our datasets. However at test-time, we expect a detector that performs well for all classes and not just the most frequent ones. In this paper, we provide a theoretical understanding of the long-trail detection problem. We show how the commonly used mean average precision evaluation metric on an unknown test-set is bound by a margin-based binary classification error on a long-tailed object-detection training set. We optimize margin-based binary classification error with a novel surrogate objective called Effective Class-Margin Loss (ECM). The ECM loss is simple, theoretically well-motivated, and outperforms other heuristic counterparts on LVIS v1 benchmark over a wide range of architecture and detectors. Code is available at https://github.com/janghyuncho/ECM-Loss.



Paperid:335
Authors:Qing Lian; Yanbo Xu; Weilong Yao; Yingcong Chen; Tong Zhang
Title: Semi-Supervised Monocular 3D Object Detection by Multi-View Consistency
Abstract:
The success of monocular 3D object detection highly relies on considerable labeled data, which is costly to obtain. To alleviate the annotation effort, we propose MVC-MonoDet, the first semi-supervised training framework that improves Monocular 3D object detection by enforcing multi-view consistency. In particular, a box-level regularization and an object-level regularization are designed to enforce the consistency of 3D bounding box predictions of the detection model across unlabeled multi-view data (stereo or video). The box-level regularizer requires the model to consistently estimate 3D boxes in different views so that the model can learn cross-view invariant features for 3D detection. The object-level regularizer employs an object-wise photometric consistency loss that mitigates 3D box estimation error through structure from motion (SFM). A key innovation in our approach to effectively utilize these consistency losses from multi-view data is a novel relative depth module that replaces the standard depth module in vanilla SFM. This technique allows the depth estimation to be coupled with the estimated 3D bounding boxes, so that the derivative of consistency regularization can be used to directly optimize the estimated 3D bounding boxes using unlabeled data. We show that the proposed semi-supervised learning techniques effectively improve the performance of 3D detection on the KITTI and nuScenes datasets. We also demonstrate that the framework is flexible and can be adapted to both stereo and video data.



Paperid:336
Authors:Han Wang; Jun Tang; Xiaodong Liu; Shanyan Guan; Rong Xie; Li Song
Title: PTSEFormer: Progressive Temporal-Spatial Enhanced TransFormer towards Video Object Detection
Abstract:
Recent years have witnessed a trend of applying context frames to boost the performance of object detection as video object detection. Existing methods usually aggregate features at one stroke to enhance the feature. These methods, however, usually lack spatial information from neighboring frames and suffer from insufficient feature aggregation. To address the issues, we perform a progressive way to introduce both temporal information and spatial information for an integrated enhancement. The temporal information is introduced by the temporal feature aggregation model (TFAM), by conducting an attention mechanism between the context frames and the target frame (i.e., the frame to be detected). Meanwhile, we employ a Spatial Transition Awareness Model (STAM) to convey the location transition information between each context frame and target frame. Built upon a transformer-based detector DETR, our PTSEFormer also follows an end-to-end fashion to avoid heavy post-processing procedures while achieving 88.1% mAP on the ImageNet VID dataset. Codes are available at https://github.com/Hon-Wong/PTSEFormer.



Paperid:337
Authors:Zhiqi Li; Wenhai Wang; Hongyang Li; Enze Xie; Chonghao Sima; Tong Lu; Yu Qiao; Jifeng Dai
Title: BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers
Abstract:
3D visual perception tasks, including 3D detection and map segmentation based on multi-camera images, are essential for autonomous driving systems. In this work, we present a new framework termed BEVFormer, which learns unified BEV representations with spatiotemporal transformers to support multiple autonomous driving perception tasks. In a nutshell, BEVFormer exploits both spatial and temporal information by interacting with spatial and temporal space through predefined grid-shaped BEV queries. To aggregate spatial information, we design spatial cross-attention that each BEV query extracts the spatial features from the regions of interest across camera views. For temporal information, we propose temporal self-attention to recurrently fuse the history BEV information. Our approach achieves the new state-of-the-art 56.9% in terms of NDS metric on the nuScenes test set, which is 9.0 points higher than previous best arts and on par with the performance of LiDAR-based baselines. We further show that BEVFormer remarkably improves the accuracy of velocity estimation and recall of objects under low visibility conditions. The code will be released at https://github.com/zhiqi-li/BEVFormer.



Paperid:338
Authors:Jiehong Lin; Zewei Wei; Changxing Ding; Kui Jia
Title: Category-Level 6D Object Pose and Size Estimation Using Self-Supervised Deep Prior Deformation Networks
Abstract:
It is difficult to precisely annotate object instances and their semantics in 3D space, and as such, synthetic data are extensively used for these tasks, e.g., category-level 6D object pose and size estimation. However, the easy annotations in synthetic domains bring the downside effect of synthetic-to-real (Sim2Real) domain gap. In this work, we aim to address this issue in the task setting of Sim2Real, unsupervised domain adaptation for category-level 6D object pose and size estimation. We propose a method that is built upon a novel Deep Prior Deformation Network, shortened as DPDN. DPDN learns to deform features of categorical shape priors to match those of object observations, and is thus able to establish deep correspondence in the feature space for direct regression of object poses and sizes. To reduce the Sim2Real domain gap, we formulate a novel self-supervised objective upon DPDN via consistency learning; more specifically, we apply two rigid transformations to each object observation in parallel, and feed them into DPDN respectively to yield dual sets of predictions; on top of the parallel learning, an inter-consistency term is employed to keep cross consistency between dual predictions for improving the sensitivity of DPDN to pose changes, while individual intra-consistency ones are used to enforce self-adaptation within each learning itself. We train DPDN on both training sets of the synthetic CAMERA25 and real-world REAL275 datasets; our results outperform the existing methods on REAL275 test set under both the unsupervised and supervised settings. Ablation studies also verify the efficacy of our designs. Our code is released publicly at https://github.com/JiehongLin/Self-DPDN.



Paperid:339
Authors:Hongyu Zhou; Zheng Ge; Songtao Liu; Weixin Mao; Zeming Li; Haiyan Yu; Jian Sun
Title: Dense Teacher: Dense Pseudo-Labels for Semi-Supervised Object Detection
Abstract:
To date, the most powerful semi-supervised object detectors (SS-OD) are based on pseudo-boxes, which need a sequence of post-processing with fine-tuned hyper-parameters. In this work, we propose replacing the sparse pseudo-boxes with the dense prediction as a united and straightforward form of pseudo-label. Compared to the pseudo-boxes, our Dense Pseudo-Label (DPL) does not involve any post-processing method, thus retaining richer information. We also introduce a region selection technique to highlight the key information while suppressing the noise carried by dense labels. We name our proposed SS-OD algorithm that leverages the DPL as Dense Teacher. On COCO and VOC, Dense Teacher shows superior performance under various settings compared with the pseudo-box-based methods. Code will be available.



Paperid:340
Authors:Pengfei Chen; Xuehui Yu; Xumeng Han; Najmul Hassan; Kai Wang; Jiachen Li; Jian Zhao; Humphrey Shi; Zhenjun Han; Qixiang Ye
Title: Point-to-Box Network for Accurate Object Detection via Single Point Supervision
Abstract:
Object detection using single point supervision has received increasing attention over the years. However, the performance gap between point supervised object detection (PSOD) and bounding box supervised detection remains large. In this paper, we attribute such a large performance gap to the failure of generating high-quality proposal bags which are crucial for multiple instance learning (MIL). To address this problem, we introduce a lightweight alternative to the off-the-shelf proposal (OTSP) method and thereby create the Point-to-Box Network (P2BNet), which can construct a inter-objects balanced proposal bag by generating proposals in an anchor-like way. By fully investigating the accurate position information, P2BNet further constructs an instance-level bag, avoiding the mixture of multiple objects. Finally, a coarse-to-fine policy in a cascade fashion is utilized to improve the IoU between proposals and ground-truth (GT). Benefiting from these strategies, P2BNet is able to produce high-quality instance-level bags for object detection. P2BNet improves the mean average precision (AP) by more than 50% relative to the previous best PSOD method on the MS COCO dataset. It also demonstrates the great potential to bridge the performance gap between point supervised and bounding-box supervised detectors. The code will be released at github.com/ucas-vg/P2BNet.



Paperid:341
Authors:Takehiko Ohkawa; Yu-Jhe Li; Qichen Fu; Ryosuke Furuta; Kris M. Kitani; Yoichi Sato
Title: Domain Adaptive Hand Keypoint and Pixel Localization in the Wild
Abstract:
We aim to improve the performance of regressing hand keypoints and segmenting pixel-level hand masks under new imaging conditions (e.g., outdoors) when we only have labeled images taken under very different conditions (e.g., indoors). In the real world, it is important that the model trained for both tasks works under various imaging conditions. However, their variation covered by existing labeled hand datasets is limited. Thus, it is necessary to adapt the model trained on the labeled images (source) to unlabeled images (target) with unseen imaging conditions. While self-training domain adaptation methods (i.e., learning from the unlabeled target images in a self-supervised manner) have been developed for both tasks, their training may degrade performance when the predictions on the target images are noisy. To avoid this, it is crucial to assign a low importance (confidence) weight to the noisy predictions during self-training. In this paper, we propose to utilize the divergence of two predictions to estimate the confidence of the target image for both tasks. These predictions are given from two separate networks, and their divergence helps identify the noisy predictions. To integrate our proposed confidence estimation into self-training, we propose a teacher-student framework where the two networks (teachers) provide supervision to a network (student) for self-training, and the teachers are learned from the student by knowledge distillation. Our experiments show its superiority over state-of-the-art methods in adaptation settings with different lighting, grasping objects, backgrounds, and camera viewpoints. Our method improves by 4% the multi-task score on HO3D compared to the latest adversarial adaptation method. We also validate our method on Ego4D, egocentric videos with rapid changes in imaging conditions outdoors.



Paperid:342
Authors:Wen Wang; Jing Zhang; Yang Cao; Yongliang Shen; Dacheng Tao
Title: Towards Data-Efficient Detection Transformers
Abstract:
Detection transformers have achieved competitive performance on the sample-rich COCO dataset. However, we show most of them suffer from significant performance drops on small-size datasets, like Cityscapes. In other words, the detection transformers are generally data-hungry. To tackle this problem, we empirically analyze the factors that affect data efficiency, through a step-by-step transition from a data-efficient RCNN variant to the representative DETR. The empirical results suggest that sparse feature sampling from local image areas holds the key. Based on this observation, we alleviate the data-hungry issue of existing detection transformers by simply alternating how key and value sequences are constructed in the cross-attention layer, with minimum modifications to the original models. Besides, we introduce a simple yet effective label augmentation method to provide richer supervision and improve data efficiency. Experiments show that our method can be readily applied to different detection transformers and improve their performance on both small-size and sample-rich datasets.



Paperid:343
Authors:Yuhang Zang; Wei Li; Kaiyang Zhou; Chen Huang; Chen Change Loy
Title: Open-Vocabulary DETR with Conditional Matching
Abstract:
Open-vocabulary object detection, which is concerned with the problem of detecting novel objects guided by natural language, has gained increasing attention from the community. Ideally, we would like to extend an open-vocabulary detector such that it can produce bounding box predictions based on user inputs in form of either natural language or exemplar image. This offers great flexibility and user experience for human-computer interaction. To this end, we propose a novel open-vocabulary detector based on DETR---hence the name OV-DETR---which, once trained, can detect any object given its class name or an exemplar image. The biggest challenge of turning DETR into an open-vocabulary detector is that it is impossible to calculate the classification cost matrix of novel classes without access to their labeled images. To overcome this challenge, we formulate the learning objective as a binary matching one between input queries (class name or exemplar image) and the corresponding objects, which learns useful correspondence to generalize to unseen queries during testing. For training, we choose to condition the Transformer decoder on the input embeddings obtained from a pre-trained vision-language model like CLIP, in order to enable matching for both text and image queries. With extensive experiments on LVIS and COCO datasets, we demonstrate that our OV-DETR---the first end-to-end Transformer-based open-vocabulary detector---achieves non-trivial improvements over current state of the arts.



Paperid:344
Authors:Chenhongyi Yang; Mateusz Ochal; Amos Storkey; Elliot J. Crowley
Title: Prediction-Guided Distillation for Dense Object Detection
Abstract:
Real-world object detection models should be cheap and accurate. Knowledge distillation (KD) can boost the accuracy of a small, light detection model by leveraging useful information from a larger teacher model. However, a key challenge is identifying the most informative features produced by the teacher for distillation. In this work, we show that only a very small fraction of features within a ground-truth bounding box are responsible for a teacher’s high detection performance. Based on this, we propose Prediction-Guided Distillation (PGD), which focuses distillation on these key predictive regions of the teacher and yields considerable gains in performance over many existing KD baselines. In addition, we propose an adaptive weighting scheme over the key regions to smooth out their influence and achieve even better performance. Our proposed approach outperforms current state-of-the-art KD baselines on a variety of advanced one-stage detection architectures. Specifically, on the COCO dataset, our method achieves between +3.1% and +4.6% AP improvement using ResNet-101 and ResNet-50 as the teacher and student backbones, respectively. On the CrowdHuman dataset, we achieve +3.2% and +2.0% improvements in MR and AP, also using these backbones. Our code is available at https://github.com/ChenhongyiYang/PGD.



Paperid:345
Authors:Yi-Ting Chen; Jinghao Shi; Zelin Ye; Christoph Mertz; Deva Ramanan; Shu Kong
Title: Multimodal Object Detection via Probabilistic Ensembling
Abstract:
Object detection with multimodal inputs can improve many safety-critical systems such as autonomous vehicles (AVs). Motivated by AVs that operate in both day and night, we study multimodal object detection with RGB and thermal cameras, since the latter provides much stronger object signatures under poor illumination. We explore strategies for fusing information from different modalities. Our key contribution is a probabilistic ensembling technique, ProbEn, a simple non-learned method that fuses together detections from multi-modalities. We derive ProbEn from Bayes’ rule and first principles that assume conditional independence across modalities. Through probabilistic marginalization, ProbEn elegantly handles missing modalities when detectors do not fire on the same object. Importantly, ProbEn also notably improves multimodal detection even when the conditional independence assumption does not hold, e.g., fusing outputs from other fusion methods (both off-the-shelf and trained in-house). We validate ProbEn on two benchmarks containing both aligned (KAIST) and unaligned (FLIR) multimodal images, showing that ProbEn outperforms prior work by more than {\bf 13\%} in relative performance!"



Paperid:346
Authors:Shiyu Zhao; Zhixing Zhang; Samuel Schulter; Long Zhao; Vijay Kumar B G; Anastasis Stathopoulos; Manmohan Chandraker; Dimitris N. Metaxas
Title: Exploiting Unlabeled Data with Vision and Language Models for Object Detection
Abstract:
Building robust and generic object detection frameworks requires scaling to larger label spaces and bigger training datasets. However, it is prohibitively costly to acquire annotations for thousands of categories at a large scale. We propose a novel method that leverages the rich semantics available in recent vision and language models to localize and classify objects in unlabeled images, effectively generating pseudo labels for object detection. Starting with a generic and class-agnostic region proposal mechanism, we use vision and language models to categorize each region of an image into any object category that is required for downstream tasks. We demonstrate the value of the generated pseudo labels in two specific tasks, open-vocabulary detection, where a model needs to generalize to unseen object categories, and semi-supervised object detection, where additional unlabeled images can be used to improve the model. Our empirical evaluation shows the effectiveness of the pseudo labels in both tasks, where we outperform competitive baselines and achieve a novel state-of-the-art for open-vocabulary object detection. Our code is available at https://github.com/xiaofeng94/VL-PLM.



Paperid:347
Authors:Junho Kim; Hojun Jang; Changwoon Choi; Young Min Kim
Title: CPO: Change Robust Panorama to Point Cloud Localization
Abstract:
We present CPO, a fast and robust algorithm that localizes a 2D panorama with respect to a 3D point cloud of a scene possibly containing changes. To robustly handle scene changes, our approach deviates from conventional feature point matching, and focuses on the spatial context provided from panorama images. Specifically, we propose efficient color histogram generation and subsequent robust localization using score maps. By utilizing the unique equivariance of spherical projections, we propose very fast color histogram generation for a large number of camera poses without explicitly rendering images for all candidate poses. We accumulate the regional consistency of the panorama and point cloud as 2D/3D score maps, and use them to weigh the input color values to further increase robustness. The weighted color distribution quickly finds good initial poses and achieves stable convergence for gradient-based optimization. CPO is lightweight and achieves effective localization in all tested scenarios, showing stable performance despite scene changes, repetitive structures, or featureless regions, which are typical challenges for visual localization with perspective cameras.



Paperid:348
Authors:Jianyun Xu; Zhenwei Miao; Da Zhang; Hongyu Pan; Kaixuan Liu; Peihan Hao; Jun Zhu; Zhengyang Sun; Hongmin Li; Xin Zhan
Title: INT: Towards Infinite-Frames 3D Detection with an Efficient Framework
Abstract:
It is natural to construct a multi-frame instead of a single-frame 3D detector for a continuous-time stream. Although increasing the number of frames might improve performance, previous multi-frame studies only used very limited frames to build their systems due to the dramatically increased computational and memory cost. To address these issues, we propose a novel on-stream training and prediction framework that, in theory, can employ an infinite number of frames while keeping the same amount of computation as a single-frame detector. This infinite framework (INT), which can be used with most existing detectors, is utilized, for example, on the popular CenterPoint, with significant latency reductions and performance improvements. We’ve also conducted extensive experiments on two large-scale datasets, nuScenes and Waymo Open Dataset, to demonstrate the scheme’s effectiveness and efficiency. By employing INT on CenterPoint, we can get around 7% (Waymo) and 15% (nuScenes) performance boost with only 2 4ms latency overhead, and currently SOTA on the Waymo 3D Detection leaderboard.



Paperid:349
Authors:Mingxiang Liao; Fang Wan; Yuan Yao; Zhenjun Han; Jialing Zou; Yuze Wang; Bailan Feng; Peng Yuan; Qixiang Ye
Title: End-to-End Weakly Supervised Object Detection with Sparse Proposal Evolution
Abstract:
Conventional methods for weakly supervised object detection (WSOD) typically enumerate dense proposals and select the discriminative proposals as objects. However, these two-stage “enumerate-and-select” methods suffer object feature ambiguity brought by dense proposals and low detection efficiency caused by the proposal enumeration procedure. In this study, we propose a sparse proposal evolution (SPE) approach, which advances WSOD from the two-stage pipeline with dense proposals to an end-to-end framework with sparse proposals. SPE is built upon a visual transformer equipped with a seed proposal generation (SPG) branch and a sparse proposal refinement (SPR) branch. SPG generates high-quality seed proposals by taking advantage of the cascaded self-attention mechanism of the visual transformer, and SPR trains the detector to predict sparse proposals which are supervised by the seed proposals in a one-to-one matching fashion. SPG and SPR are iteratively performed so that seed proposals update to accurate supervision signals and sparse proposals evolve to precise object regions. Experiments on VOC and COCO object detection datasets show that SPE outperforms the state-of-the-art end-to-end methods by 7.0% mAP and 8.1% AP50. It is an order of magnitude faster than the two-stage methods, setting the first solid baseline for end-to-end WSOD with sparse proposals.



Paperid:350
Authors:Qi Zhang; Antoni B. Chan
Title: Calibration-Free Multi-View Crowd Counting
Abstract:
Deep learning-based multi-view crowd counting (MVCC) has been proposed to handle scenes with large size, in irregular shape, or with severe occlusions. The current MVCC methods require camera calibrations in both training and testing, limiting the real application scenarios of MVCC. To extend and apply MVCC to more practical situations, in this paper we propose calibration-free multi-view crowd counting (CF-MVCC), which obtains the scene-level count directly from the density map predictions for each camera view without needing the camera calibrations in the test. Specifically, the proposed CF-MVCC method first estimates the homography matrix to align each pair of camera views, and then estimates a matching probability map for each camera-view pair. Based on the matching maps of all camera-view pairs, a weight map for each camera view is predicted, which represents how many cameras can reliably see a given pixel in the camera view. Finally, using the weight maps, the total scene-level count is obtained as a simple weighted sum of the density maps for the camera views. Experiments are conducted on several multi-view counting datasets, and promising performance is achieved compared to calibrated MVCC methods that require camera calibrations as input and use scene-level density maps as supervision.



Paperid:351
Authors:Zhenyu Li; Zehui Chen; Ang Li; Liangji Fang; Qinhong Jiang; Xianming Liu; Junjun Jiang
Title: Unsupervised Domain Adaptation for Monocular 3D Object Detection via Self-Training
Abstract:
Monocular 3D object detection (Mono3D) has achieved unprecedented success with the advent of deep learning techniques and emerging large-scale autonomous driving datasets. However, drastic performance degradation remains an unwell-studied challenge for practical cross-domain deployment as the lack of labels on the target domain. In this paper, we first comprehensively investigate the significant underlying factor of the domain gap in Mono3D, where the critical observation is a depth-shift issue caused by the geometric misalignment of domains. Then, we propose STMono3D, a new self-teaching framework for unsupervised domain adaptation on Mono3D. To mitigate the depth-shift, we introduce the geometry-aligned multi-scale training strategy to disentangle the camera parameters and guarantee the geometry consistency of domains. Based on this, we develop a teacher-student paradigm to generate adaptive pseudo labels on the target domain. Benefiting from the end-to-end framework that provides richer information of the pseudo labels, we propose the quality-aware supervision strategy to take instance-level pseudo confidences into account and improve the effectiveness of the target-domain training process. Moreover, the positive focusing training strategy and dynamic threshold are proposed to handle tremendous FN and FP pseudo samples. STMono3D achieves remarkable performance on all evaluated datasets and even surpasses fully supervised results on the KITTI 3D object detection dataset. To the best of our knowledge, this is the first study to explore effective UDA methods for Mono3D.



Paperid:352
Authors:Xiangrui Zhao; Sheng Yang; Tianxin Huang; Jun Chen; Teng Ma; Mingyang Li; Yong Liu
Title: SuperLine3D: Self-Supervised Line Segmentation and Description for LiDAR Point Cloud
Abstract:
Poles and building edges are frequently observable objects on urban roads, conveying reliable hints for various computer vision tasks. To repetitively extract them as features and perform association between discrete LiDAR frames for registration, we propose the first learning-based feature segmentation and description model for 3D lines in LiDAR point cloud. To train our model without the time consuming and tedious data labeling process, we first generate synthetic primitives for the basic appearance of target lines, and build an iterative line auto-labeling process to gradually refine line labels on real LiDAR scans. Our segmentation model can extract lines under arbitrary scale perturbations, and we use shared EdgeConv encoder layers to train the two segmentation and descriptor heads jointly. Base on the model, we can build a highly-available global registration module for point cloud registration, in conditions without initial transformation hints. Experiments have demonstrated that our line-based registration method is highly competitive to state-of-the-art point-based approaches. Our code is available at https://github.com/zxrzju/SuperLine3D.git.



Paperid:353
Authors:Yanghao Li; Hanzi Mao; Ross Girshick; Kaiming He
Title: Exploring Plain Vision Transformer Backbones for Object Detection
Abstract:
We explore the plain, non-hierarchical Vision Transformer (ViT) as a backbone network for object detection. This design enables the original ViT architecture to be fine-tuned for object detection without needing to redesign a hierarchical backbone for pre-training. With minimal adaptations for fine-tuning, our plain-backbone detector can achieve competitive results. Surprisingly, we observe: (i) it is sufficient to build a simple feature pyramid from a single-scale feature map (without the common FPN design) and (ii) it is sufficient to use window attention (without shifting) aided with very few cross-window propagation blocks. With plain ViT backbones pre-trained as Masked Autoencoders (MAE), our detector, named ViTDet, can compete with the previous leading methods that were all based on hierarchical backbones, reaching up to 61.3 AP_box on the COCO dataset using only ImageNet-1K pre-training. We hope our study will draw attention to research on plain-backbone detectors. Code for ViTDet is available in Detectron2.



Paperid:354
Authors:Ziyi Dong; Pengxu Wei; Liang Lin
Title: Adversarially-Aware Robust Object Detector
Abstract:
Object detection, as a fundamental computer vision task, has achieved a remarkable progress with the emergence of deep neural networks. Nevertheless, few works explore the adversarial robustness of object detectors to resist adversarial attacks for practical applications in various real-world scenarios. Detectors have been greatly challenged by unnoticeable perturbation, with sharp performance drop on clean images and extremely poor performance on adversarial images. In this work, we empirically explore the model training for adversarial robustness in object detection, which greatly attributes to the conflict between learning clean images and adversarial images. To mitigate this issue, we propose a Robust Detector (RobustDet) based on adversarially-aware convolution to disentangle gradients for model learning on clean and adversarial images. RobustDet also employs the Adversarial Image Discriminator (AID) and Consistent Features with Reconstruction (CFR) to ensure a reliable robustness. Extensive experiments on PASCAL VOC and MS-COCO demonstrate that our model effectively disentangles gradients and significantly enhances the detection robustness with maintaining the detection ability on clean images.



Paperid:355
Authors:Luting Wang; Xiaojie Li; Yue Liao; Zeren Jiang; Jianlong Wu; Fei Wang; Chen Qian; Si Liu
Title: HEAD: HEtero-Assists Distillation for Heterogeneous Object Detectors
Abstract:
Conventional knowledge distillation (KD) methods for object detection mainly concentrate on homogeneous teacher-student detectors. However, the design of a lightweight detector for deployment is often significantly different from a high-capacity detector. Thus, we investigate KD among heterogeneous teacher-student pairs for a wide application. We observe that the core difficulty for heterogeneous KD (hetero-KD) is the significant semantic gap between the backbone features of heterogeneous detectors due to the different optimization manners. Conventional homogeneous KD (homo-KD) methods suffer from such a gap and are hard to directly obtain satisfactory performance for hetero-KD. In this paper, we propose the HEtero-Assists Distillation (HEAD) framework, leveraging heterogeneous detection heads as assistants to guide the optimization of the student detector to reduce this gap. In HEAD, the assistant is an additional detection head with the architecture homogeneous to the teacher head attached to the student backbone. Thus, a hetero-KD is transformed into a homo-KD, allowing efficient knowledge transfer from the teacher to the student. Moreover, we extend HEAD into a Teacher-Free HEAD (TF-HEAD) framework when a well-trained teacher detector is unavailable. Our method has achieved significant improvement compared to current detection KD methods. For example, on the MS-COCO dataset, TF-HEAD helps R18 RetinaNet achieve 33.9 mAP (+2.2), while HEAD further pushes the limit to 36.2 mAP (+4.5).



Paperid:356
Authors:Zhenchao Jin; Dongdong Yu; Luchuan Song; Zehuan Yuan; Lequan Yu
Title: You Should Look at All Objects
Abstract:
Feature pyramid network (FPN) is one of the key components for object detectors. However, there is a long-standing puzzle for researchers that the detection performance of large-scale objects are usually suppressed after introducing FPN. To this end, this paper first revisits FPN in the detection framework and reveals the nature of the success of FPN from the perspective of optimization. Then, we point out that the degraded performance of large-scale objects is due to the arising of improper back-propagation paths after integrating FPN. It makes each level of the backbone network only has the ability to look at the objects within a certain scale range. Based on these analysis, two feasible strategies are proposed to enable each level of the backbone to look at all objects in the FPN-based detection frameworks. Specifically, one is to introduce auxiliary objective functions to make each backbone level directly receive the back-propagation signals of various-scale objects during training. The other is to construct the feature pyramid in a more reasonable way to avoid the irrational back-propagation paths. Extensive experiments on the COCO benchmark validate the soundness of our analysis and the effectiveness of our methods. Without bells and whistles, we demonstrate that our method achieves solid improvements (more than 2%) on various detection frameworks: one-stage, two-stage, anchor-based, anchor-free and transformer-based detectors.



Paperid:357
Authors:Xingyi Zhou; Rohit Girdhar; Armand Joulin; Philipp Krähenbühl; Ishan Misra
Title: Detecting Twenty-Thousand Classes Using Image-Level Supervision
Abstract:
Current object detectors are limited in vocabulary size due to the small scale of detection datasets. Image classifiers, on the other hand, reason about much larger vocabularies, as their datasets are larger and easier to collect. We propose Detic, which simply trains the classifiers of a detector on image classification data and thus expands the vocabulary of detectors to tens of thousands of concepts. Unlike prior work, Detic does not need complex assignment schemes to assign image labels to boxes based on model predictions, making it much easier to implement and compatible with a range of detection architectures and backbones. Our results show that Detic yields excellent detectors even for classes without box annotations. It outperforms prior work on both open-vocabulary and long-tail detection benchmarks. Detic provides a gain of 2.4 mAP for all classes and 8.3 mAP for novel classes on the open-vocabulary LVIS benchmark. On the standard LVIS benchmark, Detic obtains 41.7 mAP when evaluated on all classes, or only rare classes, hence closing the gap in performance for object categories with few samples. For the first time, we train a detector with all the twenty-one-thousand classes of the ImageNet dataset and show that it generalizes to new datasets without finetuning. Code is provided in the supplementary.



Paperid:358
Authors:Hongyang Li; Jiehong Lin; Kui Jia
Title: DCL-Net: Deep Correspondence Learning Network for 6D Pose Estimation
Abstract:
Establishment of point correspondence between camera and object coordinate systems is a promising way to solve 6D object poses. However, surrogate objectives of correspondence learning in 3D space are a step away from the true ones of object pose estimation, making the learning suboptimal for the end task. In this paper, we address this shortcoming by introducing a new method of Deep Correspondence Learning Network for direct 6D object pose estimation, shortened as DCL-Net. Specifically, DCL-Net employs dual newly proposed Feature Disengagement and Alignment (FDA) modules to establish, in the feature space, partial-to-partial correspondence and complete-to-complete one for partial object observation and its complete CAD model, respectively, which result in aggregated pose and match feature pairs from two coordinate systems; these two FDA modules thus bring complementary advantages. The match feature pairs are used to learn confidence scores for measuring the qualities of deep correspondence, while the pose ones are weighted by confidence scores for direct object pose regression. A confidence-based pose refinement network is also proposed to further improve pose precision in an iterative manner. Extensive experiments show that DCL-Net outperforms existing methods on three benchmarking datasets, including YCB-Video, LineMOD, and Oclussion-LineMOD; ablation studies also confirm the efficacy of our novel designs. Our code is released publicly at https://github.com/Gorilla-Lab-SCUT/DCL-Net.



Paperid:359
Authors:Tai Wang; Jiangmiao Pang; Dahua Lin
Title: Monocular 3D Object Detection with Depth from Motion
Abstract:
Perceiving 3D objects from monocular inputs is crucial for robotic systems, given its economy compared to multi-sensor settings. It is notably difficult as a single image can not provide any clues for predicting absolute depth values. Motivated by binocular methods for 3D object detection, we take advantage of the strong geometry structure provided by camera ego-motion for accurate object depth estimation and detection. We first make a theoretical analysis on this general two-view case and notice two challenges: 1) Cumulative errors from multiple estimations that make the direct prediction intractable; 2) Inherent dilemmas caused by static cameras and matching ambiguity. Accordingly, we establish the stereo correspondence with a geometry-aware cost volume as the alternative for depth estimation and further compensate it with monocular understanding to address the second problem. Our framework, named Depth from Motion (DfM), then uses the established geometry to lift 2D image features to the 3D space and detects 3D objects thereon. We also present a pose-free DfM to make it usable when the camera pose is unavailable. Our framework outperforms state-of-the-art methods by a large margin on the KITTI benchmark. Detailed quantitative and qualitative analyses also validate our theoretical conclusions. The code is released at https://github.com/Tai-Wang/Depth-from-Motion.



Paperid:360
Authors:Yilin Wen; Xiangyu Li; Hao Pan; Lei Yang; Zheng Wang; Taku Komura; Wenping Wang
Title: DISP6D: Disentangled Implicit Shape and Pose Learning for Scalable 6D Pose Estimation
Abstract:
Scalable 6D pose estimation for rigid objects from RGB images aims at handling multiple objects and generalizing to novel objects. Building on a well-known auto-encoding framework to cope with object symmetry and the lack of labeled training data, we achieve scalability by disentangling the latent representation of auto-encoder into shape and pose sub-spaces. The latent shape space models the similarity of different objects through contrastive metric learning, and the latent pose code is compared with canonical rotations for rotation retrieval. Because different object symmetries induce inconsistent latent pose spaces, we re-entangle the shape representation with canonical rotations to generate shape-dependent pose codebooks for rotation retrieval. We show state-of-the-art performance on two benchmarks containing textureless CAD objects without category and daily objects with categories respectively, and further demonstrate improved scalability by extending to a more challenging setting of daily objects across categories.



Paperid:361
Authors:Sanli Tang; Zhongyu Zhang; Zhanzhan Cheng; Jing Lu; Yunlu Xu; Yi Niu; Fan He
Title: Distilling Object Detectors with Global Knowledge
Abstract:
Knowledge distillation learns a lightweight student model that mimics a cumbersome teacher. Existing methods regard the knowledge as the feature of each instance or their relations, which is the instance-level knowledge only from the teacher model, i.e., the local knowledge. However, the empirical studies show that the local knowledge is much noisy in object detection tasks, especially on the blurred, occluded, or small instances. Thus, a more intrinsic approach is to measure the representations of instances w.r.t. a group of common basis vectors in the two feature spaces of the teacher and the student detectors, i.e., global knowledge. Then, the distilling algorithm can be applied as space alignment. To this end, a novel prototype generation module (PGM) is proposed to find the common basis vectors, dubbed prototypes, in the two feature spaces. Then, a robust distilling module (RDM) is applied to construct the global knowledge based on the prototypes and filtrate noisy global and local knowledge by measuring the discrepancy of the representations in two feature spaces. Experiments with Faster-RCNN and RetinaNet on PASCAL and COCO datasets show that our method achieves the best performance for distilling object detectors with various backbones, which even surpasses the performance of the teacher model. We also show that the existing methods can be easily combined with global knowledge and obtain further improvement. Code is available: https://github.com/hikvision-research/DAVAR-Lab-ML.



Paperid:362
Authors:Jianming Liang; Guanglu Song; Biao Leng; Yu Liu
Title: Unifying Visual Perception by Dispersible Points Learning
Abstract:
We present a conceptually simple, flexible, and universal visual perception head for variant visual tasks, e.g., classification, object detection, instance segmentation and pose estimation, and different frameworks, such as one-stage or two-stage pipelines. Our approach effectively identifies an object in an image while simultaneously generating a high-quality bounding box or contour-based segmentation mask or set of keypoints. The method, called UniHead, views different visual perception tasks as the dispersible points learning via the transformer encoder architecture. Given a fixed spatial coordinate, UniHead adaptively scatters it to different spatial points and reasons about their relations by transformer encoder. It directly outputs the final set of predictions in the form of multiple points, allowing us to perform different visual tasks in different frameworks with the same head design. We show extensive evaluations on ImageNet classification and all three tracks of the COCO suite of challenges, including object detection, instance segmentation and pose estimation. Without bells and whistles, UniHead can unify these visual tasks via a single visual head design and achieve comparable performance compared to expert models developed for each task. We hope our simple and universal UniHead will serve as a solid baseline and help promote universal visual perception research. Code and models are available at https://github.com/Sense-X/UniHead.



Paperid:363
Authors:Gang Li; Xiang Li; Yujie Wang; Yichao Wu; Ding Liang; Shanshan Zhang
Title: PseCo: Pseudo Labeling and Consistency Training for Semi-Supervised Object Detection
Abstract:
In this paper, we delve into two key techniques in Semi-Supervised Object Detection (SSOD), namely pseudo labeling and consistency training. We observe that these two techniques currently neglect some important properties of object detection, hindering efficient learning on unlabeled data. Specifically, for pseudo labeling, existing works only focus on the classification score yet fail to guarantee the localization precision of pseudo boxes; For consistency training, the widely adopted random-resize training only considers the label-level consistency but misses the feature-level one, which also plays an important role in ensuring the scale invariance. To address the problems incurred by noisy pseudo boxes, we design Noisy Pseudo box Learning (NPL) that includes Prediction-guided Label Assignment (PLA) and Positive-proposal Consistency Voting (PCV). PLA relies on model predictions to assign labels and makes it robust to even coarse pseudo boxes; while PCV leverages the regression consistency of positive proposals to reflect the localization quality of pseudo boxes. Furthermore, in consistency training, we propose Multi-view Scale-invariant Learning (MSL) that includes mechanisms of both label- and feature-level consistency, where feature consistency is achieved by aligning shifted feature pyramids between two images with identical content but varied scales. On COCO benchmark, our method, termed PSEudo labeling and COnsistency training (PseCo), outperforms the SOTA (Soft Teacher) by 2.0, 1.8, 2.0 points under 1%, 5%, and 10% labelling ratios, respectively. It also significantly improves the learning efficiency for SSOD, e.g., PseCo halves the training time of the SOTA approach but achieves even better performance. Code is available at https://github.com/ligang-cs/PseCo.



Paperid:364
Authors:Ziteng Cui; Yingying Zhu; Lin Gu; Guo-Jun Qi; Xiaoxiao Li; Renrui Zhang; Zenghui Zhang; Tatsuya Harada
Title: Exploring Resolution and Degradation Clues As Self-Supervised Signal for Low Quality Object Detection
Abstract:
Image restoration algorithms such as super resolution (SR) are indispensable pre-processing modules for object detection in low qual-ity images. Most of these algorithms assume the degradation is fixed andknown a priori. However, in pratical, either the real degrdation or optimalup-sampling ratio rate is unknown or differs from assumption, leading toa deteriorating performance for both the pre-processing module and theconsequent high-level task such as object detection. Here, we propose anovel self-supervised framework to detect objects in degraded low res-olution images. We utilizes the downsampling degradation as a kind oftransformation for self-supervised signals to explore the equivariant representation against various resolutions and other degradation conditions.The Auto Encoding Resolution in Self-supervision (AERIS) frameworkcould further take the advantage of advanced SR architectures with anarbitrary resolution restoring decoder to reconstruct the original corre-spondence from the degraded input image. Both the representation learn-ing and object detection are optimized jointly in an end-to-end trainingfashion. The generic AERIS frameworkcould be implemented on variousmainstream object detection architectures from CNN to Transformer.The extensive experiments show that our methods has achieved supe-rior performance compared with existing methods when facing variantdegradation situations.We will release the open source code.



Paperid:365
Authors:Wufei Ma; Angtian Wang; Alan Yuille; Adam Kortylewski
Title: Robust Category-Level 6D Pose Estimation with Coarse-to-Fine Rendering of Neural Features
Abstract:
We consider the problem of category-level 6D pose estimation from a single RGB image. Our approach represents an object category as a cuboid mesh and learns a generative model of the neural feature activations at each mesh vertex to perform pose estimation through differentiable rendering. A common problem of rendering-based approaches is that they rely on bounding box proposals, which do not convey information about the 3D rotation of the object and are not reliable when objects are partially occluded. Instead, we introduce a coarse-to-fine optimization strategy that utilizes the rendering process to estimate a sparse set of 6D object proposals, which are subsequently refined with gradient-based optimization. The key to enabling the convergence of our approach is a neural feature representation that is trained to be scale- and rotation-invariant using contrastive learning. Our experiments demonstrate an enhanced category-level 6D pose estimation performance compared to prior work, particularly under strong partial occlusion.



Paperid:366
Authors:Maoxun Yuan; Yinyan Wang; Xingxing Wei
Title: "Translation, Scale and Rotation: Cross-Modal Alignment Meets RGB-Infrared Vehicle Detection"
Abstract:
Integrating multispectral data in object detection, especially visible and infrared images, has received great attention in recent years. Since visible (RGB) and infrared (IR) images can provide complementary information to handle light variations, the paired images are used in many fields, such as multispectral pedestrian detection, RGB-IR crowd counting and RGB-IR salient object detection. Compared with natural RGB-IR images, we find detection in aerial RGB-IR images suffers from cross-modal weakly misalignment problems, which are manifested in the position, size and angle deviations of the same object. In this paper, we mainly address the challenge of cross-modal weakly misalignment in aerial RGB-IR images. Specifically, we firstly explain and analyze the cause of the weakly misalignment problem. Then, we propose a Translation-Scale-Rotation Alignment (TSRA) module to address the problem by calibrating the feature maps from these two modalities. The module predicts the deviation between two modality objects through an alignment process and utilizes Modality-Selection (MS) strategy to improve the performance of alignment. Finally, a two-stream feature alignment detector (TSFADet) based on the TSRA module is constructed for RGB-IR object detection in aerial images. With comprehensive experiments on the public DroneVehicle datasets, we verify that our method reduces the effect of the cross-modal misalignment and achieve robust detection results.



Paperid:367
Authors:Chang Xu; Jinwang Wang; Wen Yang; Huai Yu; Lei Yu; Gui-Song Xia
Title: RFLA: Gaussian Receptive Field Based Label Assignment for Tiny Object Detection
Abstract:
Detecting tiny objects is one of the main obstacles hindering the development of object detection. The performance of generic object detectors tends to drastically deteriorate on tiny object detection tasks. In this paper, we point out that either box prior in the anchor-based detector or point prior in the anchor-free detector is sub-optimal for tiny objects. Our key observation is that the current anchor-based or anchor-free label assignment paradigms will incur many outlier tiny-sized ground truth samples, leading to detectors imposing less focus on the tiny objects. To this end, we propose a Gaussian Receptive Field based Label Assignment (RFLA) strategy for tiny object detection. Specifically, RFLA first utilizes the prior information that the feature receptive field follows Gaussian distribution. Then, instead of assigning samples with IoU or center sampling strategy, a new Receptive Field Distance (RFD) is proposed to directly measure the similarity between the Gaussian receptive field and ground truth. Considering that the IoU-threshold based and center sampling strategy are skewed to large objects, we further design a Hierarchical Label Assignment (HLA) module based on RFD to achieve balanced learning for tiny objects. Extensive experiments on four datasets demonstrate the effectiveness of the proposed methods. Especially, our approach outperforms the state-of-the-art competitors with 4.0 AP points on the AI-TOD dataset. Codes are available at https://github.com/Chasel-Tsui/mmdet-rfla.



Paperid:368
Authors:Hualian Sheng; Sijia Cai; Na Zhao; Bing Deng; Jianqiang Huang; Xian-Sheng Hua; Min-Jian Zhao; Gim Hee Lee
Title: Rethinking IoU-Based Optimization for Single-Stage 3D Object Detection
Abstract:
Since Intersection-over-Union (IoU) based optimization maintains the consistency of the final IoU prediction metric and losses, it has been widely used in both regression and classification branches of single-stage 2D object detectors. Recently, several 3D object detection methods adopt IoU-based optimization and directly replace the 2D IoU with 3D IoU. However, such a direct computation in 3D is very costly due to the complex implementation and inefficient backward operations. Moreover, 3D IoU-based optimization is sub-optimal as it is sensitive to rotation and thus can cause training instability and detection performance deterioration. In this paper, we propose a novel Rotation-Decoupled IoU (RDIoU) method that can mitigate the rotation-sensitivity issue, and produce more efficient optimization objectives compared with 3D IoU during the training stage. Specifically, our RDIoU simplifies the complex interactions of regression parameters by decoupling the rotation variable as an independent term, yet preserving the geometry of 3D IoU. By incorporating RDIoU into both the regression and classification branches, the network is encouraged to learn more precise bounding boxes and concurrently overcome the misalignment issue between classification and regression. Extensive experiments on the benchmark KITTI and Waymo Open Dataset validate that our RDIoU method can bring substantial improvement for the single-stage 3D object detection. Our code will be available upon paper acceptance.



Paperid:369
Authors:Yang He; Ravi Garg; Amber Roy Chowdhury
Title: TD-Road: Top-Down Road Network Extraction with Holistic Graph Construction
Abstract:
Graph-based approaches have been becoming increasingly popular in road network extraction, in addition to segmentation-based methods. Road networks are represented as graph structures, being able to explicitly define the topology structures and avoid the ambiguity of segmentation masks, such as between a real junction area and multiple separate roads in different heights. In contrast to the bottom-up graph-based approaches, which rely on orientation information, we propose a novel top-down approach to generate road network graphs with a holistic model, namely TD-Road. We decompose road extraction as two subtasks: key point prediction and connectedness prediction. We directly apply graph structures (i.e., locations of node and connections between them) as training supervisions for neural networks and generate road graph outputs in inference, instead of learning some intermediate properties of a graph structure (e.g., orientations or distances for the next move). Our network integrates a relation inference module with key point prediction, to capture connections between neighboring points and outputs the final road graphs with no post-processing steps required. Extensive experiments are conducted on challenging datasets, including City-Scale and SpaceNet to show the effectiveness and simplicity of our method, that the proposed method achieves remarkable results compared with previous state-of-the-art methods.



Paperid:370
Authors:Shuang Wu; Wenjie Pei; Dianwen Mei; Fanglin Chen; Jiandong Tian; Guangming Lu
Title: Multi-faceted Distillation of Base-Novel Commonality for Few-Shot Object Detection
Abstract:
Most of existing methods for few-shot object detection follow the fine-tuning paradigm, which potentially assumes that the class-agnostic generalizable knowledge can be learned and transferred implicitly from base classes with abundant samples to novel classes with limited samples via such a two-stage training strategy. However, it is not necessarily true since the object detector can hardly distinguish between class-agnostic knowledge and class-specific knowledge automatically without explicit modeling. In this work we propose to learn three types of class-agnostic commonalities between base and novel classes explicitly: recognition-related semantic commonalities, localization-related semantic commonalities and distribution commonalities. We design a unified distillation framework based on a memory bank, which is able to perform distillation of all three types of commonalities jointly and efficiently. Extensive experiments demonstrate that our method can be readily integrated into most of existing fine-tuning based methods and consistently improve the performance by a large margin.



Paperid:371
Authors:Mingzhi Yuan; Zhihao Li; Qiuye Jin; Xinrong Chen; Manning Wang
Title: PointCLM: A Contrastive Learning-Based Framework for Multi-Instance Point Cloud Registration
Abstract:
Multi-instance point cloud registration is the problem of estimating multiple poses of source point cloud instances within a target point cloud. Solving this problem is challenging since inlier correspondences of one instance constitute outliers of all the other instances. Existing methods often rely on time-consuming hypothesis sampling or features leveraging spatial consistency, resulting in limited performance. In this paper, we propose PointCLM, a contrastive learning-based framework for mutli-instance point cloud registration. We first utilize contrastive learning to learn well-distributed deep representations for the input putative correspondences. Then based on these representations, we propose a outlier pruning strategy and a clustering strategy to efficiently remove outliers and assign the remaining correspondences to correct instances. Our method outperforms the state-of-the-art methods on both synthetic and real datasets by a large margin.



Paperid:372
Authors:Haotian Bai; Ruimao Zhang; Jiong Wang; Xiang Wan
Title: Weakly Supervised Object Localization via Transformer with Implicit Spatial Calibration
Abstract:
Weakly Supervised Object Localization (WSOL), which aims to localize objects by only using image-level labels, has attracted much attention because of its low annotation cost in real applications. Recent studies leverage the advantage of self-attention in visual Transformer for long-range dependency to re-active semantic regions, aiming to avoid partial activation in traditional class activation mapping (CAM). However, the long-range modeling in Transformer neglects the inherent spatial coherence of the object, and it usually diffuses the semantic-aware regions far from the object boundary, making localization results significantly larger or far smaller. To address such an issue, we introduce a simple yet effective Spatial Calibration Module (SCM) for accurate WSOL, incorporating semantic similarities of patch tokens and their spatial relationships into a unified diffusion model. Specifically, we introduce a learnable parameter to dynamically adjust the semantic correlations and spatial context intensities for effective information propagation. In practice, SCM is designed as an external module of Transformer, and can be removed during inference to reduce the computation cost. The object-sensitive localization ability is implicitly embedded into the Transformer encoder through optimization in the training phase. It enables the generated attention maps to capture the sharper object boundaries and filter the object-irrelevant background area. Extensive experimental results demonstrate the effectiveness of the proposed method, which significantly outperforms its counterpart TS-CAM on both CUB-200 and ImageNet-1K benchmarks. The code is available at https://github.com/164140757/SCM.



Paperid:373
Authors:Jinze Yu; Jiaming Liu; Xiaobao Wei; Haoyi Zhou; Yohei Nakata; Denis Gudovskiy; Tomoyuki Okuno; Jianxin Li; Kurt Keutzer; Shanghang Zhang
Title: MTTrans: Cross-Domain Object Detection with Mean Teacher Transformer
Abstract:
Recently, DEtection TRansformer (DETR), an end-to-end object detection pipeline, has achieved promising performance. However, it requires large-scale labeled data and suffers from domain shift, especially when no labeled data is available in the target domain. To solve this problem, we propose an end-to-end cross-domain detection Transformer based on the mean teacher framework, MTTrans, which can fully exploit unlabeled target domain data in object detection training and transfer knowledge between domains via pseudo labels. We further propose the comprehensive multi-level feature alignment to improve the pseudo labels generated by the mean teacher framework taking advantage of the cross-scale self-attention mechanism in Deformable DETR. Image and object features are aligned at the local, global, and instance levels with domain query-based feature alignment (DQFA), bi-level graph-based prototype alignment (BGPA), and token-wise image feature alignment (TIFA). On the other hand, the unlabeled target domain data pseudo-labeled and available for the object detection training by the mean teacher framework can lead to better feature extraction and alignment. Thus, the mean teacher framework and the comprehensive multi-level feature alignment can be optimized iteratively and mutually based on the architecture of Transformers. Extensive experiments demonstrate that our proposed method achieves state-of-the-art performance in three domain adaptation scenarios, especially the result of Sim10k to Cityscapes scenario is remarkably improved from 52.6 mAP to 57.9 mAP. Code will be released.



Paperid:374
Authors:David Ferman; Gaurav Bharaj
Title: Multi-Domain Multi-Definition Landmark Localization for Small Datasets
Abstract:
We present a novel method for multi image domain and multi-landmark definition learning for small dataset facial localization. Training a small dataset alongside a large(r) dataset helps with robust learning for the former, and provides a universal mechanism for facial landmark localization for new and/or smaller standard datasets. To this end, we propose a Vision Transformer encoder with a novel decoder with a definition agnostic shared landmark semantic group structured prior, that is learnt, as we train on more than one dataset concurrently. Due to our novel definition agnostic group prior the datasets may vary in landmark definitions and domains. During the decoder stage we use cross- and self-attention, whose output is later fed into domain/definition specific heads that minimize a Laplacian-log-likelihood loss. We achieve state-of-the-art performance on standard landmark localization datasets such as COFW and WFLW, when trained with a bigger dataset. We also show state-of-the-art performance on several varied image domain small datasets for animals, caricatures, and facial portrait paintings. Further, we contribute a small dataset (150 images) of pareidolias to show efficacy of our method. Finally, we provide several analysis and ablation studies to justify our claims.



Paperid:375
Authors:Abhinav Kumar; Garrick Brazil; Enrique Corona; Armin Parchami; Xiaoming Liu
Title: DEVIANT: Depth EquiVarIAnt NeTwork for Monocular 3D Object Detection
Abstract:
Modern neural networks use building blocks such as convolutions that are equivariant to arbitrary 2D translations. However, these vanilla blocks are not equivariant to arbitrary 3D translations in the projective manifold. Even then, all monocular 3D detectors use vanilla blocks to obtain the 3D coordinates, a task for which the vanilla blocks are not designed for. This paper takes the first step towards convolutions equivariant to arbitrary 3D translations in the projective manifold. Since the depth is the hardest to estimate for monocular detection, this paper proposes Depth EquiVarIAnt NeTwork (DEVIANT) built with existing scale equivariant steerable blocks. As a result, DEVIANT is equivariant to the depth translations in the projective manifold whereas vanilla networks are not. The additional depth equivariance forces the DEVIANT to learn consistent depth estimates, and therefore, DEVIANT achieves state-of-the-art monocular 3D detection results on KITTI and Waymo datasets in the image-only category and performs competitively to methods using extra information. Moreover, DEVIANT works better than vanilla networks in cross-dataset evaluation.



Paperid:376
Authors:Yaomin Huang; Xinmei Liu; Yichen Zhu; Zhiyuan Xu; Chaomin Shen; Zhengping Che; Guixu Zhang; Yaxin Peng; Feifei Feng; Jian Tang
Title: Label-Guided Auxiliary Training Improves 3D Object Detector
Abstract:
Detecting 3D objects from point clouds is a practical yet challenging task that has attracted increasing attention recently. In this paper, we propose a Label-Guided auxiliary training method for 3D object detection (LG3D), which serves as an auxiliary network to enhance the feature learning of existing 3D object detectors. Specifically, we propose two novel modules: a Label-Annotation-Inducer that maps annotations and point clouds in bounding boxes to task-specific representations and a Label-Knowledge-Mapper that assists the original features to obtain detection-critical representations. The proposed auxiliary network is discarded in inference and thus has no extra computational cost at test time. We conduct extensive experiments on both indoor and outdoor datasets to verify the effectiveness of our approach. For example, our proposed LG3D improves VoteNet by 2.5\% and 3.1\% mAP on the SUN RGB-D and ScanNetV2 datasets, respectively. The code is available at https://github.com/FabienCode/LG3D.



Paperid:377
Authors:Chengjian Feng; Yujie Zhong; Zequn Jie; Xiangxiang Chu; Haibing Ren; Xiaolin Wei; Weidi Xie; Lin Ma
Title: PromptDet: Towards Open-Vocabulary Detection Using Uncurated Images
Abstract:
The goal of this work is to establish a scalable pipeline for expanding an object detector towards novel/unseen categories, using zero manual annotations. To achieve that, we make the following four contributions: (i) in pursuit of generalisation, we propose a two-stage open-vocabulary object detector, where the class-agnostic object proposals are classified with a text encoder from pre-trained visual-language model; (ii) To pair the visual latent space (of RPN box proposals) with that of the pre-trained text encoder, we propose the idea of regional prompt learning to align the textual embedding space with regional visual object features; (iii) To scale up the learning procedure towards detecting a wider spectrum of objects, we exploit the available online resource via a novel self-training framework, which allows to train the proposed detector on a large corpus of noisy uncurated web images. Lastly, (iv) to evaluate our proposed detector, termed as PromptDet, we conduct extensive experiments on the challenging LVIS and MS-COCO dataset. PromptDet shows superior performance over existing approaches with fewer additional training images and zero manual annotations whatsoever. Project page with code: https://fcjian.github.io/promptdet.



Paperid:378
Authors:Yingyan Li; Yuntao Chen; Jiawei He; Zhaoxiang Zhang
Title: Densely Constrained Depth Estimator for Monocular 3D Object Detection
Abstract:
Estimating accurate 3D locations of objects from monocular images is a challenging problem because of lacking depth. Previous work shows that utilizing the object’s keypoint projection constraints to estimate multiple depth candidates boosts the detection performance. However, the existing methods can only utilize vertical edges as projection constraints for depth estimation. So these methods only use a small number of projection constraints and produce insufficient depth candidates, leading to inaccurate depth estimation. In this paper, we propose a method that utilizes dense projection constraints from edges of any direction. In this way, we employ much more projection constraints and produce considerable depth candidates. Besides, we present a graph matching weighting module to merge the depth candidates. The proposed method DCD (Densely Constrained Detector) achieves state-of-the-art performance on the KITTI and WOD benchmarks. Code is released at https://github.com/BraveGroup/DCD.



Paperid:379
Authors:Daoyi Gao; Yitong Li; Patrick Ruhkamp; Iuliia Skobleva; Magdalena Wysocki; HyunJun Jung; Pengyuan Wang; Arturo Guridi; Benjamin Busam
Title: Polarimetric Pose Prediction
Abstract:
Light has many properties that vision sensors can passively measure. Colour-band separated wavelength and intensity are arguably the most commonly used for monocular 6D object pose estimation. This paper explores how complementary polarisation information, i.e. the orientation of light wave oscillations, influences the accuracy of pose predictions. A hybrid model that leverages physical priors jointly with a data-driven learning strategy is designed and carefully tested on objects with different levels of photometric complexity. Our design significantly improves the pose accuracy compared to state-of-the-art photometric approaches and enables object pose estimation for highly reflective and transparent objects. A new multi-modal instance-level 6D object pose dataset with highly accurate pose annotations for multiple objects with varying photometric complexity is introduced as a benchmark.



Paperid:380
Authors:Shuai Chen; Xinghui Li; Zirui Wang; Victor Adrian Prisacariu
Title: DFNet: Enhance Absolute Pose Regression with Direct Feature Matching
Abstract:
We introduce a camera relocalization pipeline that combines absolute pose regression (APR) and direct feature matching. By incorporating exposure-adaptive novel view synthesis, our method successfully addresses photometric distortions in outdoor environments that existing photometric-based methods fail to handle. With domain-invariant feature matching, our solution improves pose regression accuracy using semi-supervised learning on unlabeled data. In particular, the pipeline consists of two components: Novel View Synthesizer and DFNet. The former synthesizes novel views compensating for changes in exposure and the latter regresses camera poses and extracts robust features that close the domain gap between real images and synthetic ones. Furthermore, we introduce an online synthetic data generation scheme. We show that these approaches effectively enhance camera pose estimation both in indoor and outdoor scenes. Hence, our method achieves a state-of-the-art accuracy by outperforming existing single-image APR methods by as much as 56%, comparable to 3D structure-based methods.



Paperid:381
Authors:Haoran Wei; Xin Chen; Lingxi Xie; Qi Tian
Title: Cornerformer: Purifying Instances for Corner-Based Detectors
Abstract:
Corner-based object detectors enjoy the potential of detecting arbitrarily-sized instances, yet the performance is mainly harmed by the accuracy of instance construction. Specifically, there are three factors, namely, 1) the corner keypoints are prone to false-positives; 2) incorrect matches emerge upon corner keypoint pull-push embeddings; and 3) the heuristic NMS cannot adjust the corners pull-push mechanism. Accordingly, this paper presents an elegant framework named Cornerformer that is composed of two factors. First, we build a Corner Transformer Encoder (CTE, a self-attention module) in a 2D-form to enhance the information aggregated by corner keypoints, offering stronger features for the pull-push loss to distinguish instances from each other. Second, we design an Attenuation-Auto-Adjusted NMS (A3-NMS) to maximally leverage the semantic outputs and avoid true objects from being removed. Experiments on object detection and human pose estimation show the superior performance of Cornerformer in terms of accuracy and inference speed.



Paperid:382
Authors:Guangsheng Shi; Ruifeng Li; Chao Ma
Title: PillarNet: Real-Time and High-Performance Pillar-Based 3D Object Detection
Abstract:
Real-time and high-performance 3D object detection is of critical importance for autonomous driving. Recent top-performing 3D object detectors mainly rely on point-based or 3D voxel-based convolutions, which are both computationally inefficient for onboard deployment. In contrast, pillar-based methods use solely 2D convolutions, which consume less computation resources, but they lag far behind their voxel-based counterparts in detection accuracy. In this paper, by examining the primary performance gap between pillar- and voxel-based detectors, we develop a real-time and high-performance pillar-based detector, dubbed PillarNet. The proposed PillarNet consists of a powerful encoder network for effective pillar feature learning, a neck network for spatial-semantic feature fusion and the commonly used detect head. Using only 2D convolutions, PillarNet is flexible to an optional pillar size and compatible with classical 2D CNN backbones, such as VGGNet and ResNet. Additionally, PillarNet benefits from our designed orientation-decoupled IoU regression loss along with the IoU-aware prediction branch. Extensive experimental results on large-scale nuScenes Dataset and Waymo Open Dataset demonstrate that the proposed PillarNet performs well over the state-of-the-art 3D detectors in terms of effectiveness and efficiency.



Paperid:383
Authors:Chengxin Liu; Kewei Wang; Hao Lu; Zhiguo Cao; Ziming Zhang
Title: Robust Object Detection with Inaccurate Bounding Boxes
Abstract:
Learning accurate object detectors often requires large-scale training data with precise object bounding boxes. However, labeling such data is expensive and time-consuming. As the crowd-sourcing labeling process and the ambiguities of the objects may raise noisy bounding box annotations, the object detectors will suffer from the degenerated training data. In this work, we aim to address the challenge of learning robust object detectors with inaccurate bounding boxes. Inspired by the fact that localization precision suffers significantly from inaccurate bounding boxes while classification accuracy is less affected, we propose leveraging classification as a guidance signal for refining localization results. Specifically, by treating an object as a bag of instances, we introduce an Object-Aware Multiple Instance Learning approach (OA-MIL), featured with object-aware instance selection and object-aware instance extension. The former aims to select accurate instances for training, instead of directly using inaccurate box annotations. The latter focuses on generating high-quality instances for selection. Extensive experiments on synthetic noisy datasets (i.e., noisy PASCAL VOC and MS-COCO) and a real noisy wheat head dataset demonstrate the effectiveness of our OA-MIL. Code is available at https://github.com/cxliu0/OA-MIL.



Paperid:384
Authors:Peixian Chen; Mengdan Zhang; Yunhang Shen; Kekai Sheng; Yuting Gao; Xing Sun; Ke Li; Chunhua Shen
Title: Efficient Decoder-Free Object Detection with Transformers
Abstract:
Vision transformers (ViTs) are changing the landscape of object detection tasks. A natural usage of ViTs in detection is to replace the CNN-based backbone with a transformer-based backbone, which is simple yet brings an enormous computation burden during inference. More subtle usage is the DETR family, which eliminates the need for many hand-designed components in object detection but introduces a decoder demanding an extra-long time to converge. As a result, transformer-based object detection could not prevail in large-scale applications. To overcome these issues, we propose a novel decoder-free fully transformer-based (DFFT) object detector, achieving high efficiency in both training and inference stages for the first time. We simplify objection detection to an encoder-only single-level anchor-based dense prediction problem by centering around two entry points: 1) Eliminate the training-inefficient decoder and leverage two strong encoders to preserve the accuracy of single-level feature map prediction; 2) Explore low-level semantic features for the detection task with limited computational resources. In particular, we design a novel lightweight detection-oriented transformer backbone that efficiently captures low-level features with rich semantics based on a well-conceived ablation study. Extensive experiments on the MS COCO benchmark demonstrate that DFFT{SMALL} outperforms DETR by 2.5% AP with 28% computation cost reduction and more than 10X fewer training epochs. Compared with the cutting-edge anchor-based detector RetinaNet, DFFT{SMALL} obtains over 5.5% AP gain while cutting down 70% computation cost.



Paperid:385
Authors:Yu Hong; Hang Dai; Yong Ding
Title: Cross-Modality Knowledge Distillation Network for Monocular 3D Object Detection
Abstract:
Leveraging LiDAR-based detectors or real LiDAR point data to guide monocular 3D detection has brought significant improvement, e.g., Pseudo-LiDAR methods. However, the existing methods usually apply non-end-to-end training strategies and insufficiently leverage the LiDAR information, where the rich potential of the LiDAR data has not been well exploited. In this paper, we propose the Cross-Modality Knowledge Distillation (CMKD) network for monocular 3D detection to efficiently and directly transfer the knowledge from LiDAR modality to image modality on both features and responses. Moreover, we further extend CMKD as a semi-supervised training framework by distilling knowledge from large-scale unlabeled data and significantly boost the performance. Until submission, CMKD ranks 1st among the monocular 3D detectors with publications on both KITTI test set and Waymo val set with significant performance gains compared to previous state-of-the-art methods. Our code will be released at https://github.com/Cc-Hy/CMKD.



Paperid:386
Authors:Dingfeng Shi; Yujie Zhong; Qiong Cao; Jing Zhang; Lin Ma; Jia Li; Dacheng Tao
Title: ReAct: Temporal Action Detection with Relational Queries
Abstract:
This work aims at advancing temporal action detection (TAD) using an encoder-decoder framework with action queries, similar to DETR, which has shown great success in object detection. However, the framework suffers from several problems if directly applied to TAD: the insufficient exploration of inter-query relation in the decoder, the inadequate classification training due to a limited number of training samples, and the unreliable classification scores at inference. To this end, we first propose a relational attention mechanism in the decoder, which guides the attention among queries based on their relations. Moreover, we propose two losses to facilitate and stabilize the training of action classification. Lastly, we propose to predict the localization quality of each action query at inference in order to distinguish high-quality queries. The proposed method, named ReAct, achieves the state-of-the-art performance on THUMOS14, with much lower computational costs than previous methods. Besides, extensive ablation studies are conducted to verify the effectiveness of each proposed component.



Paperid:387
Authors:Qihang Fang; Yingda Yin; Qingnan Fan; Fei Xia; Siyan Dong; Sheng Wang; Jue Wang; Leonidas J. Guibas; Baoquan Chen
Title: Towards Accurate Active Camera Localization
Abstract:
In this work, we tackle the problem of active camera localization, which controls the camera movements actively to achieve an accurate camera pose. The past solutions are mostly based on Markov Localization, which reduces the position-wise camera uncertainty for localization. These approaches localize the camera in the discrete pose space and are agnostic to the localization-driven scene property, which restricts the camera pose accuracy in the coarse scale. We propose to overcome these limitations via a novel active camera localization algorithm, composed of a passive and an active localization module. The former optimizes the camera pose in the continuous pose space by establishing point-wise camera-world correspondences. The latter explicitly models the scene and camera uncertainty components to plan the right path for accurate camera pose estimation. We validate our algorithm on the challenging localization scenarios from both synthetic and scanned real-world indoor scenes. Experimental results demonstrate that our algorithm outperforms both the state-of-the-art Markov Localization based approach and other compared approaches on the fine-scale camera pose accuracy. Code and data are released at https://github.com/qhFang/AccurateACL.



Paperid:388
Authors:Yoli Shavit; Yosi Keller
Title: Camera Pose Auto-Encoders for Improving Pose Regression
Abstract:
Absolute pose regressor (APR) networks are trained to estimate the pose of the camera given a captured image. They compute latent image representations from which the camera position and orientation are regressed. APRs provide a different tradeoff between localization accuracy, runtime, and memory, compared to structure-based localization schemes that provide state-of-the-art accuracy. In this work, we introduce Camera Pose Auto-Encoders (PAEs), multilayer perceptrons that are trained via a Teacher-Student approach to encode camera poses using APRs as their teachers. We show that the resulting latent pose representations can closely reproduce APR performance and demonstrate their effectiveness for related tasks. Specifically, we propose a light-weight test-time optimization in which the closest train poses are encoded and used to refine camera position estimation. This procedure achieves a new state-of-the-art position accuracy for APRs, on both the CambridgeLandmarks and 7Scenes benchmarks. We also show that train images can be reconstructed from the learned pose encoding, paving the way for integrating visual information from the train set at a low memory cost. Our code and pre-trained models are available at https://github.com/yolish/camera-pose-auto-encoders.



Paperid:389
Authors:Chiyu Max Jiang; Mahyar Najibi; Charles R. Qi; Yin Zhou; Dragomir Anguelov
Title: Improving the Intra-Class Long-Tail in 3D Detection via Rare Example Mining
Abstract:
Continued improvements in deep learning architectures have steadily advanced the overall performance of 3D object detectors to levels on par with humans for certain tasks and datasets, where the overall performance is mostly driven by common examples. However, even the best performing models suffer from the most naive mistakes when it comes to rare examples that do not appear frequently in the training data, such as vehicles with irregular geometries. Most studies in the long-tail literature focus on class-imbalanced classification problems with known imbalanced label counts per class, but they are not directly applicable to the intra-class long-tail examples in problems with large intra-class variations such as 3D object detection, where instances with the same class label can have drastically varied properties such as shapes and sizes. Other works propose to mitigate this problem using active learning based on the criteria of uncertainty, difficulty, or diversity. In this study, we identify a new conceptual dimension - rareness - to mine new data for improving the long-tail performance of models. We show that rareness, as opposed to difficulty, is the key to data-centric improvements for 3D detectors, since rareness is the result of a lack in data support while difficulty is related to the fundamental ambiguity in the problem. We propose a general and effective method to identify the rareness of objects based on density estimation in the feature space using flow models, and propose a principled cost-aware formulation for mining rare object tracks, which improves overall model performance, but more importantly - significantly improves the performance for rare objects (by 30.97%).



Paperid:390
Authors:Lei Zhu; Qian Chen; Lujia Jin; Yunfei You; Yanye Lu
Title: Bagging Regional Classification Activation Maps for Weakly Supervised Object Localization
Abstract:
Classification activation map (CAM), utilizing the classification structure to generate pixel-wise localization maps, is a crucial mechanism for weakly supervised object localization (WSOL). However, CAM directly uses the classifier trained on image-level features to locate objects, making it prefers to discern global discriminative factors rather than regional object cues. Thus only the discriminative locations are activated when feeding pixel-level features into this classifier. To solve this issue, this paper elaborates a plug-and-play mechanism called BagCAMs to better project a well-trained classifier for the localization task without refining or re-training the baseline structure. Our BagCAMs adopts a proposed regional localizer generation (RLG) strategy to define a set of regional localizers and then derive them from a well-trained classifier. These regional localizers can be viewed as the base learner that only discerns region-wise object factors for localization tasks, and their results can be effectively weighted by our BagCAMs to form the final localization map. Experiments indicate that adopting our proposed BagCAMs can improve the performance of baseline WSOL methods to a great extent and obtains state-of-the-art performance on three WSOL benchmarks. Code are released at https://github.com/zh460045050/BagCAMs.



Paperid:391
Authors:Zhiheng Wu; Yue Lu; Xingyu Chen; Zhengxing Wu; Liwen Kang; Junzhi Yu
Title: UC-OWOD: Unknown-Classified Open World Object Detection
Abstract:
Open World Object Detection (OWOD) is a challenging computer vision problem that requires detecting unknown objects and gradually learning the identified unknown classes. However, it cannot distinguish unknown instances as multiple unknown classes. In this work, we propose a novel OWOD problem called Unknown-Classified Open World Object Detection (UC-OWOD). UC-OWOD aims to detect unknown instances and classify them into different unknown classes. Besides, we formulate the problem and devise a two-stage object detector to solve UC-OWOD. First, unknown label-aware proposal and unknown-discriminative classification head are used to detect known and unknown objects. Then, similarity-based unknown classification and unknown clustering refinement modules are constructed to distinguish multiple unknown classes. Moreover, two novel evaluation protocols are designed to evaluate unknown-class detection. Abundant experiments and visualizations prove the effectiveness of the proposed method. Code is available at https://github.com/JohnWuzh/UC-OWOD.



Paperid:392
Authors:Michał J. Tyszkiewicz; Kevis-Kokitsi Maninis; Stefan Popov; Vittorio Ferrari
Title: RayTran: 3D Pose Estimation and Shape Reconstruction of Multiple Objects from Videos with Ray-Traced Transformers
Abstract:
We propose a transformer-based neural network architecture for multi-object 3D reconstruction from RGB videos. It relies on two alternative ways to represent its knowledge: as a global 3D grid of features and an array of view-specific 2D grids. We progressively exchange information between the two with a dedicated bidirectional attention mechanism. We exploit knowledge about the image formation process to significantly sparsify the attention weight matrix, making our architecture feasible on current hardware, both in terms of memory and computation. We attach a DETR-style head on top of the 3D feature grid in order to detect the objects in the scene and to predict their 3D pose and 3D shape. Compared to previous methods, our architecture is single stage, end-to-end trainable, and it can reason holistically about a scene from multiple video frames without needing a brittle tracking step. We evaluate our method on the challenging Scan2CAD dataset, where we outperform (1) recent state-of-the-art methods for 3D object pose estimation from RGB videos; and (2) a strong alternative method combining Multi-view Stereo with RGB-D CAD alignment. We plan to release our source code.



Paperid:393
Authors:Xinyi Li; Haibin Ling
Title: GTCaR: Graph Transformer for Camera Re-Localization
Abstract:
Camera re-localization or absolute pose regression is the centerpiece in numerous computer vision tasks such as visual odometry, structure from motion (SfM) and SLAM. In this paper we propose a neural network approach with a graph Transformer backbone, namely GTCaR (Graph Transformer for Camera Re-localization), to address the multi-view camera re-localization problem. In contrast with prior work where the pose regression is mainly guided by photometric consistency, GTCaR effectively fuses the image features, camera pose information and inter-frame relative camera motions into encoded graph attributes and is trained towards the graph consistency and pose accuracy combined instead, yielding significantly higher computational efficiency. By leveraging graph Transformer layers with edge features and enabling the adjacency tensor, GTCaR dynamically captures the global attention and thus endows the pose graph with evolving structures to achieve improved robustness and accuracy. In addition, optional temporal Transformer layers actively enhance the spatiotemporal inter-frame relation for sequential inputs. Evaluation of the proposed network on various public benchmarks demonstrates that GTCaR outperforms state-of-the-art approaches.



Paperid:394
Authors:Emeç Erçelik; Ekim Yurtsever; Mingyu Liu; Zhijie Yang; Hanzhen Zhang; Pınar Topçam; Maximilian Listl; Yılmaz Kaan Çaylı; Alois Knoll
Title: 3D Object Detection with a Self-Supervised Lidar Scene Flow Backbone
Abstract:
State-of-the-art lidar-based 3D object detection methods rely on supervised learning and large labeled datasets. However, annotating lidar data is resource-consuming, and depending only on supervised learning limits the applicability of trained models. Self-supervised training strategies can alleviate these issues by learning a general point cloud backbone model for downstream 3D vision tasks. Against this backdrop, we show the relationship between self-supervised multi-frame flow representations and single-frame 3D detection hypotheses. Our main contribution leverages learned flow and motion representations and combines a self-supervised backbone with a supervised 3D detection head. First, a self-supervised scene flow estimation model is trained with cycle consistency. Then, the point cloud encoder of this model is used as the backbone of a single-frame 3D object detection head model. This second 3D object detection model learns to utilize motion representations to distinguish dynamic objects exhibiting different movement patterns. Experiments on KITTI and nuScenes benchmarks show that the proposed self-supervised pre-training increases 3D detection performance significantly. https://github.com/emecercelik/ssl-3d-detection.git"



Paperid:395
Authors:Mingfei Gao; Chen Xing; Juan Carlos Niebles; Junnan Li; Ran Xu; Wenhao Liu; Caiming Xiong
Title: Open Vocabulary Object Detection with Pseudo Bounding-Box Labels
Abstract:
Despite great progress in object detection, most existing methods work only on a limited set of object categories, due to the tremendous human effort needed for bounding-box annotations of training data. To alleviate the problem, recent open vocabulary and zero-shot detection methods attempt to detect novel object categories beyond those seen during training. They achieve this goal by training on a pre-defined base categories to induce generalization to novel objects. However, their potential is still constrained by the small set of base categories available for training. To enlarge the set of base classes, we propose a method to automatically generate pseudo bounding-box annotations of diverse objects from large-scale image-caption pairs. Our method leverages the localization ability of pre-trained vision-language models to generate pseudo bounding-box labels and then directly uses them for training object detectors. Experimental results show that our method outperforms the state-of-the-art open vocabulary detector by 8% AP on COCO novel categories, by 6.3% AP on PASCAL VOC, by 2.3% AP on Objects365 and by 2.8% AP on LVIS. Code is available at https://github.com/salesforce/PB-OVD.



Paperid:396
Authors:Wenjie Pei; Shuang Wu; Dianwen Mei; Fanglin Chen; Jiandong Tian; Guangming Lu
Title: Few-Shot Object Detection by Knowledge Distillation Using Bag-of-Visual-Words Representations
Abstract:
While fine-tuning based methods for few-shot object detection have achieved remarkable progress, a crucial challenge that has not been addressed well is the potential class-specific overfitting on base classes and sample-specific overfitting on novel classes. In this work we design a novel knowledge distillation framework to guide the learning of the object detector and thereby restrain the overfitting in both the pre-training stage on base classes and fine-tuning stage on novel classes. To be specific, we first present a novel Position-Aware Bag-of-Visual-Words model for learning a representative bag of visual words (BoVW) from a limited size of image set, which is used to encode general images based on the similarities between the learned visual words and an image. Then we perform knowledge distillation based on the fact that an image should have consistent BoVW representations in two different feature spaces. To this end, we pre-learn a feature space independently from the object detection, and encode images using BoVW in this space. The obtained BoVW representation for an image can be considered as distilled knowledge to guide the learning of object detector: the extracted features by the object detector for the same image are expected to derive the consistent BoVW representations with the distilled knowledge. Extensive experiments validate the effectiveness of our method and demonstrate the superiority over other state-of-the-art methods.



Paperid:397
Authors:Babak Ehteshami Bejnordi; Amirhossein Habibian; Fatih Porikli; Amir Ghodrati
Title: SALISA: Saliency-Based Input Sampling for Efficient Video Object Detection
Abstract:
High-resolution images are widely adopted for high-performance object detection in videos. However, processing high-resolution inputs comes with high computation costs, and naive down-sampling of the input to reduce the computation costs quickly degrades the detection performance. In this paper, we propose SALISA, a novel non-uniform SALiency-based Input SAmpling technique for video object detection that allows for heavy down-sampling of unimportant background regions while preserving the fine-grained details of a high-resolution image. The resulting image is spatially smaller, leading to reduced computational costs while enabling a performance comparable to a high-resolution input. To achieve this, we propose a differentiable resampling module based on a thin plate spline spatial transformer network (TPS-STN). This module is regularized by a novel loss to provide an explicit supervision signal to learn to “magnify” salient regions. We report state-of-the-art results in the low compute regime on the ImageNet-VID and UA-DETRAC video object detection datasets. We demonstrate that on both datasets, the mAP of an EfficientDet-D1 (EfficientDet-D2) gets on par with EfficientDet-D2 (EfficientDet-D3) at a much lower computational cost. We also show that SALISA significantly improves the detection of small objects. In particular, SALISA with an EfficientDet-D1 detector improves the detection of small objects by 77%, and remarkably also outperforms EfficientDet-D3 baseline.



Paperid:398
Authors:Dongli Tan; Jiang-Jiang Liu; Xingyu Chen; Chao Chen; Ruixin Zhang; Yunhang Shen; Shouhong Ding; Rongrong Ji
Title: ECO-TR: Efficient Correspondences Finding via Coarse-to-Fine Refinement
Abstract:
Abstract. Modeling sparse and dense image matching within a unified functional model has recently attracted increasing research interest. However, existing efforts mainly focus on improving matching accuracy while ignoring its efficiency, which is crucial for real-world applications. In this paper, we propose an efficient structure named Efficient Correspondence Transformer (ECO-TR) by finding correspondences in a coarse-to-fine manner, which significantly improves the efficiency of functional model. To achieve this, multiple transformer blocks are stage-wisely connected to gradually refine the predicted coordinates upon a shared multi-scale feature extraction network. Given a pair of images and for arbitrary query coordinates, all the correspondences are predicted within a single feed-forward pass. We further propose an adaptive query-clustering strategy and an uncertainty-based outlier detection module to cooperate with the proposed framework for faster and better predictions. Experiments on various sparse and dense matching tasks demonstrate the superiority of our method in both efficiency and effectiveness against existing state-of-the-arts. Project page: https://dltan7.github.io/ecotr/.



Paperid:399
Authors:Yangzheng Wu; Mohsen Zand; Ali Etemad; Michael Greenspan
Title: Vote from the Center: 6 DoF Pose Estimation in RGB-D Images by Radial Keypoint Voting
Abstract:
We propose a novel keypoint voting scheme based on intersecting spheres, that is more accurate than existing schemes and allows for fewer, more disperse keypoints. The scheme is based upon the distance between points, which as a 1D quantity can be regressed more accurately than the 2D and 3D vector and offset quantities regressed in previous work, yielding more accurate keypoint localization. The scheme forms the basis of the proposed RCVPose method for 6 DoF pose estimation of 3D objects in RGB-D data, which is particularly effective at handling occlusions. A CNN is trained to estimate the distance between the 3D point corresponding to the depth mode of each RGB pixel, and a set of 3 disperse keypoints defined in the object frame. At inference, a sphere centered at each 3D point is generated, of radius equal to this estimated distance. The surfaces of these spheres vote to increment a 3D accumulator space, the peaks of which indicate keypoint locations. The proposed radial voting scheme is more accurate than previous vector or offset schemes, and is robust to disperse keypoints. Experiments demonstrate RCVPose to be highly accurate and competitive, achieving state-of-theart results on the LINEMOD (99.7%) and YCB-Video (97.2%) datasets, notably scoring +4.9% higher (71.1%) than previous methods on the challenging Occlusion LINEMOD dataset, and on average outperforming all other published results from the BOP benchmark for these 3 datasets. Our code is available at http://www.github.com/aaronwool/rcvpose.



Paperid:400
Authors:Konstantinos Panagiotis Alexandridis; Jiankang Deng; Anh Nguyen; Shan Luo
Title: Long-Tailed Instance Segmentation Using Gumbel Optimized Loss
Abstract:
Major advancements have been made in the field of object detection and segmentation recently. However, when it comes to rare categories, the state-of-the-art methods fail to detect them, resulting in a significant performance gap between rare and frequent categories. In this paper, we identify that Sigmoid or Softmax functions used in deep detectors are a major reason for low performance and are suboptimal for long-tailed detection and segmentation. To address this, we develop a Gumbel Optimized Loss (GOL), for long-tailed detection and segmentation. It aligns with the Gumbel distribution of rare classes in imbalanced datasets, considering the fact that most classes in long-tailed detection have low expected probability. The proposed GOL significantly outperforms the best state-of-the-art method by 1.1% on AP, and boosts the overall segmentation by 9.0% and detection by 8.0%, particularly improving detection of rare classes by 20.3%, compared to Mask-RCNN, on LVIS dataset. Code available at: https://github.com/kostas1515/ GOL.



Paperid:401
Authors:Jinhyung Park; Chenfeng Xu; Yiyang Zhou; Masayoshi Tomizuka; Wei Zhan
Title: DetMatch: Two Teachers Are Better than One for Joint 2D and 3D Semi-Supervised Object Detection
Abstract:
While numerous 3D detection works leverage the complementary relationship between RGB images and point clouds, developments in the broader framework of semi-supervised object recognition remain uninfluenced by multi-modal fusion. Current methods develop independent pipelines for 2D and 3D semi-supervised learning despite the availability of paired image and point cloud frames. Observing that the distinct characteristics of each sensor cause them to be biased towards detecting different objects, we propose DetMatch, a flexible framework for joint semi-supervised learning on 2D and 3D modalities. By identifying objects detected in both sensors, our pipeline generates a cleaner, more robust set of pseudo-labels that both demonstrates stronger performance and stymies single-modality error propagation. Further, we leverage the richer semantics of RGB images to rectify incorrect 3D class predictions and improve localization of 3D boxes. Evaluating our method on the challenging KITTI and Waymo datasets, we improve upon strong semi-supervised learning methods and observe higher quality pseudo-labels. Code will be released here: https://github.com/Divadi/DetMatch.



Paperid:402
Authors:Mohsen Zand; Ali Etemad; Michael Greenspan
Title: ObjectBox: From Centers to Boxes for Anchor-Free Object Detection
Abstract:
We present ObjectBox, a novel single-stage anchor-free and highly generalizable object detection approach. As opposed to both existing anchor-based and anchor-free detectors, which are more biased toward specific object scales in their label assignments, we use only object center locations as positive samples and treat all objects equally in different feature levels regardless of the objects’ sizes or shapes. Specifically, our label assignment strategy considers the object center locations as shape- and size-agnostic anchors in an anchor-free fashion, and allows learning to occur at all scales for every object. To support this, we define new regression targets as the distances from two corners of the center cell location to the four sides of the bounding box. Moreover, to handle scale-variant objects, we propose a tailored IoU loss to deal with boxes with different sizes. As a result, our proposed object detector does not need any dataset-dependent hyperparameters to be tuned across datasets. We evaluate our method on MS-COCO 2017 and PASCAL VOC 2012 datasets, and compare our results to state-of-the-art methods. We observe that ObjectBox performs favorably in comparison to prior works. Furthermore, we perform rigorous ablation experiments to evaluate different components of our method. Our code is available at: https://github.com/MohsenZand/ObjectBox.



Paperid:403
Authors:Qunjie Zhou; Sérgio Agostinho; Aljoša Ošep; Laura Leal-Taixé
Title: Is Geometry Enough for Matching in Visual Localization?
Abstract:
In this paper, we propose to go beyond the well-established approach to vision-based localization that relies on visual descriptor matching between a query image and a 3D point cloud. While matching keypoints via visual descriptors makes localization highly accurate, it has significant storage demands, raises privacy concerns and requires update to the descriptors in the long-term. To elegantly address those practical challenges for large-scale localization, we present GoMatch, an alternative to visual-based matching that solely relies on geometric information for matching image keypoints to maps, represented as sets of bearing vectors. Our novel bearing vectors representation of 3D points, significantly relieves the cross-modal challenge in geometric-based matching that prevented prior work to tackle localization in a realistic environment. With additional careful architecture design, GoMatch improves over prior geometric-based matching work with a reduction of (10.67m,95.7deg) and (1.43m, 34.7deg) in average median pose errors on Cambridge Landmarks and 7-Scenes, while requiring as little as 1.5/1.7% of storage capacity in comparison to the best visual-based matching methods. This confirms its potential and feasibility for real-world localization and opens the door to future efforts in advancing city-scale visual localization methods that do not require storing visual descriptors.



Paperid:404
Authors:Pei Sun; Mingxing Tan; Weiyue Wang; Chenxi Liu; Fei Xia; Zhaoqi Leng; Dragomir Anguelov
Title: SWFormer: Sparse Window Transformer for 3D Object Detection in Point Clouds
Abstract:
3D object detection in point clouds is a core component for modern robotics and autonomous driving systems. A key challenge in 3D object detection comes from the inherent sparse nature of point occupancy within the 3D scene. In this paper, we propose Sparse Window Transformer (SWFormer ), a scalable and accurate model for 3D object detection, which can take full advantage of the sparsity of point clouds. Built upon the idea of Swin Transformers, SWFormer operates on the sparse voxels and processes variable-length sparse windows efficiently using a bucketing scheme. In addition to self-attention within each spatial window, our SWFormer also captures cross-window correlation with multi-scale feature fusion and window shifting operations. To further address the unique challenge of detecting 3D objects accurately from sparse features, we propose a new voxel diffusion technique. Experimental results on the Waymo Open Dataset show our SWFormer achieves state-of-the-art 73.36 L2 mAPH on vehicle and pedestrian for 3D object detection on the official test set, outperforming all previous single-stage and two-stage models.



Paperid:405
Authors:Yu Zhang; Junle Yu; Xiaolin Huang; Wenhui Zhou; Ji Hou
Title: PCR-CG: Point Cloud Registration via Deep Explicit Color and Geometry
Abstract:
In this paper, we introduce PCR-CG: a novel 3D point cloud registration module explicitly embedding the color signals into geometry representation. Different from the previous methods that used only geometry representation, our module is specifically designed to effectively correlate color and geometry for the point cloud registration task. Our key contribution is a 2D-3D cross-modality learning algorithm that embeds the features learned from color signals to the geometry representation. With our designed 2D-3D projection module, the pixel features in a square region centered at correspondences perceived from images are effectively correlated with point cloud representations. In this way, the overlap regions can be inferred not only from point cloud but also from the texture appearances. Adding color is non-trivial. We compare against a variety of baselines designed for adding color to 3D, such as exhaustively adding per-pixel features or RGB values in an implicit manner. We leverage Predator as our baseline method and incorporate our module into it. Our experimental results indicate a significant improvement on the 3DLoMatch benchmark. With the help of our module, we achieve a significant improvement of 6.5% registration recall over our baseline method. To validate the effectiveness of 2D features on 3D, we ablate different 2D pre-trained networks and show a positive correlation between the pre-trained weights and task performance. Our study reveals a significant advantage of correlating explicit deep color features to the point cloud in the registration task.



Paperid:406
Authors:Younho Jang; Wheemyung Shin; Jinbeom Kim; Simon Woo; Sung-Ho Bae
Title: GLAMD: Global and Local Attention Mask Distillation for Object Detectors
Abstract:
Knowledge distillation (KD) is a well-known model compression strategy to improve models’ performance with fewer parameters. However, recent KD approaches for object detection have faced two limitations. First, they distill nearby foreground regions, ignoring potentially useful background information. Second, they only consider global contexts, thereby the student model can hardly learn local details from the teacher model. To overcome such challenging issues, we propose a novel knowledge distillation method, GLAMD, distilling both global and local knowledge from the teacher. We divide the feature maps into several patches and apply an attention mechanism for both the entire feature area and each patch to extract the global context as well as local details simultaneously. Our method outperforms the state-of-the-art methods with 40.8 AP on COCO2017 dataset, which is 3.4 AP higher than the student model (ResNet50 based Faster R-CNN) and 0.7 AP higher than the previous global attention-based distillation method.



Paperid:407
Authors:Danila Rukhovich; Anna Vorontsova; Anton Konushin
Title: FCAF3D: Fully Convolutional Anchor-Free 3D Object Detection
Abstract:
Recently, promising applications in robotics and augmented reality have attracted considerable attention to 3D object detection from point clouds. In this paper, we present FCAF3D -- a first-in-class fully convolutional anchor-free indoor 3D object detection method. It is a simple yet effective method that uses a voxel representation of a point cloud and processes voxels with sparse convolutions. FCAF3D can handle large-scale scenes with minimal runtime through a single fully convolutional feed-forward pass. Existing 3D object detection methods make prior assumptions on the geometry of objects, and we argue that it limits their generalization ability. To eliminate prior assumptions, we propose a novel parametrization of oriented bounding boxes that allows obtaining better results in a purely data-driven way. The proposed method achieves state-of-the-art 3D object detection results in terms of mAP@0.5 on ScanNet V2 (+4.5), SUN RGB-D (+3.5), and S3DIS (+20.5) datasets.The code and models are available at https://github.com/samsunglabs/fcaf3d"



Paperid:408
Authors:Guodong Wang; Yunhong Wang; Jie Qin; Dongming Zhang; Xiuguo Bao; Di Huang
Title: Video Anomaly Detection by Solving Decoupled Spatio-Temporal Jigsaw Puzzles
Abstract:
Video Anomaly Detection (VAD) is an important topic in computer vision. Motivated by the recent advances in self-supervised learning, this paper addresses VAD by solving an intuitive yet challenging pretext task, i.e., spatio-temporal jigsaw puzzles, which is cast as a multi-label fine-grained classification problem. Our method exhibits several advantages over existing works: 1) the spatio-temporal jigsaw puzzles are decoupled in terms of spatial and temporal dimensions, responsible for capturing highly discriminative appearance and motion features, respectively; 2) full permutations are used to provide abundant jigsaw puzzles covering various difficulty levels, allowing the network to distinguish subtle spatio-temporal differences between normal and abnormal events; and 3) the pretext task is tackled in an end-to-end manner without relying on any pre-trained models. Our method outperforms state-of-the-art counterparts on three public benchmarks. Especially on ShanghaiTech Campus, the result is superior to reconstruction and prediction-based methods by a large margin.



Paperid:409
Authors:Muhammad Maaz; Hanoona Rasheed; Salman Khan; Fahad Shahbaz Khan; Rao Muhammad Anwer; Ming-Hsuan Yang
Title: Class-Agnostic Object Detection with Multi-modal Transformer
Abstract:
What constitutes an object? This has been a long-standing question in computer vision. Towards this goal, numerous learning-free and learning-based approaches have been developed to score objectness. However, they generally do not scale well across new domains and for unseen objects. In this paper, we advocate that existing methods lack a top-down supervision signal governed by human-understandable semantics. For the first time in literature, we demonstrate that Multi-modal Vision Transformers (MViT) trained with aligned image-text pairs can effectively bridge this gap. Our extensive experiments across various domains and novel objects show the state-of-the-art performance of MViTs to localize generic objects in images. Based on the observation that existing MViTs do not include multi-scale feature processing and usually require longer training schedules, we develop an efficient and flexible MViT architecture using multi-scale and deformable self-attention. We show the significance of MViT proposals in a diverse range of applications including open-world object detection, salient and camouflage object detection, supervised and self-supervised detection tasks. Further, MViTs can adaptively generate proposals given a specific language query and thus offer enhanced interactability. Code: https://git.io/J1HPY.



Paperid:410
Authors:Hao Li; Zehan Zhang; Xian Zhao; Yulong Wang; Yuxi Shen; Shiliang Pu; Hui Mao
Title: Enhancing Multi-modal Features Using Local Self-Attention for 3D Object Detection
Abstract:
LiDAR and Camera sensors have complementary properties: LiDAR senses accurate positioning, while camera provides rich texture and color information. Fusing these two modalities can intuitively improve the performance of 3D detection. Most multi-modal fusion methods use networks to extract features of LiDAR and camera modality respectively, then simply add or concancate them together. We argue that these two kinds of signals are completely different, so it is not proper to combine these two heterogeneous features directly. In this paper, we propose EMMF-Det to do multi-modal fusion leveraging range and camera images. EMMF-Det uses self-attention mechanism to do feature re-weighting on these two modalities interactively, which can enchance the features with color, texture and localiztion information provided by LiDAR and camera signals. On the Waymo Open Dataset, EMMF-Det acheives the state-of-the-art performance. Besides this, evaluation on self-built dataset further proves the effectiveness of our method.



Paperid:411
Authors:Georg Hess; Christoffer Petersson; Lennart Svensson
Title: Object Detection As Probabilistic Set Prediction
Abstract:
Accurate uncertainty estimates are essential for deploying deep object detectors in safety-critical systems. The development and evaluation of probabilistic object detectors have been hindered by shortcomings in existing performance measures, which tend to involve arbitrary thresholds or limit the detector’s choice of distributions. In this work, we propose to view object detection as a set prediction task where detectors predict the distribution over the set of objects. Using the negative log-likelihood for random finite sets, we present a proper scoring rule for evaluating and training probabilistic object detectors. The proposed method can be applied to existing probabilistic detectors, is free from thresholds, and enables fair comparison between architectures. Three different types of detectors are evaluated on the COCO dataset. Our results indicate that the training of existing detectors is optimized toward non-probabilistic metrics. We hope to encourage the development of new object detectors that can accurately estimate their own uncertainty.



Paperid:412
Authors:Zhi Li; Lu He; Huijuan Xu
Title: Weakly-Supervised Temporal Action Detection for Fine-Grained Videos with Hierarchical Atomic Actions
Abstract:
Action understanding has evolved into the era of fine granularity, as most human behaviors in real life have only minor differences. To detect these fine-grained actions accurately in a label-efficient way, we tackle the problem of weakly-supervised fine-grained temporal action detection in videos for the first time. Without the careful design to capture subtle differences between fine-grained actions, previous weakly-supervised models for general action detection cannot perform well in fine-grained settings. We propose to model actions as the combinations of reusable atomic actions which are automatically discovered from data through self-supervised clustering, in order to capture the commonality and individuality of fine-grained actions. The learnt atomic actions, represented by visual concepts, are further mapped to fine and coarse action labels leveraging the semantic label hierarchy. Our approach constructs a visual representation hierarchy of four levels: clip level, atomic action level, fine action class level and coarse action class level, with supervision at each level. Extensive experiments on two large-scale fine-grained video datasets, FineAction and FineGym, show the benefit of our proposed weakly-supervised model for fine-grained action detection, and it achieves state-of-the-art results.



Paperid:413
Authors:Lin Huang; Tomas Hodan; Lingni Ma; Linguang Zhang; Luan Tran; Christopher D. Twigg; Po-Chen Wu; Junsong Yuan; Cem Keskin; Robert Wang
Title: Neural Correspondence Field for Object Pose Estimation
Abstract:
We propose a method for estimating the 6DoF pose of a rigid object with an available 3D model from a single RGB image. Unlike classical correspondence-based methods which predict 3D object coordinates at pixels of the input image, the proposed method predicts 3D object coordinates at 3D query points sampled in the camera frustum. The move from pixels to 3D points, which is inspired by recent PIFu-style methods for 3D reconstruction, enables reasoning about the whole object, including its (self-)occluded parts. For a 3D query point associated with a pixel-aligned image feature, we train a fully-connected neural network to predict: (i) the corresponding 3D object coordinates, and (ii) the signed distance to the object surface, with the first defined only for query points in the surface vicinity. We call the mapping realized by this network as Neural Correspondence Field. The object pose is then robustly estimated from the predicted 3D-3D correspondences by the Kabsch-RANSAC algorithm. The proposed method achieves state-of-the-art results on three BOP datasets and is shown superior especially in challenging cases with occlusion. The project website is at: linhuang17.github.io/NCF.



Paperid:414
Authors:Elijah Cole; Kimberly Wilber; Grant Van Horn; Xuan Yang; Marco Fornoni; Pietro Perona; Serge Belongie; Andrew Howard; Oisin Mac Aodha
Title: On Label Granularity and Object Localization
Abstract:
Weakly supervised object localization (WSOL) aims to learn representations that encode object location using only image-level category labels. However, many objects can be labeled at different levels of granularity. Is it an animal, a bird, or a great horned owl? Which image-level labels should we use? In this paper we study the role of label granularity in WSOL. To facilitate this investigation we introduce iNatLoc500, a new large-scale fine-grained benchmark dataset for WSOL. Surprisingly, we find that choosing the right training label granularity provides a much larger performance boost than choosing the best WSOL algorithm. We also show that changing the label granularity can significantly improve data efficiency.



Paperid:415
Authors:Sanghoon Lee; Youngmin Oh; Donghyeon Baek; Junghyup Lee; Bumsub Ham
Title: OIMNet++: Prototypical Normalization and Localization-Aware Learning for Person Search
Abstract:
We address the task of person search, that is, localizing and re-identifying query persons from a set of raw scene images. Recent approaches are typically built upon OIMNet, a pioneer work on person search, that learns joint person representations for performing both detection and person re-identification (reID) tasks. To obtain the representations, they extract features from pedestrian proposals, and then project them on a unit hypersphere with L2 normalization. These methods also incorporate all positive proposals, that sufficiently overlap with the ground truth, equally to learn person representations for reID. We have found that 1) the L2 normalization without considering feature distributions degenerates the discriminative power of person representations, and 2) positive proposals often also depict background clutter and person overlaps, which could encode noisy features to person representations. In this paper, we introduce OIMNet++ that addresses the aforementioned limitations. To this end, we introduce a novel normalization layer, dubbed ProtoNorm, that calibrates features from pedestrian proposals, while considering a long-tail distribution of person IDs, enabling L2 normalized person representations to be discriminative. We also propose a localization-aware feature learning scheme that encourages better-aligned proposals to contribute more in learning discriminative representations. Experimental results and analysis on standard person search benchmarks demonstrate the effectiveness of OIMNet++.



Paperid:416
Authors:Ruoqi Li; Chongyang Zhang; Hao Zhou; Chao Shi; Yan Luo
Title: Out-of-Distribution Identification: Let Detector Tell Which I Am Not Sure
Abstract:
The superior performance of object detectors is often established under the condition that the test samples are in the same distribution as the training data. However, in most practical applications, out-of-distribution (OOD) instances are inevitable and usually lead to detection uncertainty. In this work, the Feature structured OOD-IDentification (FOOD-ID) model is proposed to reduce the uncertainty of detection results by identifying the OOD instances. Instead of outputting each detection result directly, FOOD-ID uses a likelihood-based measuring mechanism to identify whether the feature satisfies the corresponding class distribution and outputs the OOD results separately. Specifically, the clustering-oriented feature structuration is firstly developed using class-specified prototypes and Attractive-Repulsive loss for more discriminative feature representation and more compact distribution. With the structured features space, the density distribution of all training categories is estimated based on a class-conditional normalizing flow, which is then used for the OOD identification in the test stage. The proposed FOOD-ID can be easily applied to various object detectors including anchor-based frameworks and anchor-free frameworks. Extensive experiments on the PASCAL VOC dataset and an industrial defect dataset demonstrate that FOOD-ID achieves satisfactory OOD identification performance, with which the certainty of detection results is improved significantly.



Paperid:417
Authors:Cheng Zhang; Tai-Yu Pan; Tianle Chen; Jike Zhong; Wenjin Fu; Wei-Lun Chao
Title: Learning with Free Object Segments for Long-Tailed Instance Segmentation
Abstract:
One fundamental challenge in building an instance segmentation model for a large number of classes in complex scenes is the lack of training examples, especially for rare objects. In this paper, we explore the possibility to increase the training examples without laborious data collection and annotation. We find that an abundance of instance segments can potentially be obtained freely from object-centric images, according to two insights: (i) an object-centric image usually contains one salient object in a simple background; (ii) objects from the same class often share similar appearances or similar contrasts to the background. Motivated by these insights, we propose a simple and scalable framework FreeSeg for extracting and leveraging these ""free"" object foreground segments to facilitate model training in long-tailed instance segmentation. Concretely, we investigate the similarity among object-centric images of the same class to propose candidate segments of foreground instances, followed by a novel ranking of segment quality. The resulting high-quality object segments can then be used to augment the existing long-tailed datasets, e.g., by copying and pasting the segments onto the original training images. Extensive experiments show that FreeSeg yields substantial improvements on top of strong baselines and achieves state-of-the-art accuracy for segmenting rare object categories. Our code is publicly available at https://github.com/czhang0528/FreeSeg.



Paperid:418
Authors:YuXuan Liu; Nikhil Mishra; Maximilian Sieb; Yide Shentu; Pieter Abbeel; Xi Chen
Title: Autoregressive Uncertainty Modeling for 3D Bounding Box Prediction
Abstract:
3D bounding boxes are a widespread intermediate representation in many computer vision applications. However, predicting them is a challenging task, largely due to partial observability, which motivates the need for a strong sense of uncertainty. While many recent methods have explored better architectures for consuming sparse and unstructured point cloud data, we hypothesize that there is room for improvement in the modeling of the output distribution and explore how this can be achieved using an autoregressive prediction head. Additionally, we release a simulated dataset, COB-3D, which highlights new types of ambiguity that arise in real-world robotics applications, where 3D bounding box prediction has largely been underexplored. We propose methods for leveraging our autoregressive model to make high confidence predictions and meaningful uncertainty measures, achieving strong results on SUN-RGBD, Scannet, KITTI, and our new dataset. Code and dataset are available at https://bbox.yuxuanliu.com.



Paperid:419
Authors:Rui Qiu; Ming Xu; Yuyao Yan; Jeremy S. Smith; Xi Yang
Title: 3D Random Occlusion and Multi-layer Projection for Deep Multi-Camera Pedestrian Localization
Abstract:
Although deep-learning based methods for monocular pedestrian detection have made a great progress, they are still vulnerable to heavy occlusions. Using multi-view information fusion is a potential solution but has limited applications, due to the lack of annotated training samples in existing multi-view datasets, which increases the risk of overfitting. To address this problem, a data augmentation method is proposed to randomly generate 3D cylinder occlusions, on the ground plane, which are of the average size of pedestrians and projected to multiple views, to relieve the impact of overfitting in the training. Moreover, the feature map of each view is projected to multiple parallel planes at different heights, by using homographies, which allows the CNNs to fully utilize the features across the height of each pedestrian to infer the locations of pedestrians on the ground plane. The proposed 3DROM method has a greatly improved performance in comparison with the state-of-the-art deep-learning based methods for multi-view pedestrian detection.



Paperid:420
Authors:Wuyang Chen; Xianzhi Du; Fan Yang; Lucas Beyer; Xiaohua Zhai; Tsung-Yi Lin; Huizhong Chen; Jing Li; Xiaodan Song; Zhangyang Wang; Denny Zhou
Title: A Simple Single-Scale Vision Transformer for Object Detection and Instance Segmentation
Abstract:
This work presents a simple vision transformer design as a strong baseline for object localization and instance segmentation tasks. Transformers recently demonstrate competitive performance in image classification tasks. To adopt ViT to object detection and dense prediction tasks, many works inherit the multistage design from convolutional networks and highly customized ViT architectures. Behind this design, the goal is to pursue a better trade-off between computational cost and effective aggregation of multiscale global contexts. However, existing works adopt multistage architectural design as a black-box solution without a clear understanding of its true benefits. In this paper, we comprehensively study three architecture design choices on ViT -- spatial reduction, doubled channels, and multiscale features -- and demonstrate that a vanilla ViT architecture can fulfill this goal without handcrafting multiscale features, maintaining the original ViT design philosophy. We further complete a scaling rule to optimize our model’s trade-off on accuracy and computation cost / model size. By leveraging a constant feature resolution and hidden size throughout the encoder blocks, we propose a simple and compact ViT architecture called Universal Vision Transformer (UViT) that achieves strong performance on COCO object detection and instance segmentation benchmark. Our code will be released upon acceptance.



Paperid:421
Authors:Matthias Minderer; Alexey Gritsenko; Austin Stone; Maxim Neumann; Dirk Weissenborn; Alexey Dosovitskiy; Aravindh Mahendran; Anurag Arnab; Mostafa Dehghani; Zhuoran Shen; Xiao Wang; Xiaohua Zhai; Thomas Kipf; Neil Houlsby
Title: Simple Open-Vocabulary Object Detection with Vision Transformers
Abstract:
Combining simple architectures with large-scale pre-training has led to massive improvements in image classification. For object detection, pre-training and scaling approaches are less well established, especially in the long-tailed and open-vocabulary setting, where training data is relatively scarce. In this paper, we propose a strong recipe for transferring image-text models to open-vocabulary object detection. We use a standard Vision Transformer architecture with minimal modifications, contrastive image-text pre-training, and end-to-end detection fine-tuning. Our analysis of the scaling properties of this setup shows that increasing image-level pre-training and model size yield consistent improvements on the downstream detection task. We provide the adaptation strategies and regularizations needed to attain very strong performance on zero-shot text-conditioned and one-shot image-conditioned object detection. Code and models will be made available on GitHub.



Paperid:422
Authors:Yutong Lin; Chen Li; Yue Cao; Zheng Zhang; Jianfeng Wang; Lijuan Wang; Zicheng Liu; Han Hu
Title: "A Simple Approach and Benchmark for 21,000-Category Object Detection"
Abstract:
Current object detection systems and benchmarks typically handle a limited number of categories, up to about a thousand categories. This paper scales the number of categories for object detection systems and benchmarks up to 21,000, by leveraging existing object detection and image classification data. Unlike previous efforts that usually transfer knowledge from base detectors to image classification data, we propose to rely more on a reverse information flow from a base image classifier to object detection data. In this framework, the large-vocabulary classification capability is first learnt thoroughly using only the image classification data. In this step, the image classification problem is reformulated as a special configuration of object detection that treats the entire image as a special RoI. Then, a simple multi-task learning approach is used to join the image classification and object detection data, with the backbone and the RoI classification branch shared between two tasks. This two-stage approach, though very simple without a sophisticated process such as multi-instance learning (MIL) to generate pseudo labels for object proposals on the image classification data, performs rather strongly that it surpasses previous large-vocabulary object detection systems on a standard evaluation protocol of tailored LVIS. Considering that the tailored LVIS evaluation only accounts for a few hundred novel object categories, we present a new evaluation benchmark that assesses the detection of all 21,841 object classes in the ImageNet-21K dataset. The baseline approach and evaluation benchmark will be publicly available at https://github.com/SwinTransformer/Simple-21K-Detection. We hope these would ease future research on large-vocabulary object detection.



Paperid:423
Authors:Chenxin Li; Mingbao Lin; Zhiyuan Ding; Nie Lin; Yihong Zhuang; Yue Huang; Xinghao Ding; Liujuan Cao
Title: Knowledge Condensation Distillation
Abstract:
Knowledge Distillation (KD) transfers the knowledge from a high-capacity teacher network to strengthen a smaller student. Existing methods focus on excavating the knowledge hints and transferring the whole knowledge to the student. However, the knowledge redundancy arises since the knowledge shows different values to the student at different learning stages. In this paper, we propose Knowledge Condensation Distillation (KCD). Specifically, the knowledge value on each sample is dynamically estimated, based on which an Expectation-Maximization (EM) framework is forged to iteratively condense a compact knowledge set from the teacher to guide the student learning. Our approach is easy to build on top of the off-the-shelf KD methods, with no extra training parameters and negligible computation overhead. Thus, it presents one new perspective for KD, in which the student that actively identifies teacher’s knowledge in line with its aptitude can learn to learn more effectively and efficiently. Experiments on standard benchmarks manifest that the proposed KCD can well boost the performance of student model with even higher distillation efficiency. Code is available at https://github.com/dzy3/KCD.



Paperid:424
Authors:Yufei Guo; Yuanpei Chen; Liwen Zhang; YingLei Wang; Xiaode Liu; Xinyi Tong; Yuanyuan Ou; Xuhui Huang; Zhe Ma
Title: Reducing Information Loss for Spiking Neural Networks
Abstract:
The Spiking Neural Network (SNN) has attracted more and more attention recently. It adopts binary spike signals to transmit information. Benefitting from the information passing paradigm of SNNs, the multiplications of activations and weights can be replaced by additions, which are more energy-efficient. However, its “Hard Reset"" mechanism for the firing activity would ignore the difference among membrane potentials when the membrane potential is above the firing threshold, causing information loss. Meanwhile, quantifying the membrane potential to 0/1 spikes at the firing instants will inevitably introduce the quantization error thus bringing about information loss too. To address these problems, we propose a “Soft Reset"" mechanism for the supervised training-based SNNs, which will drive the membrane potential to a dynamic reset potential according to its magnitude, and Membrane Potential Rectifier (MPR) to reduce the quantization error via redistributing the membrane potential to a range close to the spikes. Experimental results show that the SNNs with the proposed “Soft Reset"" mechanism and MPR outperform their vanilla counterparts on both static and dynamic datasets.



Paperid:425
Authors:Zhendong Yang; Zhe Li; Mingqi Shao; Dachuan Shi; Zehuan Yuan; Chun Yuan
Title: Masked Generative Distillation
Abstract:
Knowledge distillation has been applied to various tasks successfully. The current distillation algorithm usually improves students’ performance by imitating the output of the teacher. This paper shows that teachers can also improve students’ representation power by guiding students’ feature recovery. From this point of view, we propose Masked Generative Distillation (MGD), which is simple: we mask random pixels of the student’s feature and force it to generate the teacher’s full feature through a simple block. MGD is a truly general feature-based distillation method, which can be utilized on various tasks, including image classification, object detection, semantic segmentation and instance segmentation. We experiment on different models with extensive datasets and the results show that all the students achieve excellent improvements. Notably, we boost ResNet-18 from 69.90% to 71.69% ImageNet top-1 accuracy, RetinaNet with ResNet-50 backbone from 37.4 to 41.0 Boundingbox mAP, SOLO based on ResNet-50 from 33.1 to 36.2 Mask mAP and DeepLabV3 based on ResNet-18 from 73.20 to 76.02 mIoU. Our codes are available at https://github.com/yzd-v/MGD.



Paperid:426
Authors:Yunshan Zhong; Mingbao Lin; Mengzhao Chen; Ke Li; Yunhang Shen; Fei Chao; Yongjian Wu; Rongrong Ji
Title: Fine-Grained Data Distribution Alignment for Post-Training Quantization
Abstract:
While post-training quantization receives popularity mostly due to its evasion in accessing the original complete training dataset, its poor performance also stems from scarce images. To alleviate this limitation, in this paper, we leverage the synthetic data introduced by zero-shot quantization with calibration dataset and propose a fine-grained data distribution alignment (FDDA) method to boost the performance of post-training quantization. The method is based on two important properties of batch normalization statistics (BNS) we observed in deep layers of the trained network, i.e., inter-class separation and intra-class incohesion. To preserve this fine-grained distribution information: 1) We calculate the per-class BNS of the calibration dataset as the BNS centers of each class and propose a BNS-centralized loss to force the synthetic data distributions of different classes to be close to their own centers. 2) We add Gaussian noise into the centers to imitate the incohesion and propose a BNS-distorted loss to force the synthetic data distribution of the same class to be close to the distorted centers. By utilizing these two fine-grained losses, our method manifests the state-of-the-art performance on ImageNet, especially when both the first and last layers are quantized to the low-bit.



Paperid:427
Authors:Jingwen Ye; Yifang Fu; Jie Song; Xingyi Yang; Songhua Liu; Xin Jin; Mingli Song; Xinchao Wang
Title: Learning with Recoverable Forgetting
Abstract:
Life-long learning aims at learning a sequence of tasks without forgetting the previously acquired knowledge. However, the involved training data may not be life-long legitimate due to privacy or copyright reasons. In practical scenarios, for instance, the model owner may wish to enable or disable the knowledge of specific tasks or specific samples from time to time. Such flexible control over knowledge transfer, unfortunately, has been largely overlooked in previous incremental or decremental learning methods, even at a problem-setup level. In this paper, we explore a novel learning scheme, termed as \textbf{L}earning w\textbf{I}th \textbf{R}ecoverable \textbf{F}orgetting (LIRF), that explicitly handles the task- or sample-specific knowledge removal and recovery. Specifically, LIRF brings in two innovative schemes, namely knowledge \emph{deposit} and \emph{withdrawal}, which allow for isolating user-designated knowledge from a pre-trained network and injecting it back when necessary. During the knowledge deposit process, the specified knowledge is extracted from the target network and stored in a deposit module, while the insensitive or general knowledge of the target network is preserved and further augmented. During knowledge withdrawal, the taken-off knowledge is added back to the target network. The deposit and withdraw processes only demand for a few epochs of finetuning on the removal data, ensuring both data and time efficiency. We conduct experiments on several datasets, and demonstrate that the proposed LIRF strategy yields encouraging results with gratifying generalization capability.



Paperid:428
Authors:Jiajun Liang; Linze Li; Zhaodong Bing; Borui Zhao; Yao Tang; Bo Lin; Haoqiang Fan
Title: Efficient One Pass Self-Distillation with Zipf's Label Smoothing
Abstract:
Self-distillation exploits non-uniform soft supervision from itself during training and improves performance without any runtime cost. However, the overhead during training is often overlooked, and yet reducing time and memory overhead during training is increasingly important in the giant models’ era. This paper proposes an efficient self-distillation method named Zipf’s Label Smoothing (Zipf’s LS), which uses the on-the-fly prediction of a network to generate soft supervision that conforms to Zipf distribution without using any contrastive samples or auxiliary parameters. Our idea comes from an empirical observation that when the network is duly trained the output values of a network’s final softmax layer, after sorting by the magnitude and averaged across samples, should follow a distribution reminiscent to Zipf’s Law in the word frequency statistics of natural languages. By enforcing this property on the sample level and throughout the whole training period, we find that the prediction accuracy can be greatly improved. Using ResNet50 on the INAT21 fine-grained classification dataset, our technique achieves +3.61% accuracy gain compared to the vanilla baseline, and 0.88% more gain against the previous label smoothing or self-distillation strategies. The implementation is publicly available at https://github.com/megvii-research/zipfls.



Paperid:429
Authors:Jinhyuk Park; Albert No
Title: Prune Your Model before Distill It
Abstract:
Knowledge distillation transfers the knowledge from a cumbersome teacher to a small student. Recent results suggest that the student-friendly teacher is more appropriate to distill since it provides more transferrable knowledge. In this work, we propose the novel framework, “prune, then distill,” that prunes the model first to make it more transferrable and then distill it to the student. We provide several exploratory examples where the pruned teacher teaches better than the original unpruned networks. We further show theoretically that the pruned teacher plays the role of regularizer in distillation, which reduces the generalization error. Based on this result, we propose a novel neural network compression scheme where the student network is formed based on the pruned teacher and then apply the “prune, then distill” strategy. The code is available at https://github.com/ososos888/prune-then-distill.



Paperid:430
Authors:Zhongnan Qu; Cong Liu; Lothar Thiele
Title: Deep Partial Updating: Towards Communication Efficient Updating for On-Device Inference
Abstract:
Emerging edge intelligence applications require the server to retrain and update deep neural networks deployed on remote edge nodes to leverage newly collected data samples. Unfortunately, it may be impossible in practice to continuously send fully updated weights to these edge nodes due to the highly constrained communication resource. In this paper, we propose the weight-wise deep partial updating paradigm, which smartly selects a small subset of weights to update in each server-to-edge communication round, while achieving a similar performance compared to full updating. Our method is established through analytically upper-bounding the loss difference between partial updating and full updating, and only updates the weights which make the largest contributions to the upper bound. Extensive experimental results demonstrate the efficacy of our partial updating methodology which achieves a high inference accuracy while updating a rather small number of weights.



Paperid:431
Authors:Zhikai Li; Liping Ma; Mengjuan Chen; Junrui Xiao; Qingyi Gu
Title: Patch Similarity Aware Data-Free Quantization for Vision Transformers
Abstract:
Vision transformers have recently gained great success on various computer vision tasks; nevertheless, their high model complexity makes it challenging to deploy on resource-constrained devices. Quantization is an effective approach to reduce model complexity, and data-free quantization, which can address data privacy and security concerns during model deployment, has received widespread interest. Unfortunately, all existing methods, such as BN regularization, were designed for convolutional neural networks and cannot be applied to vision transformers with significantly different model architectures. In this paper, we propose PSAQ-ViT, a Patch Similarity Aware data-free Quantization framework for Vision Transformers, to enable the generation of ""realistic"" samples based on the vision transformer’s unique properties for calibrating the quantization parameters. Specifically, we analyze the self-attention module’s properties and reveal a general difference (patch similarity) in its processing of Gaussian noise and real images. The above insights guide us to design a relative value metric to optimize the Gaussian noise to approximate the real images, which are then utilized to calibrate the quantization parameters. Extensive experiments and ablation studies are conducted on various benchmarks to validate the effectiveness of PSAQ-ViT, which can even outperform the real-data-driven methods.



Paperid:432
Authors:Jonghyun Bae; Woohyeon Baek; Tae Jun Ham; Jae W. Lee
Title: "L3: Accelerator-Friendly Lossless Image Format for High-Resolution, High-Throughput DNN Training"
Abstract:
The training process of deep neural networks (DNNs) is usually pipelined with stages for data preparation on CPUs followed by gradient computation on accelerators like GPUs. In an ideal pipeline, the end-to-end training throughput is eventually limited by the throughput of the accelerator, not by that of data preparation. In the past, the DNN training pipeline achieved a near-optimal throughput by utilizing datasets encoded with a lightweight, lossy image format like JPEG. However, as high-resolution, losslessly-encoded datasets become more popular for applications requiring high accuracy, a performance problem arises in the data preparation stage due to low-throughput image decoding on the CPU. Thus, we propose L3, a custom lightweight, lossless image format for high-resolution, high-throughput DNN training. The decoding process of L3 is effectively parallelized on the accelerator, thus minimizing CPU intervention for data preparation during DNN training. L3 achieves a 9.29x higher data preparation throughput than PNG, the most popular lossless image format, for the Cityscapes dataset on NVIDIA A100 GPU, which leads to 1.71x higher end-to-end training throughput. Compared to JPEG and WebP, two popular lossy image formats, L3 provides up to 1.77x and 2.87x higher end-to-end training throughput for ImageNet, respectively, at equivalent metric performance.



Paperid:433
Authors:Can Ufuk Ertenli; Emre Akbas; Ramazan Gokberk Cinbis
Title: Streaming Multiscale Deep Equilibrium Models
Abstract:
We present StreamDEQ, a method that infers frame-wise representations on videos with minimal per-frame computation. In contrast to conventional methods where compute time grows at least linearly with the network depth, we aim to update the representations in a continuous manner. For this purpose, we leverage the recently emerging implicit layer model which infers the representation of an image by solving a fixed-point problem. Our main insight is to leverage the slowly changing nature of videos and use the previous frame representation as an initial condition on each frame. This scheme effectively recycles the recent inference computations and greatly reduces the needed processing time. Through extensive experimental analysis, we show that StreamDEQ is able to recover near-optimal representations in a few frames time, and maintain an up-to-date representation throughout the video duration. Our experiments on video semantic segmentation and video object detection show that StreamDEQ achieves on par accuracy with the baseline (standard MDEQ) while being more than 3x faster.



Paperid:434
Authors:Sein Park; Yeongsang Jang; Eunhyeok Park
Title: Symmetry Regularization and Saturating Nonlinearity for Robust Quantization
Abstract:
Robust quantization improves the tolerance of networks for various implementations, allowing the maintenance of accuracy in a different bit-width or quantization policy. It offers appealing candidates, especially when the target objective (i.e., energy consumption and performance) is not static and implementations are fragmented. In this work, we perform extensive analyses to identify the sources of quantization error and present three insights to robustify the network against quantization: reduction of error propagation, range clamping for error minimization, and inherited robustness against quantization. Based on these insights, we propose two novel methods called symmetry regularization (SymReg) and saturating nonlinearity (SatNL). Applying the proposed methods during training can enhance the robustness of arbitrary neural networks against quantization on existing post-training quantization (PTQ) and quantization-aware training (QAT) algorithms, obtaining a single weight flexible enough to maintain the output quality at various conditions. We conduct extensive studies and validate the effectiveness of the proposed methods.



Paperid:435
Authors:Huanyu Wang; Wenhu Zhang; Shihao Su; Hui Wang; Zhenwei Miao; Xin Zhan; Xi Li
Title: SP-Net: Slowly Progressing Dynamic Inference Networks
Abstract:
Dynamic inference networks improve computational efficiency by executing a subset of network components, i.e., executing path, conditioned on input sample. Prevalent methods typically assign routers to computational blocks so that a computational block can be skipped or executed. However, such inference mechanisms are prone to suffer instability in the optimization of dynamic inference networks. First, a dynamic inference network is more sensitive to its routers than its computational blocks. Second, the components executed by the network vary with samples, resulting in unstable feature evolution throughout the network. To alleviate the problems above, we propose a slowly progressing dynamic inference network to stabilize the optimization. First, we design a dynamic auxiliary module to slow down the progress in routers. Moreover, we regularize the feature evolution directions across the network to smoothen the feature extraction. As a result, we conduct extensive experiments on three widely used benchmarks and show that our proposed SP-Nets achieve state-of-the-art performance in terms of efficiency and accuracy.



Paperid:436
Authors:Tan Wang; Qianru Sun; Sugiri Pranata; Karlekar Jayashree; Hanwang Zhang
Title: Equivariance and Invariance Inductive Bias for Learning from Insufficient Data
Abstract:
We are interested in learning robust models from insufficient data, without the need for any externally pre-trained checkpoints. First, compared to sufficient data, we show why insufficient data renders the model more easily biased to the limited training environments that are usually different from testing. For example, if all the training swan samples are “white”, the model may wrongly use the “white” environment to represent the intrinsic class swan. Then, we justify that equivariance inductive bias can retain the class feature while invariance inductive bias can remove the environmental feature, leaving the class feature that generalizes to any environmental changes in testing. To impose them on learning, for equivariance, we demonstrate that any off-the-shelf contrastive-based self-supervised feature learning method can be deployed; for invariance, we propose a class-wise invariant risk minimization (IRM) that efficiently tackles the challenge of missing environmental annotation in conventional IRM. State-of-the-art experimental results on real-world benchmarks (VIPriors, ImageNet100 and NICO) validate the great potential of equivariance and invariance in data-efficient learning. The code is available at https://github.com/Wangt-CN/EqInv.



Paperid:437
Authors:Chen Tang; Kai Ouyang; Zhi Wang; Yifei Zhu; Wen Ji; Yaowei Wang; Wenwu Zhu
Title: Mixed-Precision Neural Network Quantization via Learned Layer-Wise Importance
Abstract:
The exponentially large discrete search space in mixed-precision quantization (MPQ) makes it hard to determine the optimal bit-width for each layer. Previous works usually resort to iterative search methods on the training set, which consume hundreds or even thousands of GPU-hours. In this study, we reveal that some unique learnable parameters in quantization, namely the scale factors in the quantizer, can serve as importance indicators of a layer, reflecting the contribution of that layer to the final accuracy at certain bit-widths. These importance indicators naturally perceive the numerical transformation during quantization-aware training, which can precisely provide quantization sensitivity metrics of layers. However, a deep network always contains hundreds of such indicators, and training them one by one would lead to an excessive time cost. To overcome this issue, we propose a joint training scheme that can obtain all indicators at once. It considerably speeds up the indicators training process by parallelizing the original sequential training processes. With these learned importance indicators, we formulate the MPQ search problem as a one-time integer linear programming (ILP) problem. That avoids the iterative search and significantly reduces search time without limiting the bit-width search space. For example, MPQ search on ResNet18 with our indicators takes only 0.06 seconds, which improves time efficiency exponentially compared to iterative search methods. Also, extensive experiments show our approach can achieve SOTA accuracy on ImageNet for far-ranging models with various constraints (e.g., BitOps, compress rate).



Paperid:438
Authors:Matthew Dutson; Yin Li; Mohit Gupta
Title: Event Neural Networks
Abstract:
Video data is often repetitive; for example, the contents of adjacent frames are usually strongly correlated. Such redundancy occurs at multiple levels of complexity, from low-level pixel values to textures and high-level semantics. We propose Event Neural Networks (EvNets), which leverage this redundancy to achieve considerable computation savings during video inference. A defining characteristic of EvNets is that each neuron has state variables that provide it with long-term memory, which allows low-cost, high-accuracy inference even in the presence of significant camera motion. We show that it is possible to transform a wide range of neural networks into EvNets without re-training. We demonstrate our method on state-of-the-art architectures for both high- and low-level visual processing, including pose recognition, object detection, optical flow, and image enhancement. We observe roughly an order-of-magnitude reduction in computational costs compared to conventional networks, with minimal reductions in model accuracy.



Paperid:439
Authors:Junting Pan; Adrian Bulat; Fuwen Tan; Xiatian Zhu; Lukasz Dudziak; Hongsheng Li; Georgios Tzimiropoulos; Brais Martinez
Title: EdgeViTs: Competing Light-Weight CNNs on Mobile Devices with Vision Transformers
Abstract:
Self-attention based models such as vision transformers (ViTs) have emerged as a very competitive architecture alternative to convolutional neural networks (CNNs) in computer vision. Despite increasingly stronger variants with ever-higher recognition accuracies, due to the quadratic complexity of self-attention, existing ViTs are typically demanding in computation and model size. Although several successful design choices (e.g., the convolutions and hierarchical multi-stage structure) of prior CNNs have been reintroduced into recent ViTs, they are still not sufficient to meet the limited resource requirements of mobile devices. This motivates a very recent attempt to develop light ViTs based on the state-of-the-art MobileNet-v2, but still leaves a performance gap behind. In this work, pushing further along this under-studied direction we introduce EdgeViTs, a new family of light-weight ViTs that, for the first time, enable attention-based vision models to compete with the best light-weight CNNs in the tradeoff between accuracy and on-device efficiency. This is realized by introducing a highly cost-effective local-global-local (LGL) information exchange bottleneck based on optimal integration of self-attention and convolutions. For device-dedicated evaluation, rather than relying on inaccurate proxies like the number of FLOPs or parameters, we adopt a practical approach of focusing directly on on-device latency and, for the first time, energy efficiency. Extensive experiments on image classification, object detection, and semantic segmentation validate the high efficiency of our EdgeViTs when compared to the state-of-the-art efficient CNNs and ViTs in terms of accuracy-efficiency tradeoff on mobile hardware. Specifically, we show that our models are Pareto-optimal when both accuracy-latency and accuracy-energy tradeoffs are considered, achieving strict dominance over other ViTs in almost all cases and competing with the most efficient CNNs.



Paperid:440
Authors:Qinghao Hu; Gang Li; Qiman Wu; Jian Cheng
Title: PalQuant: Accelerating High-Precision Networks on Low-Precision Accelerators
Abstract:
Recently low-precision deep learning accelerators (DLAs) have become popular due to their advantages in chip area and energy consumption, yet the low-precision quantized models on these DLAs bring in severe accuracy degradation. One way to achieve both high accuracy and efficient inference is to deploy high-precision neural networks on low-precision DLAs, which is rarely studied. In this paper, we propose the PArallel Low-precision Quantization (PalQuant) method that approximates high-precision computations via learning parallel low-precision representations from scratch. In addition, we present a novel cyclic shuffle module to boost the cross-group information communication between parallel low-precision groups. Extensive experiments demonstrate that PalQuant has superior performance to state-of-the-art quantization methods in both accuracy and inference speed, e.g., for ResNet-18 network quantization, PalQuant can obtain 0.52 % higher accuracy and 1.78 times speedup simultaneously over their 4-bit counter-part on a state-of-the-art 2-bit accelerator. Code is available at https://github.com/huqinghao/PalQuant.



Paperid:441
Authors:Shangqian Gao; Feihu Huang; Yanfu Zhang; Heng Huang
Title: Disentangled Differentiable Network Pruning
Abstract:
In this paper, we propose a novel channel pruning method for compression and acceleration of Convolutional Neural Networks (CNNs). Many existing channel pruning works try to discover compact sub-networks by optimizing a regularized loss function through differentiable operations. Usually, a learnable parameter is used to characterize each channel, which entangles the width and channel importance. In this setting, the FLOPs or parameter constraints implicitly restrict the search space of the pruned model. To solve the aforementioned problems, we propose optimizing each layer’s width by relaxing the hard equality constraint used in previous works. The relaxation is inspired by the definition of the top-$k$ operation. By doing so, we partially disentangle the learning of width and channel importance, which enables independent parametrization for width and importance and makes pruning more flexible. We also introduce soft top-$k$ to improve the learning of width. Moreover, to make pruning more efficient, we use two neural networks to parameterize the importance and width. The importance generation network considers both inter-channel and inter-layer relationships. The width generation network has similar functions. In addition, our method can be easily optimized by popular SGD methods, which enjoys the benefits of differentiable pruning. Extensive experiments on CIFAR-10 and ImageNet show that our method is competitive with state-of-the-art methods.



Paperid:442
Authors:Sheng Xu; Yanjing Li; Bohan Zeng; Teli Ma; Baochang Zhang; Xianbin Cao; Peng Gao; Jinhu Lü
Title: IDa-Det: An Information Discrepancy-Aware Distillation for 1-Bit Detectors
Abstract:
Knowledge distillation (KD) has been proven to be useful for training compact object detection models. However, we observe that KD is often effective when the teacher model and student counterpart share similar proposal information. This explains why existing KD methods are less effective for 1-bit detectors, caused by a significant information discrepancy between the real-valued teacher and the 1-bit student. This paper presents an Information Discrepancy-aware strategy (IDa-Det) to distill 1-bit detectors that can effectively eliminate information discrepancies and significantly reduce the performance gap between a 1-bit detector and its real-valued counterpart. We formulate the distillation process as a bi-level optimization formulation. At the inner level, we select the representative proposals with maximum information discrepancy. We then introduce a novel entropy distillation loss to reduce the disparity based on the selected proposals. Extensive experiments demonstrate IDa-Det’s superiority over state-of-the-art 1-bit detectors and KD methods on both PASCAL VOC and COCO datasets. IDa-Det achieves a 76.9\% mAP for a 1-bit Faster-RCNN with ResNet-18 backbone. Our code is open-sourced on https://github.com/SteveTsui/IDa-Det.



Paperid:443
Authors:Yizeng Han; Yifan Pu; Zihang Lai; Chaofei Wang; Shiji Song; Junfeng Cao; Wenhui Huang; Chao Deng; Gao Huang
Title: Learning to Weight Samples for Dynamic Early-Exiting Networks
Abstract:
Early exiting is an effective paradigm for improving the inference efficiency of deep networks. By constructing classifiers with varying resource demands (the exits), such networks allow easy samples to be output at early exits, removing the need for executing deeper layers. While existing works mainly focus on the architectural design of multi-exit networks, the training strategies for such models are largely left unexplored. The current state-of-the-art models treat all samples the same during training. However, the early-exiting behavior during testing has been ignored, leading to a gap between training and testing. In this paper, we propose to bridge this gap by sample weighting. Intuitively, easy samples, which generally exit early in the network during inference, should contribute more to training early classifiers. The training of hard samples (mostly exit from deeper layers), however, should be emphasized by the late classifiers. Our work proposes to adopt a weight prediction network to weight the loss of different training samples at each exit. This weight prediction network and the backbone model are jointly optimized under a \emph{meta-learning} framework with a novel optimization objective. By bringing the adaptive behavior during inference into the training phase, we show that the proposed weighting mechanism consistently improves the trade-off between classification accuracy and inference efficiency. Code is available at https://github.com/LeapLabTHU/L2W-DEN.



Paperid:444
Authors:Zhijun Tu; Xinghao Chen; Pengju Ren; Yunhe Wang
Title: AdaBin: Improving Binary Neural Networks with Adaptive Binary Sets
Abstract:
This paper studies the Binary Neural Networks (BNNs) in which weights and activations are both binarized into 1-bit values, thus greatly reducing the memory usage and computational complexity. Since the modern deep neural networks are of sophisticated design with complex architecture for the accuracy reason, the diversity on distributions of weights and activations is very high. Therefore, the conventional sign function cannot be well used for effectively binarizing full precision values in BNNs. To this end, we present a simple yet effective approach called AdaBin to adaptively obtain the optimal binary sets {b_1, b_2} (b_1, b_2 belong to R) of weights and activations for each layer instead of a fixed set (i.e., {-1, +1}). In this way, the proposed method can better fit different distributions and increase the representation ability of binarized features. In practice, we use the center position and distance of 1-bit values to define a new binary quantization function. For the weights, we propose an equalization method to align the symmetrical center of binary distribution to real-valued distribution, and minimize the Kullback-Leibler divergence of them. Meanwhile, we introduce a gradient-based optimization method to get these two parameters for activations, which are jointly trained in an end-to-end manner. Experimental results on benchmark models and datasets demonstrate that the proposed AdaBin is able to achieve state-of-the-art performance. For instance, we obtain a 66.4% Top-1 accuracy on the ImageNet using ResNet-18 architecture, and a 69.4 mAP on PASCAL VOC using SSD300.



Paperid:445
Authors:Mohsen Fayyaz; Soroush Abbasi Koohpayegani; Farnoush Rezaei Jafari; Sunando Sengupta; Hamid Reza Vaezi Joze; Eric Sommerlade; Hamed Pirsiavash; Jürgen Gall
Title: Adaptive Token Sampling for Efficient Vision Transformers
Abstract:
While state-of-the-art vision transformer models achieve promising results in image classification, they are computationally expensive and require many GFLOPs. Although the GFLOPs of a vision transformer can be decreased by reducing the number of tokens in the network, there is no setting that is optimal for all input images. In this work, we therefore introduce a differentiable parameter-free Adaptive Token Sampler (ATS) module, which can be plugged into any existing vision transformer architecture. ATS empowers vision transformers by scoring and adaptively sampling significant tokens. As a result, the number of tokens is not constant anymore and varies for each input image. By integrating ATS as an additional layer within the current transformer blocks, we can convert them into much more efficient vision transformers with an adaptive number of tokens. Since ATS is a parameter-free module, it can be added to the off-the-shelf pre-trained vision transformers as a plug and play module, thus reducing their GFLOPs without any additional training. Moreover, due to its differentiable design, one can also train a vision transformer equipped with ATS. We evaluate the efficiency of our module in both image and video classification tasks by adding it to multiple SOTA vision transformers. Our proposed module improves the SOTA by reducing their computational costs (GFLOPs) by 2$\times$, while preserving their accuracy on the ImageNet, Kinetics-400, and Kinetics-600 datasets.



Paperid:446
Authors:Christopher Subia-Waud; Srinandan Dasmahapatra
Title: Weight Fixing Networks
Abstract:
Modern iterations of deep learning models contain millions (billions) of unique parameters--each represented by a $b$-bit number. Popular attempts at compressing neural networks (such as pruning and quantisation) have shown that many of the parameters are superfluous, which we can remove (pruning) or express with $b’ < b$ bits (quantisation) without hindering performance. Here we look to go much further in minimising the information content of networks. Rather than a channel or layer-wise encoding, we look to lossless whole-network quantisation to minimise the entropy and number of unique parameters in a network. We propose a new method, which we call Weight Fixing Networks (WFN) that we design to realise four model outcome objectives: i) very few unique weights, ii) low-entropy weight encodings, iii) unique weight values which are amenable to energy-saving versions of hardware multiplication, and iv) lossless task-performance. Some of these goals are conflicting. To best balance these conflicts, we combine a few novel (and some well-trodden) tricks; a novel regularisation term, (i, ii) a view of clustering cost as relative distance change (i, ii, iv), and a focus on whole-network re-use of weights (i, iii). Our Imagenet experiments demonstrate lossless compression using 56x fewer unique weights and a 1.9x lower weight-space entropy than SOTA quantisation approaches.



Paperid:447
Authors:Zhuofan Zong; Kunchang Li; Guanglu Song; Yali Wang; Yu Qiao; Biao Leng; Yu Liu
Title: Self-Slimmed Vision Transformer
Abstract:
Vision transformers (ViTs) have become the popular structures and outperformed convolutional neural networks (CNNs) on various vision tasks. However, such powerful transformers bring a huge computation burden, because of the exhausting token-to-token comparison. The previous works focus on dropping insignificant tokens to reduce the computational cost of ViTs. But when the dropping ratio increases, this hard manner will inevitably discard the vital tokens, which limits its efficiency. To solve the issue, we propose a generic self-slimmed learning approach for vanilla ViTs, namely SiT. Specifically, we first design a novel Token Slimming Module (TSM), which can boost the inference efficiency of ViTs by dynamic token aggregation. As a general method of token hard dropping, our TSM softly integrates redundant tokens into fewer informative ones. It can dynamically zoom visual attention without cutting off discriminative token relations in the images, even with a high slimming ratio. Furthermore, we introduce a concise Feature Recalibration Distillation (FRD) framework, wherein we design a reverse version of TSM (RTSM) to recalibrate the unstructured token in a flexible auto-encoder manner. Due to the similar structure between teacher and student, our FRD can effectively leverage structure knowledge for better convergence. Finally, we conduct extensive experiments to evaluate our SiT. It demonstrates that our method can speed up ViTs by 1.7x with negligible accuracy drop, and even speed up ViTs by 3.6x while maintaining 97% of their performance. Surprisingly, by simply arming LV-ViT with our SiT, we achieve new state-of-the-art performance on ImageNet. Code is available at https://github.com/Sense-X/SiT.



Paperid:448
Authors:Biao Qian; Yang Wang; Hongzhi Yin; Richang Hong; Meng Wang
Title: Switchable Online Knowledge Distillation
Abstract:
Online Knowledge Distillation (OKD) improves the involved models by reciprocally exploiting the difference between teacher and student. Several crucial bottlenecks over the gap between them -- e.g., Why and when does a large gap harm the performance, especially for student? How to quantify the gap between teacher and student? -- have received limited formal study. In this paper, we propose Switchable Online Knowledge Distillation (SwitOKD), to answer these questions.Instead of focusing on the accuracy gap at test phase by the existing arts, the core idea of SwitOKD is to adaptively calibrate the gap at training phase, namely distillation gap, via a switching strategy be-tween two modes -- expert mode (pause the teacher while keep the student learning) and learning mode (restart the teacher). To possess an appropriate distillation gap, we further devise an adaptive switching threshold, which provides a formal criterion as to when to switch to learning mode or expert mode, and thus improves the student’s performance. Meanwhile, the teacher benefits from our adaptive switching threshold and keeps basically on a par with other online arts. We further extend SwitOKD to multiple networks with two basis topologies. Finally, extensive experiments and analysis validate the merits of SwitOKD for classification over the state-of-the-arts. Our code is available at https://github.com/hfutqian/SwitOKD.



Paperid:449
Authors:Hadi M. Dolatabadi; Sarah Erfani; Christopher Leckie
Title: $\ell_\infty$-Robustness and Beyond: Unleashing Efficient Adversarial Training
Abstract:
Neural networks are vulnerable to adversarial attacks: adding well-crafted, imperceptible perturbations to their input can modify their output. Adversarial training is one of the most effective approaches in training robust models against such attacks. However, it is much slower than vanilla training of neural networks since it needs to construct adversarial examples for the entire training data at every iteration, hampering its effectiveness. Recently, Fast Adversarial Training (FAT) was proposed that can obtain robust models efficiently. However, the reasons behind its success are not fully understood, and more importantly, it can only train robust models for $\ell_\infty$-bounded attacks as it uses FGSM during training. In this paper, by leveraging the theory of coreset selection, we show how selecting a small subset of training data provides a general, more principled approach toward reducing the time complexity of robust training. Unlike existing methods, our approach can be adapted to a wide variety of training objectives, including TRADES, $\ell_p$-PGD, and Perceptual Adversarial Training (PAT). Our experimental results indicate that our approach speeds up adversarial training by 2-3 times while experiencing a slight reduction in the clean and robust accuracy.



Paperid:450
Authors:Tianli Zhao; Xi Sheryl Zhang; Wentao Zhu; Jiaxing Wang; Sen Yang; Ji Liu; Jian Cheng
Title: Multi-Granularity Pruning for Model Acceleration on Mobile Devices
Abstract:
For practical deep neural network design on mobile devices, it is essential to consider the constraints incurred by the computational resources and the inference latency in various applications. Among deep network acceleration approaches, pruning is a widely adopted practice to balance the computational resource consumption and the accuracy, where unimportant connections can be removed either channel-wisely or randomly with a minimal impact on model accuracy. The coarse-grained channel pruning instantly results in a significant latency reduction, while the fine-grained weight pruning is more flexible to retain accuracy. In this paper, we present a unified framework for the Joint Channel pruning and Weight pruning, named JCW, which achieves an optimal pruning proportion between channel and weight pruning. To fully optimize the trade-off between latency and accuracy, we further develop a tailored multi-objective evolutionary algorithm in the JCW framework, which enables one single round search to obtain the optimal candidate architectures for various deployment requirements. Extensive experiments demonstrate that the JCW achieves a better trade-off between the latency and accuracy against previous state-of-the-art pruning methods on the ImageNet classification dataset.



Paperid:451
Authors:Naoki Okamoto; Tsubasa Hirakawa; Takayoshi Yamashita; Hironobu Fujiyoshi
Title: Deep Ensemble Learning by Diverse Knowledge Distillation for Fine-Grained Object Classification
Abstract:
Ensemble of networks with bidirectional knowledge distillation does not significantly improve on the performance of ensemble of networks without bidirectional knowledge distillation. We think that this is because there is a relationship between the knowledge in knowledge distillation and the individuality of networks in the ensemble. In this paper, we propose a knowledge distillation for ensemble by optimizing the elements of knowledge distillation as hyperparameters. The proposed method uses graphs to represent diverse knowledge distillations. It automatically designs the knowledge distillation for the optimal ensemble by optimizing the graph structure to maximize the ensemble accuracy. Graph optimization and evaluation experiments using Stanford Dogs, Stanford Cars, CUB-200-2011, CIFAR-10, and CIFAR-100 show that the proposed method achieves higher ensemble accuracy than conventional ensembles.



Paperid:452
Authors:Hyundong Jin; Eunwoo Kim
Title: Helpful or Harmful: Inter-Task Association in Continual Learning
Abstract:
When optimizing sequentially incoming tasks, deep neural networks generally suffer from catastrophic forgetting due to their lack of ability to maintain knowledge from old tasks. This may lead to a significant performance drop of the previously learned tasks. To alleviate this problem, studies on continual learning have been conducted as a countermeasure. Nevertheless, it suffers from an increase in computational cost due to the expansion of the network size or a change in knowledge that is favorably linked to previous tasks. In this work, we propose a novel approach to differentiate helpful and harmful information for old tasks using a model search to learn a current task effectively. Given a new task, the proposed method discovers an underlying association knowledge from old tasks, which can provide additional support in acquiring the new task knowledge. In addition, by introducing a sensitivity measure to the loss of the current task from the associated tasks, we find cooperative relations between tasks while alleviating harmful interference. We apply the proposed approach to both task- and class-incremental scenarios in continual learning, using a wide range of datasets from small to large scales. Experimental results show that the proposed method outperforms a large variety of continual learning approaches for the experiments while effectively alleviating catastrophic forgetting.



Paperid:453
Authors:Xingrun Xing; Yangguang Li; Wei Li; Wenrui Ding; Yalong Jiang; Yufeng Wang; Jing Shao; Chunlei Liu; Xianglong Liu
Title: Towards Accurate Binary Neural Networks via Modeling Contextual Dependencies
Abstract:
Existing Binary Neural Networks (BNNs) mainly operate on local convolutions with binarization function. However, such simple bit operations lack the ability of modeling contextual dependencies, which is critical for learning discriminative deep representations in vision models. In this work, we tackle this issue by presenting new designs of binary neural modules, which enables BNNs to learn effective contextual dependencies. First, we propose a binary multi-layer perceptron (MLP) block as an alternative to binary convolution blocks to directly model contextual dependencies. Both short-range and long-range feature dependencies are modeled by binary MLPs, where the former provides local inductive bias and the latter breaks limited receptive field in binary convolutions. A sparse contextual interaction is achieved with the long-short range binary MLP block. Second, to improve the robustness of binary models with contextual dependencies, we compute the contextual dynamic embeddings to determine the binarization thresholds in general binary convolutional blocks. Armed with our binary MLP blocks and improved binary convolution, we build the BNNs with explicit Contextual Dependency modeling, termed as BCDNet. On the standard ImageNet-1K classification benchmark, the BCDNet achieves 72.3% Top-1 accuracy and outperforms leading binary methods by a large margin. In particular, the proposed BCDNet exceeds the state-of-the-art ReActNet-A by 2.9% Top-1 accuracy with similar operations. Our codes is available at https://github.com/Sense-GVT/BCDNet.



Paperid:454
Authors:Chien-Yu Lin; Anish Prabhu; Thomas Merth; Sachin Mehta; Anurag Ranjan; Maxwell Horton; Mohammad Rastegari
Title: SPIN: An Empirical Evaluation on Sharing Parameters of Isotropic Networks
Abstract:
Recent isotropic networks, such as ConvMixer and Vision Transformers, have found significant success across visual recognition tasks, matching or outperforming non-isotropic Convolutional Neural Networks. Isotropic architectures are particularly well-suited to cross-layer weight sharing, an effective neural network compression technique. In this paper, we perform an empirical evaluation on methods for sharing parameters in isotropic networks (SPIN). We present a framework to formalize major weight sharing design decisions and perform a comprehensive empirical evaluation of this design space. Guided by our experimental results, we propose a weight sharing strategy to generate a family of models with better overall efficiency, in terms of FLOPs and parameters versus accuracy, compared to traditional scaling methods alone, for example compressing ConvMixer by 1.9x while improving accuracy on ImageNet. Finally, we perform a qualitative study to further understand the behavior of weight sharing in isotropic architectures. The code is available at https://github.com/apple/ml-spin.



Paperid:455
Authors:Seunghyun Lee; Byung Cheol Song
Title: Ensemble Knowledge Guided Sub-network Search and Fine-Tuning for Filter Pruning
Abstract:
Conventional NAS-based pruning algorithms aim to find the sub-network with the best validation performance. However, validation performance does not successfully represent test performance, i.e., potential performance. Also, although fine-tuning the pruned network to restore the performance drop is an inevitable process, few studies have handled this issue. This paper proposes a novel sub-network search and fine-tuning method that is named Ensemble Knowledge Guidance (EKG). First, we experimentally prove that the fluctuation of the loss landscape is an effective metric to evaluate the potential performance. In order to search a sub-network with the smoothest loss landscape at a low cost, we propose a pseudo-supernet built by an ensemble sub-network knowledge distillation. Next, we propose a novel fine-tuning that re-uses the information of the search phase. We store the interim sub-networks, that is, the by-products of the search phase, and transfer their knowledge into the pruned network. Note that EKG is easy to be plugged-in and computationally efficient. For example, in the case of ResNet-50, about 45% of FLOPS is removed without any performance drop in only 315 GPU hours.



Paperid:456
Authors:Yuzhang Shang; Dan Xu; Ziliang Zong; Liqiang Nie; Yan Yan
Title: Network Binarization via Contrastive Learning
Abstract:
Neural network binarization accelerates deep models by quantizing their weights and activations into 1-bit. However, there is still a huge performance gap between Binary Neural Networks (BNNs) and their full-precision (FP) counterparts. As the quantization error caused by weights binarization has been reduced in earlier works, the activations binarization becomes the major obstacle for further improvement of the accuracy. BNN characterises a unique and interesting structure, where the binary and latent FP activations exist in the same forward pass (i.e. Binarize(a_F) = a_B). To mitigate the information degradation caused by the binarization operation from FP to binary activations, we establish a novel contrastive learning framework while training BNNs through the lens of Mutual Information (MI) maximization. MI is introduced as the metric to measure the information shared between binary and the FP activations, which assists binarization with contrastive learning. Specifically, the representation ability of the BNNs is greatly strengthened via pulling the positive pairs with binary and full-precision activations from the same input samples, as well as pushing negative pairs from different samples (the number of negative pairs can be exponentially large). This benefits the downstream tasks, not only classification but also segmentation and depth estimation, etc. The experimental results show that our method can be implemented as a pile-up module on existing state-of-the-art binarization methods and can remarkably improve the performance over them on CIFAR-10/100 and ImageNet, in addition to the great generalization ability on NYUD-v2.



Paperid:457
Authors:Yuzhang Shang; Dan Xu; Bin Duan; Ziliang Zong; Liqiang Nie; Yan Yan
Title: Lipschitz Continuity Retained Binary Neural Network
Abstract:
Relying on the premise that the performance of a binary neural network can be largely restored with eliminated quantization error between full-precision weight vectors and their corresponding binary vectors, existing works of network binarization frequently adopt the idea of model robustness to reach the aforementioned objective. However, robustness remains to be an ill-defined concept without solid theoretical support. In this work, we introduce the Lipschitz continuity, a well-defined functional property, as the rigorous criteria to define the model robustness for BNN. We then propose to retain the Lipschitz continuity as a regularization term to improve the model robustness. Particularly, while the popular Lipschitz-involved regularization methods often collapse in BNN due to its extreme sparsity, we design the Retention Matrices to approximate spectral norms of the targeted weight matrices, which can be deployed as the approximation for the Lipschitz constant of BNNs without the exact Lipschitz constant computation (NP-hard). Our experiments prove that our BNN-specific regularization method can effectively strengthen the robustness of BNN (testified on ImageNet-C), achieving state-of-the-art performance on CIFAR and ImageNet.



Paperid:458
Authors:Zhenglun Kong; Peiyan Dong; Xiaolong Ma; Xin Meng; Wei Niu; Mengshu Sun; Xuan Shen; Geng Yuan; Bin Ren; Hao Tang; Minghai Qin; Yanzhi Wang
Title: SPViT: Enabling Faster Vision Transformers via Latency-Aware Soft Token Pruning
Abstract:
Recently, Vision Transformer (ViT) has continuously established new milestones in the computer vision field, while the high computation and memory cost makes its propagation in industrial production difficult. Considering the computation complexity, the internal data pattern of ViTs, and the edge device deployment, we propose a latency-aware soft token pruning framework, SPViT, which can be set up on vanilla Transformers of both flatten and hierarchical structures, such as DeiTs and Swin-Transformers (Swin). More concretely, we design a dynamic attention-based multi-head token selector, which is a lightweight module for adaptive instance-wise token selection. We further introduce a soft pruning technique, which integrates the less informative tokens chosen by the selector module into a package token rather than discarding them completely. SPViT is bound to the trade-off between accuracy and latency requirements of specific edge devices through our proposed latency-aware training strategy. Experiment results show that SPViT significantly reduces the computation cost of ViTs with comparable performance on image classification. Moreover, SPViT can guarantee the identified model meets the latency specifications of mobile devices and FPGA, and even achieve the real-time execution of DeiT-T on mobile devices. For example, SPViT reduces the latency of DeiT-T to 26 ms (26% 41% superior to existing works) on the mobile device with 0.25% 4% higher top-1 accuracy on ImageNet. Our code is released at https://github.com/PeiyanFlying/SPViT"



Paperid:459
Authors:Ryan Humble; Maying Shen; Jorge Albericio Latorre; Eric Darve; Jose Alvarez
Title: Soft Masking for Cost-Constrained Channel Pruning
Abstract:
Structured channel pruning has been shown to significantly accelerate inference time for convolution neural networks (CNNs) on modern hardware, with a relatively minor loss of network accuracy. Recent works permanently zero these channels during training, which we observe to significantly hamper final accuracy, particularly as the fraction of the network being pruned increases. We propose Soft Masking for cost-constrained Channel Pruning (SMCP) to allow pruned channels to adaptively return to the network while simultaneously pruning towards a target cost constraint. By adding a soft mask re-parameterization of the weights and channel pruning from the perspective of removing input channels, we allow gradient updates to previously pruned channels and the opportunity for the channels to later return to the network. We then formulate input channel pruning as a global resource allocation problem. Our method outperforms prior works on both the ImageNet classification and PASCAL VOC detection datasets.



Paperid:460
Authors:Sangyun Oh; Hyeonuk Sim; Jounghyun Kim; Jongeun Lee
Title: Non-uniform Step Size Quantization for Accurate Post-Training Quantization
Abstract:
Quantization is a very effective optimization technique to reduce hardware cost and memory footprint of deep neural network (DNN) accelerators. In particular, post-training quantization (PTQ) is often preferred as it does not require a full dataset or costly retraining. However, performance of PTQ lags significantly behind that of quantization-aware training especially for low-precision networks (<= 4-bit). In this paper we propose a novel PTQ scheme to bridge the gap, with minimal impact on hardware cost. The main idea of our scheme is to increase arithmetic precision while retaining the same representational precision. The excess arithmetic precision enables us to better match the input data distribution while also presenting a new optimization problem, to which we propose a novel search-based solution. Our scheme is based on logarithmic-scale quantization, which can help reduce hardware cost through the use of shifters instead of multipliers. Our evaluation results using various DNN models on challenging computer vision tasks (image classification, object detection, semantic segmentation) show superior accuracy compared with the state-of-the-art PTQ methods at various low-bit precisions.



Paperid:461
Authors:Haoran You; Baopu Li; Zhanyi Sun; Xu Ouyang; Yingyan Lin
Title: SuperTickets: Drawing Task-Agnostic Lottery Tickets from Supernets via Jointly Architecture Searching and Parameter Pruning
Abstract:
Neural architecture search (NAS) has demonstrated amazing success in searching for efficient deep neural networks (DNNs) from a given supernet. In parallel, the lottery ticket hypothesis has shown that DNNs contain small subnetworks that can be trained from scratch to achieve a comparable or higher accuracy than original DNNs. As such, it is currently a common practice to develop efficient DNNs via a pipeline of first search and then prune. Nevertheless, doing so often requires a search-train-prune-retrain process and thus prohibitive computational cost. In this paper, we discover for the first time that both efficient DNNs and their lottery subnetworks (i.e., lottery tickets) can be directly identified from a supernet, which we term as SuperTickets, via a two-in-one training scheme with jointly architecture searching and parameter pruning. Moreover, we develop a progressive and unified SuperTickets identification strategy that allows the connectivity of subnetworks to change during supernet training, achieving better accuracy and efficiency trade-offs than conventional sparse training. Finally, we evaluate whether such identified SuperTickets drawn from one task can transfer well to other tasks, validating their potential of handling multiple tasks simultaneously. Extensive experiments and ablation studies on three tasks and four benchmark datasets validate that our proposed SuperTickets achieve boosted accuracy and efficiency trade-offs than both typical NAS and pruning pipelines, regardless of having retraining or not. Codes and pretrained models are available at https://github.com/RICE-EIC/SuperTickets.



Paperid:462
Authors:Yi Sun; Jian Li; Xin Xu
Title: Meta-GF: Training Dynamic-Depth Neural Networks Harmoniously
Abstract:
Most state-of-the-art deep neural networks use static inference graphs, which makes it impossible for such networks to dynamically adjust the depth or width of the network according to the complexity of the input data. Different from these static models, depth-adaptive neural networks, e.g. the multi-exit networks, aim at improving the computation efficiency by conducting adaptive inference conditioned on the input. To achieve adaptive inference, multiple output exits are attached at different depths of the multi-exit networks. Unfortunately, these exits usually interfere with each other in the training stage. The interference would reduce performance of the models and cause negative influences on the convergence speed. To address this problem, we investigate the gradient conflict of these multi-exit networks, and propose a novel meta-learning based training paradigm namely Meta-GF(meta gradient fusion) to harmoniously train these exits. Different from existing approaches, Meta-GF takes account of the importances of the shared parameters to each exit, and fuses the gradients of each exit by the meta-learned weights. Experimental results on CIFAR and ImageNet verify the effectiveness of the proposed method. Furthermore, the proposed Meta-GF requires no modification on the network structures and can be directly combined with previous training techniques. The code is available at https://github.com/SYVAE/MetaGF.



Paperid:463
Authors:Sayeed Shafayet Chowdhury; Nitin Rathi; Kaushik Roy
Title: Towards Ultra Low Latency Spiking Neural Networks for Vision and Sequential Tasks Using Temporal Pruning
Abstract:
Spiking Neural Networks (SNNs) can be energy efficient alternatives to commonly used deep neural networks (DNNs). However, computation over multiple timesteps increases latency and energy and incurs memory access overhead of membrane potentials. Hence, latency reduction is pivotal to obtain SNNs with high energy efficiency. But, reducing latency can have an adverse effect on accuracy. To optimize the accuracy-energy-latency trade-off, we propose a temporal pruning method which starts with an SNN of T timesteps, and reduces T every iteration of training, with threshold and leak as trainable parameters. This results in a continuum of SNNs from T timesteps, all the way up to unit timestep. Training SNNs directly with 1 timestep results in convergence failure due to layerwise spike vanishing and difficulty in finding optimum thresholds. The proposed temporal pruning overcomes this by enabling the learning of suitable layerwise thresholds with backpropagation by maintaining sufficient spiking activity. Using the proposed algorithm, we achieve top-1 accuracy of 93.05%, 70.15% and 69.00% on CIFAR-10, CIFAR-100 and ImageNet, respectively with VGG16, in just 1 timestep. Note, SNNs with leaky-integrate-and-fire (LIF) neurons behave as Recurrent Neural Networks (RNNs), with the membrane potential retaining information of previous inputs. The proposed SNNs also enable performing sequential tasks such as reinforcement learning on Cartpole and Atari pong environments using only 1 to 5 timesteps.



Paperid:464
Authors:Kirill Solodskikh; Vladimir Chikin; Ruslan Aydarkhanov; Dehua Song; Irina Zhelavskaya; Jiansheng Wei
Title: Towards Accurate Network Quantization with Equivalent Smooth Regularizer
Abstract:
Neural network quantization techniques have been a prevailing way to reduce the inference time and storage cost of full-precision models for mobile devices. However, they still suffer from accuracy degradation due to inappropriate gradients in the optimization phase, especially for low-bit precision network and low-level vision tasks. To alleviate this issue, this paper defines a family of equivalent smooth regularizers for neural network quantization, named as SQR, which represents the equivalent of actual quantization error. Based on the definition, we propose a novel QSin regularizer as an instance to evaluate the performance of SQR, and also build up an algorithm to train the network for integer weight and activation. Extensive experimental results on classification and SR tasks reveal that the proposed method achieves higher accuracy than other prominent quantization approaches. Especially for SR task, our method alleviates the plaid artifacts effectively for quantized networks in terms of visual quality.



Paperid:465
Authors:Vladimir Chikin; Kirill Solodskikh; Irina Zhelavskaya
Title: Explicit Model Size Control and Relaxation via Smooth Regularization for Mixed-Precision Quantization
Abstract:
While Deep Neural Networks (DNNs) quantization leads to a significant reduction in computational and storage costs, it reduces model capacity and therefore, usually leads to an accuracy drop. One of the possible ways to overcome this issue is to use different quantization bit-widths for different layers. The main challenge of the mixed-precision approach is to define the bit-widths for each layer, while staying under memory and latency requirements. Motivated by this challenge, we introduce a novel technique for explicit complexity control of DNNs quantized to mixed-precision, which uses smooth optimization on the surface containing neural networks of constant size. Furthermore, we introduce a family of smooth quantization regularizers, which can be used jointly with our complexity control method for both post-training mixed-precision quantization and quantization-aware training. Our approach can be applied to any neural network architecture. Experiments show that the proposed techniques reach state-of-the-art results.



Paperid:466
Authors:Han-Byul Kim; Eunhyeok Park; Sungjoo Yoo
Title: BASQ: Branch-Wise Activation-Clipping Search Quantization for Sub-4-Bit Neural Networks
Abstract:
In this paper, we propose Branch-wise Activation-clipping Search Quantization (BASQ), which is a novel quantization method for low-bit activation. BASQ optimizes clip value in continuous search space while simultaneously searching L2 decay weight factor for updating clip value in discrete search space. We also propose a novel block structure for low precision that works properly on both MobileNet and ResNet structures with branch-wise searching. We evaluate the proposed methods by quantizing both weights and activations to 4-bit or lower. Contrary to the existing methods which are effective only for redundant networks, e.g., ResNet-18, or highly optimized networks, e.g., MobileNet-v2, our proposed method offers constant competitiveness on both types of networks across low precisions from 2 to 4-bits. Specifically, our 2-bit MobileNet-v2 offers top-1 accuracy of 64.71% on ImageNet, outperforming the existing method by a large margin (2.8%), and our 4-bit MobileNet-v2 gives 71.98% which is comparable to the full-precision accuracy 71.88% while our uniform quantization method offers comparable accuracy of 2-bit ResNet-18 to the state-of-the-art non-uniform quantization method.



Paperid:467
Authors:Geng Yuan; Sung-En Chang; Qing Jin; Alec Lu; Yanyu Li; Yushu Wu; Zhenglun Kong; Yanyue Xie; Peiyan Dong; Minghai Qin; Xiaolong Ma; Xulong Tang; Zhenman Fang; Yanzhi Wang
Title: You Already Have It: A Generator-Free Low-Precision DNN Training Framework Using Stochastic Rounding
Abstract:
Stochastic rounding is a critical technique used in low-precision deep neural networks (DNNs) training to ensure good model accuracy. However, it requires a large number of random numbers generated on the fly. This is not a trivial task on the hardware platforms such as FPGA and ASIC. The widely used solution is to introduce random number generators with extra hardware costs. In this paper, we innovatively propose to employ the stochastic property of DNN training process itself and directly extract random numbers from DNNs in a self-sufficient manner. We propose different methods to obtain random numbers from different sources in neural networks and a generator-free framework is proposed for low-precision DNN training on a variety of deep learning tasks. Moreover, we evaluate the quality of the extracted random numbers and find that high-quality random numbers widely exist in DNNs, while their quality can even pass the NIST test suite.



Paperid:468
Authors:Yufei Guo; Liwen Zhang; Yuanpei Chen; Xinyi Tong; Xiaode Liu; YingLei Wang; Xuhui Huang; Zhe Ma
Title: Real Spike: Learning Real-Valued Spikes for Spiking Neural Networks
Abstract:
Brain-inspired spiking neural networks (SNNs) have recently drawn more and more attention due to their event-driven and energy efficient characteristics. The integration of storage and computation paradigm on neuromorphic hardwares makes SNNs much different from Deep Neural Networks (DNNs). In this paper, we argue that SNNs may not benefit from the weight-sharing mechanism, which can effectively reduce parameters and improve inference efficiency in DNNs, in some hardwares, and assume that an SNN with unshared convolution kernels could perform better. Motivated by this assumption, a training-inference decoupling method for SNNs named as Real Spike is proposed, which not only enjoys both unshared convolution kernels and binary spikes in inference time but also aintains both shared convolution kernels and Real-valued Spikes during training. This decoupling mechanism of SNN is realized by a re-parameterization technique. Furthermore, based on the training-inference-decoupled idea, a series of other ways for constructing Real Spike on different levels are presented, which also enjoy shared convolutions in the inference and are friendly to both neuromorphic and non-neuromorphic hardware platforms. A theoretical proof is given to clarify that the Real Spike-based SNN network is superior to its vanilla counterpart. Experimental results show that all different Real Spike versions can consistently improve the SNN performance. Moreover, the proposed method outperforms the state-of-the-art models on both non-spiking static and neuromorphic datasets.



Paperid:469
Authors:Vaikkunth Mugunthan; Eric Lin; Vignesh Gokul; Christian Lau; Lalana Kagal; Steve Pieper
Title: FedLTN: Federated Learning for Sparse and Personalized Lottery Ticket Networks
Abstract:
Federated learning (FL) enables clients to collaboratively train a model, while keeping their local training data decentralized. However, high communication costs, data heterogeneity across clients, and lack of personalization techniques hinder the development of FL. In this paper, we propose FedLTN, a novel approach motivated by the well-known Lottery Ticket Hypothesis to learn sparse and personalized lottery ticket networks (LTNs) for communication-efficient and personalized FL under non-identically and independently distributed (non-IID) data settings. Preserving batch-norm statistics of local clients, postpruning without rewinding, and aggregation of LTNs using server momentum ensures that our approach significantly outperforms existing state-of-the-art solutions. Experiments on CIFAR-10 and TinyImageNet datasets show the efficacy of our approach in learning personalized models while significantly reducing communication costs.



Paperid:470
Authors:Joshua Andle; Salimeh Yasaei Sekeh
Title: Theoretical Understanding of the Information Flow on Continual Learning Performance
Abstract:
Continual learning (CL) requires a model to continually learn new tasks with incremental available information while retaining previous knowledge. Despite the numerous previous approaches to CL, most of them still suffer forgetting, expensive memory cost, or lack sufficient theoretical understanding. While different CL training regimes have been extensively studied empirically, insufficient attention has been paid to the underlying theory. In this paper, we establish a probabilistic framework to analyze information flow through layers in networks for sequential tasks and its impact on learning performance. Our objective is to optimize the information preservation between layers while learning new tasks. This manages task-specific knowledge passing throughout the layers while maintaining model performance on previous tasks. Our analysis provides novel insights into information adaptation within the layers during incremental task learning. We provide empirical evidence and practically highlight the performance improvement across multiple tasks. Code is available at https://github.com/Sekeh-Lab/InformationFlow-CL.



Paperid:471
Authors:Youngeun Kim; Yuhang Li; Hyoungseob Park; Yeshwanth Venkatesha; Ruokai Yin; Priyadarshini Panda
Title: Exploring Lottery Ticket Hypothesis in Spiking Neural Networks
Abstract:
Spiking Neural Networks (SNNs) have recently emerged as a new generation of low-power deep neural networks, which is suitable to be implemented on low-power mobile/edge devices. As such devices have limited memory storage, neural pruning on SNNs has been widely explored in recent years. Most existing SNN pruning works focus on shallow SNNs (2 6 layers), however, deeper SNNs (>16 layers) are proposed by state-of-the-art SNN works, which is difficult to be compatible with the current SNN pruning work. To scale up a pruning technique towards deep SNNs, we investigate Lottery Ticket Hypothesis (LTH) which states that dense networks contain smaller subnetworks (i.e., winning tickets) that achieve comparable performance to the dense networks. Our studies on LTH reveal that the winning tickets consistently exist in deep SNNs across various datasets and architectures, providing up to 97% sparsity without huge performance degradation. However, the iterative searching process of LTH brings a huge training computational cost when combined with the multiple timesteps of SNNs. To alleviate such heavy searching cost, we propose Early-Time (ET) ticket where we find the important weight connectivity from a smaller number of timesteps. The proposed ET ticket can be seamlessly combined with a common pruning techniques for finding winning tickets, such as Iterative Magnitude Pruning (IMP) and Early-Bird (EB) tickets. Our experiment results show that the proposed ET ticket reduces search time by up to 38% compared to IMP or EB methods. Code is available at Github.



Paperid:472
Authors:Juseung Yun; Janghyeon Lee; Hyounguk Shon; Eojindl Yi; Seung Hwan Kim; Junmo Kim
Title: On the Angular Update and Hyperparameter Tuning of a Scale-Invariant Network
Abstract:
Modern deep neural networks are equipped with normalization layers such as batch normalization or layer normalization to enhance and stabilize training dynamics. If a network contains such normalization layers, the optimization objective is invariant to the scale of the neural network parameters. The scale-invariance induces the neural network’s output to be only affected by the weights’ direction and not the weights’ scale. We first find a common feature of good hyperparameter combinations on such a scale-invariant network, including learning rate, weight decay, number of data samples, and batch size. Then we observe that hyperparameter setups that lead to good performance show similar degrees of angular update during one epoch. Using a stochastic differential equation, we analyze the angular update and show how each hyperparameter affects it. With this relationship, we can derive a simple hyperparameter tuning method and apply it to the efficient hyperparameter search.



Paperid:473
Authors:Pavlo Molchanov; Jimmy Hall; Hongxu Yin; Jan Kautz; Nicolo Fusi; Arash Vahdat
Title: LANA: Latency Aware Network Acceleration
Abstract:
We introduce latency-aware network acceleration (LANA)-an approach that builds on neural architecture search technique to accelerate neural networks. LANA consists of two phases: in the first phase, it trains many alternative operations for every layer of a target network using layer-wise feature map distillation. In the second phase, it solves the combinatorial selection of efficient operations using a novel constrained integer linear optimization (ILP) approach. ILP brings unique properties as it (i) performs NAS within a few seconds to minutes, (ii) easily satisfies budget constraints, (iii) works on the layer-granularity, (iv) supports a huge search space O(10^100), surpassing prior search approaches in efficacy and efficiency. In extensive experiments, we show that LANA yields efficient and accurate models constrained by a target latency budget, while being significantly faster than other techniques. We analyze three popular network architectures: EfficientNetV1, EfficientNetV2 and ResNeST, and achieve accuracy improvement (up to 3.0%) for all models when compressing larger models. LANA achieves significant speed-ups (up to 5x) with minor to no accuracy drop on GPU and CPU.



Paperid:474
Authors:Zhe Wang; Jie Lin; Xue Geng; Mohamed M. Sabry Aly; Vijay Chandrasekhar
Title: RDO-Q: Extremely Fine-Grained Channel-Wise Quantization via Rate-Distortion Optimization
Abstract:
Allocating different bit widths to different channels and quantizing them independently bring higher quantization precision and accuracy. Most of prior works use equal bit width to quantize all layers or channels, which is sub-optimal. On the other hand, it is very challenging to explore the hyperparameter space of channel bit widths, as the search space increases exponentially with the number of channels, which could be tens of thousand in a deep neural network. In this paper, we address the problem of efficiently exploring the hyperparameter space of channel bit widths. We formulate the quantization of deep neural networks as a rate-distortion optimization problem, and present an ultra-fast algorithm to search the bit allocation of channels. Our approach has only linear time complexity and can find the optimal bit allocation within a few minutes on CPU. In addition, we provide an effective way to improve the performance on target hardware platforms. We restrict the bit rate (size) of each layer to allow as many weights and activations as possible to be stored on-chip, and incorporate hardware-aware constraints into our objective function. The hardware-aware constraints do not cause additional overhead to optimization, and have very positive impact on hardware performance. Experimental results show that our approach achieves state-of-the-art results on four deep neural networks, ResNet-18, ResNet-34, ResNet-50, and MobileNet-v2, on ImageNet. Hardware simulation results demonstrate that our approach is able to bring up to 3.5x and 3.0x speedups on two deep-learning accelerators, TPU and Eyeriss, respectively.



Paperid:475
Authors:Ahmet Caner Yüzügüler; Nikolaos Dimitriadis; Pascal Frossard
Title: U-Boost NAS: Utilization-Boosted Differentiable Neural Architecture Search
Abstract:
Optimizing resource utilization in target platforms is key to achieving high performance during DNN inference. While optimizations have been proposed for inference latency, memory footprint, and energy consumption, prior hardware-aware neural architecture search (NAS) methods have omitted resource utilization, preventing DNNs to take full advantage of the target inference platforms. Modeling resource utilization efficiently and accurately is challenging, especially for widely-used array-based inference accelerators such as Google TPU. In this work, we propose a novel hardware-aware NAS framework that does not only optimize for task accuracy and inference latency, but also for resource utilization. We also propose and validate a new computational model for resource utilization in inference accelerators. By using the proposed NAS framework and the proposed resource utilization model, we achieve 2.8 - 4x speedup for DNN inference compared to prior hardware-aware NAS methods while attaining similar or improved accuracy in image classification on CIFAR-10 and Imagenet-100 datasets.



Paperid:476
Authors:Zhihang Yuan; Chenhao Xue; Yiqi Chen; Qiang Wu; Guangyu Sun
Title: PTQ4ViT: Post-Training Quantization for Vision Transformers with Twin Uniform Quantization
Abstract:
Quantization is one of the most effective methods to compress neural networks, which has achieved great success on convolutional neural networks (CNNs). Recently, vision transformers have demonstrated great potential in computer vision. However, previous post-training quantization methods performed not well on vision transformer, resulting in more than 1% accuracy drop even in 8-bit quantization. Therefore, we analyze the problems of quantization on vision transformers. We observe the distributions of activation values after softmax and GELU functions are quite different from the Gaussian distribution. We also observe that common quantization metrics, such as MSE and cosine distance, are inaccurate to determine the optimal scaling factor. In this paper, we propose the twin uniform quantization method to reduce the quantization error on these activation values. And we propose to use a Hessian guided metric to evaluate different scaling factors, which improves the accuracy of calibration with a small cost. To enable the fast quantization of vision transformers, we develop an efficient framework, PTQ4ViT. Experiments show the quantized vision transformers achieve near-lossless prediction accuracy (less than 0.5% drop at 8-bit quantization) on the ImageNet classification task.



Paperid:477
Authors:Jiseok Youn; Jaehun Song; Hyung-Sin Kim; Saewoong Bahk
Title: Bitwidth-Adaptive Quantization-Aware Neural Network Training: A Meta-Learning Approach
Abstract:
Deep neural network quantization with adaptive bitwidths has gained increasing attention due to the ease of model deployment on various platforms with different resource budgets. In this paper, we propose a meta-learning approach to achieve this goal. Specifically, we propose MEBQAT, a simple yet effective way of bitwidth-adaptive quantization aware training (QAT) where meta-learning is effectively combined with QAT by redefining meta-learning tasks to incorporate bitwidths. After being deployed on a platform, MEBQAT allows the (meta-)trained model to be quantized to any candidate bitwidth then helps to conduct inference without much accuracy drop from quantization. Moreover, with a few-shot learning scenario, MEBQAT can also adapt a model to any bitwidth as well as any unseen target classes by adding conventional optimization or metric-based meta-learning. We design variants of MEBQAT to support both (1) a bitwidth-adaptive quantization scenario and (2) a new few-shot learning scenario where both quantization bitwidths and target classes are jointly adapted. We experimentally demonstrate their validity in multiple QAT schemes. By comparing their performance to (bitwidth-dedicated) QAT, existing bitwidth adaptive QAT and vanilla meta-learning, we find that merging bitwidths into meta-learning tasks achieves a higher level of robustness.



Paperid:478
Authors:Yao Lu; Wen Yang; Yunzhe Zhang; Zuohui Chen; Jinyin Chen; Qi Xuan; Zhen Wang; Xiaoniu Yang
Title: Understanding the Dynamics of DNNs Using Graph Modularity
Abstract:
There are good arguments to support the claim that deep neural networks (DNNs) capture better feature representations than the previous hand-crafted feature engineering, which leads to a significant performance improvement. In this paper, we move a tiny step towards understanding the dynamics of feature representations over layers. Specifically, we model the process of class separation of intermediate representations in pre-trained DNNs as the evolution of communities in dynamic graphs. Then, we introduce modularity, a generic metric in graph theory, to quantify the evolution of communities. In the preliminary experiment, we find that modularity roughly tends to increase as the layer goes deeper and the degradation and plateau arise when the model complexity is great relative to the dataset. Through an asymptotic analysis, we prove that modularity can be broadly used for different applications. For example, modularity provides new insights to quantify the difference between feature representations. More crucially, we demonstrate that the degradation and plateau in modularity curves represent redundant layers in DNNs and can be pruned with minimal impact on performance, which provides theoretical guidance for layer pruning. Our code is available at https://github.com/yaolu-zjut/Dynamic-Graphs-Construction.



Paperid:479
Authors:Gianni Franchi; Xuanlong Yu; Andrei Bursuc; Emanuel Aldea; Severine Dubuisson; David Filliat
Title: Latent Discriminant Deterministic Uncertainty
Abstract:
Predictive uncertainty estimation is essential for deploying Deep Neural Networks in real-world autonomous systems. However, most successful approaches are computationally intensive. In this work, we attempt to address these challenges in the context of autonomous driving perception tasks. Recently proposed Deterministic Uncertainty Methods (DUM) can only partially meet such requirements as their scalability to complex computer vision tasks is not obvious. In this work we advance a scalable and effective DUM for high-resolution semantic segmentation, that relaxes the Lipschitz constraint typically hindering practicality of such architectures. We learn a discriminant latent space by leveraging a distinction maximization layer over an arbitrarily-sized set of trainable prototypes. Our approach achieves competitive results over Deep Ensembles, the state-of-the-art for uncertainty prediction, on image classification, segmentation and monocular depth estimation tasks.



Paperid:480
Authors:Simon Vandenhende; Dhruv Mahajan; Filip Radenovic; Deepti Ghadiyaram
Title: Making Heads or Tails: Towards Semantically Consistent Visual Counterfactuals
Abstract:
A visual counterfactual explanation replaces image regions in a query image with regions from a distractor image such that the system’s decision on the transformed image changes to the distractor class. In this work, we present a novel framework for computing visual counterfactual explanations based on two key ideas. First, we enforce that the replaced and replacer regions contain the same semantic part, resulting in more semantically consistent explanations. Second, we use multiple distractor images in a computationally efficient way and obtain more discriminative explanations with fewer region replacements. Our approach is 27 % more semantically consistent and an order of magnitude faster than a competing method on three fine-grained image recognition datasets. We highlight the utility of our counterfactuals over existing works through machine teaching experiments where we teach humans to classify different bird species. We also complement our explanations with the vocabulary of parts and attributes that contributed the most to the system’s decision. In this task as well, we obtain state-of-the-art results when using our counterfactual explanations relative to existing works, reinforcing the importance of semantically consistent explanations. Source code is available at https://github.com/facebookresearch/visual-counterfactuals.



Paperid:481
Authors:Sunnie S. Y. Kim; Nicole Meister; Vikram V. Ramaswamy; Ruth Fong; Olga Russakovsky
Title: HIVE: Evaluating the Human Interpretability of Visual Explanations
Abstract:
As AI technology is increasingly applied to high-impact, high-risk domains, there have been a number of new methods aimed at making AI models more human interpretable. Despite the recent growth of interpretability work, there is a lack of systematic evaluation of proposed techniques. In this work, we introduce HIVE (Human Interpretability of Visual Explanations), a novel human evaluation framework that assesses the utility of explanations to human users in AI-assisted decision making scenarios, and enables falsifiable hypothesis testing, cross-method comparison, and human-centered evaluation of visual interpretability methods. To the best of our knowledge, this is the first work of its kind. Using HIVE, we conduct IRB-approved human studies with nearly 1000 participants and evaluate four methods that represent the diversity of computer vision interpretability works: GradCAM, BagNet, ProtoPNet, and ProtoTree. Our results suggest that explanations engender human trust, even for incorrect predictions, yet are not distinct enough for users to distinguish between correct and incorrect predictions. We open-source HIVE to enable future studies and encourage more human-centered approaches to interpretability research.



Paperid:482
Authors:Uddeshya Upadhyay; Shyamgopal Karthik; Yanbei Chen; Massimiliano Mancini; Zeynep Akata
Title: BayesCap: Bayesian Identity Cap for Calibrated Uncertainty in Frozen Neural Networks
Abstract:
High-quality calibrated uncertainty estimates are crucial for numerous real-world applications, especially for deep learning-based deployed ML systems. While Bayesian deep learning techniques allow uncertainty estimation, training them with large-scale datasets is an expensive process that does not always yield models competitive with non-Bayesian counterparts. Moreover, many of the high-performing deep learning models that are already trained and deployed are non-Bayesian in nature, and do not provide uncertainty estimates. To address these issues, we propose BayesCap that learns a Bayesian identity mapping for the frozen model, allowing uncertainty estimation. BayesCap is a memory-efficient method that can be trained on a small fraction of the original dataset, enhancing pretrained non-Bayesian computer vision models by providing calibrated uncertainty estimates for the predictions without (i) hampering the performance of the model and (ii) the need for expensive retraining the model from scratch. The proposed method is agnostic to various architectures and tasks. We show the efficacy of our method on a wide variety of tasks with a diverse set of architectures, including image super-resolution, deblurring, inpainting, and crucial application such as medical image translation. Moreover, we apply the derived uncertainty estimates to detect out-of-distribution samples in critical scenarios like depth estimation in autonomous driving.



Paperid:483
Authors:Osman Tursun; Simon Denman; Sridha Sridharan; Clinton Fookes
Title: SESS: Saliency Enhancing with Scaling and Sliding
Abstract:
High-quality saliency maps are essential in several machine learning application areas including explainable AI and weakly supervised object detection and segmentation. Many techniques have been developed to generate better saliency using neural networks. However, they are often limited to specific saliency visualisation methods or saliency issues. We propose a novel saliency enhancing approach called \textbf{SESS} (\textbf{S}aliency \textbf{E}nhancing with \textbf{S}caling and \textbf{S}liding). It is a method and model agnostic extension to existing saliency map generation methods. With SESS, existing saliency approaches become robust to scale variance, multiple occurrences of target objects, presence of distractors and generate less noisy and more discriminative saliency maps. SESS improves saliency by fusing saliency maps extracted from multiple patches at different scales from different areas, and combines these individual maps using a novel fusion scheme that incorporates channel-wise weights and spatial weighted average. To improve efficiency, we introduce a pre-filtering step that can exclude uninformative saliency maps to improve efficiency while still enhancing overall results. We evaluate SESS on object recognition and detection benchmarks where it achieves significant improvement. The code is released publicly to enable researchers to verify performance and further development. Code is available at https://github.com/neouyghur/SESS.



Paperid:484
Authors:Roni Paiss; Hila Chefer; Lior Wolf
Title: No Token Left Behind: Explainability-Aided Image Classification and Generation
Abstract:
The application of zero-shot learning in computer vision has been revolutionized by the use of image-text matching models. The most notable example, CLIP, has been widely used for both zero-shot classification and guiding generative models with a text prompt. However, the zero-shot use of CLIP is unstable with respect to the phrasing of the input text, making it necessary to carefully engineer the prompts used. We find that this instability stems from a selective similarity score, which is based only on a subset of the semantically meaningful input tokens. To mitigate it, we present a novel explainability-based approach, which adds a loss term to ensure that CLIP focuses on all relevant semantic parts of the input, in addition to employing the CLIP similarity loss used in previous works. When applied to one-shot classification through prompt engineering, our method yields an improvement in the recognition rate, without additional training or fine-tuning. Additionally, we show that CLIP guidance of generative models using our method significantly improves the generated images. Finally, we demonstrate a novel use of CLIP guidance for text-based image generation with spatial conditioning on object location, by requiring the image explainability heatmap for each object to be confined to a pre-determined bounding box.



Paperid:485
Authors:Dawid Rymarczyk; Łukasz Struski; Michał Górszczak; Koryna Lewandowska; Jacek Tabor; Bartosz Zieliński
Title: Interpretable Image Classification with Differentiable Prototypes Assignment
Abstract:
Existing prototypical-based models address the black-box nature of deep learning. However, they are sub-optimal as they often assume separate prototypes for each class, require multi-step optimization, make decisions based on prototype absence (so-called negative reasoning process), and derive vague prototypes. To address those shortcomings, we introduce ProtoPool, an interpretable prototype-based model with positive reasoning and three main novelties. Firstly, we reuse prototypes in classes, which significantly decreases their number. Secondly, we allow automatic, fully differentiable assignment of prototypes to classes, which substantially simplifies the training process. Finally, we propose a new focal similarity function that contrasts the prototype from the background and consequently concentrates on more salient visual features. We show that ProtoPool obtains state-of-the-art accuracy on the CUB-200-2011 and the Stanford Cars datasets, substantially reducing the number of prototypes. We provide a theoretical analysis of the method and a user study to show that our prototypes capture more salient features than those obtained with competitive methods. We made the code available at https://github.com/gmum/ProtoPool.



Paperid:486
Authors:Yunhao Ge; Yao Xiao; Zhi Xu; Xingrui Wang; Laurent Itti
Title: "Contributions of Shape, Texture, and Color in Visual Recognition"
Abstract:
We investigate the contributions of three important features of the human visual system (HVS)---shape, texture, and color ---to object classification. We build a humanoid vision engine (HVE) that explicitly and separately computes shape, texture, and color features from images. The resulting feature vectors are then concatenated to support the final classification. We show that HVE can summarize and rank-order the contributions of the three features to object recognition. We use human experiments to confirm that both HVE and humans predominantly use some specific features to support the classification of specific classes (e.g., texture is the dominant feature to distinguish a zebra from other quadrupeds, both for humans and HVE). With the help of HVE, given any environment (dataset), we can summarize the most important features for the whole task (global; e.g., color is the most important feature overall for classification with the CUB dataset), and for each class (local; e.g., shape is the most important feature to recognize boats in the iLab-20M dataset). To demonstrate more usefulness of HVE, we use it to simulate the open-world zero-shot learning ability of humans with no attribute labeling. Finally, we show that HVE can also simulate human imagination ability with the combination of different features.



Paperid:487
Authors:Paul Jacob; Éloi Zablocki; Hédi Ben-Younes; Mickaël Chen; Patrick Pérez; Matthieu Cord
Title: STEEX: Steering Counterfactual Explanations with Semantics
Abstract:
As deep learning models are increasingly used in safety-critical applications, explainability and trustworthiness become major concerns. For simple images, such as low-resolution face portraits, synthesizing visual counterfactual explanations has recently been proposed as a way to uncover the decision mechanisms of a trained classification model. In this work, we address the problem of producing counterfactual explanations for high-quality images and complex scenes. Leveraging recent semantic-to-image models, we propose a new generative counterfactual explanation framework that produces plausible and sparse modifications which preserve the overall scene structure. Furthermore, we introduce the concept of ""region-targeted counterfactual explanations"", and a corresponding framework, where users can guide the generation of counterfactuals by specifying a set of semantic regions of the query image the explanation must be about. Extensive experiments are conducted on challenging datasets including high-quality portraits (CelebAMask-HQ) and driving scenes (BDD100k).



Paperid:488
Authors:Jindong Gu; Volker Tresp; Yao Qin
Title: Are Vision Transformers Robust to Patch Perturbations?
Abstract:
Recent advances in Vision Transformer (ViT) have demonstrated its impressive performance in image classification, which makes it a promising alternative to Convolutional Neural Network (CNN). Unlike CNNs, ViT represents an input image as a sequence of image patches. The patch-based input image representation makes the following question interesting: How does ViT perform when individual input image patches are perturbed with natural corruptions or adversarial perturbations, compared to CNNs? In this work, we study the robustness of ViT to patch-wise perturbations. Surprisingly, we {find} that ViTs are more robust to naturally corrupted patches than CNNs, whereas they are more vulnerable to adversarial patches. Furthermore, we discover that the attention mechanism greatly affects the robustness of vision transformers. Specifically, the attention module can help improve the robustness of ViT by effectively ignoring natural corrupted patches. However, when ViTs are attacked by an adversary, the attention mechanism can be easily fooled to focus more on the adversarially perturbed patches and cause a mistake. Based on our analysis, we propose a simple temperature-scaling based method to {improve} the robustness of ViT against adversarial patches. Extensive qualitative and quantitative experiments are performed to support our findings, understanding, and improvement of ViT robustness to patch-wise perturbations across a set of transformer-based architectures.



Paperid:489
Authors:Gautam Machiraju; Sylvia Plevritis; Parag Mallick
Title: A Dataset Generation Framework for Evaluating Megapixel Image Classifiers \& Their Explanations
Abstract:
Deep learning-based megapixel image classifiers have exceptional prediction performance in a number of domains, including clinical pathology. However, extracting reliable, human-interpretable model explanations has remained challenging. Because real-world megapixel images often contain latent image features highly correlated with image labels, it is difficult to distinguish correct explanations from incorrect ones. Furthering this issue are the flawed assumptions and designs of today’s classifiers. To investigate classification and explanation performance, we introduce a framework to (a) generate synthetic control images that reflect common properties of megapixel images and (b) evaluate average test-set correctness. By benchmarking two commonplace Convolutional Neural Networks (CNNs), we demonstrate how this interpretability evaluation framework can inform architecture selection beyond classification performance -- in particular, we show that a simple Attention-based architecture identifies salient objects in all seven scenarios, while a standard CNN fails to do so in six scenarios. This work carries widespread applicability to any megapixel imaging domain.



Paperid:490
Authors:Stefan Kolek; Duc Anh Nguyen; Ron Levie; Joan Bruna; Gitta Kutyniok
Title: Cartoon Explanations of Image Classifiers
Abstract:
We present CartoonX (Cartoon Explanation), a novel model-agnostic explanation method tailored towards image classifiers and based on the rate-distortion explanation (RDE) framework. Natural images are roughly piece-wise smooth signals---also called cartoon-like images---and tend to be sparse in the wavelet domain. CartoonX is the first explanation method to exploit this by requiring its explanations to be sparse in the wavelet domain, thus extracting the relevant piece-wise smooth part of an image instead of relevant pixel-sparse regions. We demonstrate that CartoonX can reveal novel valuable explanatory information, particularly for misclassifications. Moreover, we show that CartoonX achieves a lower distortion with fewer coefficients than state-of-the-art methods.



Paperid:491
Authors:Quan Zheng; Ziwei Wang; Jie Zhou; Jiwen Lu
Title: Shap-CAM: Visual Explanations for Convolutional Neural Networks Based on Shapley Value
Abstract:
Explaining deep convolutional neural networks has been recently drawing increasing attention since it helps to understand the networks’ internal operations and why they make certain decisions. Saliency maps, which emphasize salient regions largely connected to the network’s decision-making, are one of the most common ways for visualizing and analyzing deep networks in the computer vision community. However, saliency maps generated by existing methods cannot represent authentic information in images due to the unproven proposals about the weights of activation maps which lack solid theoretical foundation and fail to consider the relations between each pixels. In this paper, we develop a novel post-hoc visual explanation method called Shap-CAM based on class activation mapping. Unlike previous gradient-based approaches, Shap-CAM gets rid of the dependence on gradients by obtaining the importance of each pixels through Shapley value. We demonstrate that Shap-CAM achieves better visual performance and fairness for interpreting the decision making process. Our approach outperforms previous methods on both recognition and localization tasks.



Paperid:492
Authors:Jiazhen Ji; Huan Wang; Yuge Huang; Jiaxiang Wu; Xingkun Xu; Shouhong Ding; ShengChuan Zhang; Liujuan Cao; Rongrong Ji
Title: Privacy-Preserving Face Recognition with Learnable Privacy Budgets in Frequency Domain
Abstract:
Face recognition technology has been used in many fields due to its high recognition accuracy, including the face unlocking of mobile devices, community access control systems, and city surveillance. As the current high accuracy is guaranteed by very deep network structures, facial images often need to be transmitted to third-party servers with high computational power for inference. However, facial images visually reveal the user’s identity information. In this process, both untrusted service providers and malicious users can significantly increase the risk of a personal privacy breach. Current privacy-preserving approaches to face recognition are often accompanied by many side effects, such as a significant increase in inference time or a noticeable decrease in recognition accuracy. This paper proposes a privacy-preserving face recognition method using differential privacy in the frequency domain. Due to the utilization of differential privacy, it offers a guarantee of privacy in theory. Meanwhile, the loss of accuracy is very slight. This method first converts the original image to the frequency domain and removes the direct component termed DC. Then a privacy budget allocation method can be learned based on the loss of the back-end face recognition network within the differential privacy framework. Finally, it adds the corresponding noise to the frequency domain features. Our method performs very well with several classical face recognition test sets according to the extensive experiments.



Paperid:493
Authors:Zhaodong Sun; Xiaobai Li
Title: Contrast-Phys: Unsupervised Video-Based Remote Physiological Measurement via Spatiotemporal Contrast
Abstract:
Video-based remote physiological measurement utilizes face videos to measure the blood volume change signal, which is also called remote photoplethysmography (rPPG). Supervised methods for rPPG measurements achieve state-of-the-art performance. However, supervised rPPG methods require face videos and ground truth physiological signals for model training. In this paper, we propose an unsupervised rPPG measurement method that does not require ground truth signals for training. We use a 3DCNN model to generate multiple rPPG signals from each video in different spatiotemporal locations and train the model with a contrastive loss where rPPG signals from the same video are pulled together while those from different videos are pushed away. We test on five public datasets, including RGB videos and NIR videos. The results show that our method outperforms the previous unsupervised baseline and achieves accuracies very close to the current best supervised rPPG methods on all five datasets. Furthermore, we also demonstrate that our approach can run at a much faster speed and is more robust to noises than the previous unsupervised baseline. Our code is available at https://github.com/zhaodongsun/contrast-phys.



Paperid:494
Authors:Yuchen Liu; Yabo Chen; Wenrui Dai; Mengran Gou; Chun-Ting Huang; Hongkai Xiong
Title: Source-Free Domain Adaptation with Contrastive Domain Alignment and Self-Supervised Exploration for Face Anti-Spoofing
Abstract:
Despite promising success in intra-dataset tests, existing face anti-spoofing (FAS) methods suffer from poor generalization ability under domain shift. This problem can be solved by aligning source and target data. However, due to privacy and security concerns of human faces, source data are usually inaccessible during adaptation for practical deployment, where only a pre-trained source model and unlabeled target data are available. In this paper, we propose a novel Source-free Domain Adaptation framework for Face Anti-Spoofing, namely SDA-FAS, that addresses the problems of source knowledge adaptation and target data exploration under the source-free setting. For source knowledge adaptation, we present novel strategies to realize self-training and domain alignment. We develop a contrastive domain alignment module to align conditional distribution across different domains by aggregating the features of fake and real faces separately. We demonstrate in theory that the pre-trained source model is equivalent to the source data as source prototypes for supervised contrastive learning in domain alignment. The source-oriented regularization is also introduced into self-training to alleviate the self-biasing problem. For target data exploration, self-supervised learning is employed with specified patch shuffle data augmentation to explore intrinsic spoofing features for unseen attack types. To our best knowledge, SDA-FAS is the first attempt that jointly optimizes the source-adapted knowledge and target self-supervised exploration for FAS. Extensive experiments on thirteen cross-dataset testing scenarios show that the proposed framework outperforms the state-of-the-art methods by a large margin.



Paperid:495
Authors:Yingjie Chen; Huasong Zhong; Chong Chen; Chen Shen; Jianqiang Huang; Tao Wang; Yun Liang; Qianru Sun
Title: On Mitigating Hard Clusters for Face Clustering
Abstract:
Face clustering is a promising way to scale up face recognition systems using large-scale unlabeled face images. It remains challenging to identify small or sparse face image clusters that we call hard clusters, which is caused by the heterogeneity, i.e., high variations in size and sparsity, of the clusters. Consequently, the conventional way of using a uniform threshold (to identify clusters) often leads to a terrible misclassification for the samples that should belong to hard clusters. We tackle this problem by leveraging the neighborhood information of samples and inferring the cluster memberships (of samples) in a probabilistic way. We introduce two novel modules, Neighborhood-Diffusion-based Density (NDDe) and Transition-Probability-based Distance (TPDi), based on which we can simply apply the standard Density Peak Clustering algorithm with a uniform threshold. Our experiments on multiple benchmarks show that each module contributes to the final performance of our method, and by incorporating them into other advanced face clustering methods, these two modules can boost the performance of these methods to a new state-of-the-art.



Paperid:496
Authors:Jiaheng Liu; Zhipeng Yu; Haoyu Qin; Yichao Wu; Ding Liang; Gangming Zhao; Ke Xu
Title: OneFace: One Threshold for All
Abstract:
Face recognition (FR) has witnessed remarkable progress with the surge of deep learning. Current FR evaluation protocols usually adopt different thresholds to calculate the True Accept Rate (TAR) under a pre-defined False Accept Rate (FAR) for different datasets. How- ever, in practice, when the FR model is deployed on industry systems (e.g., hardware devices), only one fixed threshold is adopted for all scenarios to distinguish whether a face image pair belongs to the same identity. Therefore, current evaluation protocols using different thresholds for different datasets are not fully compatible with the practical evaluation scenarios with one fixed threshold, and it is critical to measure the performance of FR models by using one threshold for all datasets. In this paper, we rethink the limitations of existing evaluation protocols for FR and propose to evaluate the performance of FR models from a new perspective. Specifically, in our OneFace, we first propose the One- Threshold-for-All (OTA) evaluation protocol for FR, which utilizes one fixed threshold called as Calibration Threshold to measure the performance on different datasets. Then, to improve the performance of FR models under the OTA protocol, we propose the Threshold Consistency Penalty (TCP) to improve the consistency of the thresholds among multiple domains, which includes Implicit Domain Division (IDD) as well as Calibration and Domain Thresholds Estimation (CDTE). Extensive experimental results on multiple benchmark datasets demonstrate the effectiveness of our proposed method for FR.



Paperid:497
Authors:Wanhua Li; Zhexuan Cao; Jianjiang Feng; Jie Zhou; Jiwen Lu
Title: Label2Label: A Language Modeling Framework for Multi-Attribute Learning
Abstract:
Objects are usually associated with multiple attributes, and these attributes often exhibit high correlations. Modeling complex relationships between attributes poses a great challenge for multi-attribute learning. This paper proposes a simple yet generic framework named Label2Label to exploit the complex attribute correlations. Label2Label is the first attempt for multi-attribute prediction from the perspective of language modeling. Specifically, it treats each attribute label as a ""word"" describing the sample. As each sample is annotated with multiple attribute labels, these ""words"" will naturally form an unordered but meaningful ""sentence"", which depicts the semantic information of the corresponding sample. Inspired by the remarkable success of pre-training language models in NLP, Label2Label introduces an image-conditioned masked language model, which randomly masks some of the ""word"" tokens from the label ""sentence"" and aims to recover them based on the masked ""sentence"" and the context conveyed by image features. Our intuition is that the instance-wise attribute relations are well grasped if the neural net can infer the missing attributes based on the context and the remaining attribute hints. Label2Label is conceptually simple and empirically powerful. Without incorporating task-specific prior knowledge and highly specialized network designs, our approach achieves state-of-the-art results on three different multi-attribute learning tasks, compared to highly customized domain-specific methods. Code is available at https://github.com/Li-Wanhua/Label2Label.



Paperid:498
Authors:Gee-Sern Hsu; Rui-Cang Xie; Zhi-Ting Chen; Yu-Hong Lin
Title: AgeTransGAN for Facial Age Transformation with Rectified Performance Metrics
Abstract:
We propose the AgeTransGAN for facial age transformation and the improvements to the metrics for performance evaluation. The AgeTransGAN is composed of an encoder-decoder generator and a conditional multitask discriminator with an age classifier embedded. The generator exploits cycle-generation consistency, age classification and cross-age identity consistency to disentangle the identity and age characteristics during training. The discriminator fuses age features with the target age group label and collaborates with the embedded age classifier to warrant the desired age traits made on the generated images. As many previous work use the Face++ APIs as the metrics for performance evaluation, we reveal via experiments the inappropriateness of using the Face++ as the metrics for the face verification and age estimation of juniors. To rectify the Face++ metrics, we made the Cross-Age Face (CAF) dataset which contains 4000 face images of 520 individuals taken from their childhood to seniorhood. The CAF is one of the very few datasets that offer much more images of the same individuals across large age gaps than the popular FG-Net. We use the CAF to rectify the face verification thresholds of the Face++ APIs across different age gaps. We also use the CAF and the FFHQ-Aging datasets to compare the age estimation performance of the Face++ APIs and an age estimator made by our own, and propose rectified metrics for performance evaluation. We compare the performance of the AgeTransGAN and state-of-the-art approaches by using the existing and rectified metrics.



Paperid:499
Authors:Zhihao Gu; Taiping Yao; Yang Chen; Shouhong Ding; Lizhuang Ma
Title: Hierarchical Contrastive Inconsistency Learning for Deepfake Video Detection
Abstract:
With the rapid development of Deepfake techniques, the capacity of generating hyper-realistic faces has aroused public concerns in recent years. The temporal inconsistency which derives from the contrast of facial movements between pristine and forged videos can serve as an efficient cue in identifying Deepfakes. However, most existing approaches tend to impose binary supervision to model it, which restricts them to only focusing on the category-level discrepancies. In this paper, we propose a novel Hierarchical Contrastive Inconsistency Learning framework (HCIL) with a two-level contrastive paradigm. Specially, sampling multiply snippets to form the input, HCIL performs contrastive learning from both local and global perspectives to capture more general and intrinsical temporal inconsistency between real and fake videos. Moreover, we also incorporate a region-adaptive module for intra-snippet inconsistency mining and an inter-snippet fusion module for cross-snippet information fusion, which further facilitates the inconsistency learning. Extensive experiments and visualizations demonstrate the effectiveness of our method against SOTA competitors on four Deepfake video datasets, \emph{i.e.,} FaceForensics++, Celeb-DF, DFDC, and Wild-Deepfake.



Paperid:500
Authors:Bingqi Ma; Guanglu Song; Boxiao Liu; Yu Liu
Title: Rethinking Robust Representation Learning under Fine-Grained Noisy Faces
Abstract:
Learning robust feature representation from large-scale noisy faces stands out as one of the key challenges in high-performance face recognition. Recent attempts have been made to cope with this challenge by alleviating the intra-class conflict and inter-class conflict. However, the unconstrained noise type in each conflict still makes it difficult for these algorithms to perform well. To better understand this, we reformulate the noise type of faces in each class with a more fine-grained manner as N-identities|K-clusters|C-conflicts. Different types of noisy faces can be generated by adjusting the values of N, K, and C. Based on this unified formulation, we found that the main barrier behind the noise-robust representation learning is the flexibility of the algorithm under different N, K, and C. For this potential problem, we constructively propose a new method, named Evolving Sub-centers Learning (ESL), to find optimal hyperplanes to accurately describe the latent space of massive noisy faces. More specifically, we initialize M sub-centers for each class and ESL encourages it to be automatically aligned to N-identities|K-clusters|C-conflicts faces via producing, merging, and dropping operations. Images belonging to the same identity in noisy faces can effectively converge to the same sub-center and samples with different identities will be pushed away. We inspect its effectiveness with an elaborate ablation study on synthetic noisy datasets different N, K, and C. Without any bells and whistles, ESL can achieve significant performance gains over state-of-the-art methods in large-scale noisy faces.



Paperid:501
Authors:Sungho Shin; Joosoon Lee; Junseok Lee; Yeonguk Yu; Kyoobin Lee
Title: Teaching Where to Look: Attention Similarity Knowledge Distillation for Low Resolution Face Recognition
Abstract:
Deep learning has achieved outstanding performance for face recognition benchmarks, but performance reduces significantly for low resolution (LR) images. We propose an attention similarity knowledge distillation approach, which transfers attention maps obtained from a high resolution (HR) network as a teacher into an LR network as a student to boost LR recognition performance. Inspired by humans being able to approximate an object’s region from an LR image based on prior knowledge obtained from HR images, we designed the knowledge distillation loss using the cosine similarity to make the student network’s attention resemble the teacher network’s attention. Experiments on various LR face related benchmarks confirmed the proposed method generally improved recognition performances on LR settings, outperforming state-of-the-art results by simply transferring well-constructed attention maps. The code and pretrained models are publicly available in the https://github.com/gist-ailab/teaching-where-to-look.



Paperid:502
Authors:Tohar Lukov; Na Zhao; Gim Hee Lee; Ser-Nam Lim
Title: Teaching with Soft Label Smoothing for Mitigating Noisy Labels in Facial Expressions
Abstract:
Recent studies have highlighted the problem of noisy labels in large scale in-the-wild facial expressions datasets due to the uncertainties caused by ambiguous facial expressions, low-quality facial images, and the subjectiveness of annotators. To solve the problem of noisy labels, we propose Soft Label Smoothing (SLS), which smooths out multiple high-confidence classes in the logits by assigning them a probability based on the corresponding confidence, and at the same time assigning a fixed low probability to the low-confidence classes. Specifically, we introduce what we call the Smooth Operator Framework for Teaching (SOFT), based on a mean-teacher (MT) architecture where SLS is applied over the teacher’s logits. We find that the smoothed teacher’s logit provides a beneficial supervision to the student via a consistency loss -- at 30\% noise rate, SLS leads to 15\% reduction in the error rate compared with MT. Overall, SOFT beats the state of the art at mitigating noisy labels by a significant margin for both symmetric and asymmetric noise. Our code is available at https://github.com/toharl/soft.



Paperid:503
Authors:Shuai Shen; Wanhua Li; Zheng Zhu; Yueqi Duan; Jie Zhou; Jiwen Lu
Title: Learning Dynamic Facial Radiance Fields for Few-Shot Talking Head Synthesis
Abstract:
Talking head synthesis is an emerging technology with wide applications in film dubbing, virtual avatars and online education. Recent NeRF-based methods generate more natural talking videos, as they better capture the 3D structural information of faces. However, a specific model needs to be trained for each identity with a large dataset. In this paper, we propose Dynamic Facial Radiance Fields (DFRF) for few-shot talking head synthesis, which can rapidly generalize to an unseen identity with few training data. Different from the existing NeRF-based methods which directly encode the 3D geometry and appearance of a specific person into the network, our DFRF conditions face radiance field on 2D appearance images to learn the face prior. Thus the facial radiance field can be flexibly adjusted to the new identity with few reference images. Additionally, for better modeling of the facial deformations, we propose a differentiable face warping module conditioned on audio signals to deform all reference images to the query space. Extensive experiments show that with only tens of seconds of training clip available, our proposed DFRF can synthesize natural and high-quality audio-driven talking head videos for novel identities with only 40k iterations. We highly recommend readers view our supplementary video for intuitive comparisons. Code is available in https://sstzal.github.io/DFRF/.



Paperid:504
Authors:Jiaheng Liu; Haoyu Qin; Yichao Wu; Jinyang Guo; Ding Liang; Ke Xu
Title: CoupleFace: Relation Matters for Face Recognition Distillation
Abstract:
Knowledge distillation is an effective method to im- prove the performance of a lightweight neural network (i.e., student model) by transferring the knowledge of a well- performed neural network (i.e., teacher model), which has been widely applied in many computer vision tasks, includ- ing face recognition. Nevertheless, the current face recogni- tion distillation methods usually utilize the Feature Consis- tency Distillation (FCD) (e.g., L 2 distance) on the learned embeddings extracted by the teacher and student models for each sample, which is not able to fully transfer the knowl- edge from the teacher to the student for face recognition. In this work, we observe that mutual relation knowledge between samples is also important to improve the discrim- inative ability of the learned representation of the student model, and propose an effective face recognition distilla- tion method called CoupleFace by additionally introducing the Mutual Relation Distillation (MRD) into existing distil- lation framework. Specifically, in MRD, we first propose to mine the informative mutual relations, and then intro- duce the Relation-Aware Distillation (RAD) loss to trans- fer the mutual relation knowledge of the teacher model to the student model. Extensive experimental results on multi- ple benchmark datasets demonstrate the effectiveness of our proposed CoupleFace for face recognition.



Paperid:505
Authors:Feng Liu; Minchul Kim; Anil Jain; Xiaoming Liu
Title: Controllable and Guided Face Synthesis for Unconstrained Face Recognition
Abstract:
Although significant advances have been made in face recognition (FR), FR in unconstrained environments remains challenging due to the domain gap between the semi-constrained training datasets and unconstrained testing scenarios. To address this problem, we propose a controllable face synthesis model (CFSM) that can mimic the distribution of target datasets in a style latent space. CFSM learns a linear subspace with orthogonal bases in the style latent space with precise control over the diversity and degree of synthesis. Furthermore, the pre-trained synthesis model can be guided by the FR model, making the resulting images more beneficial for FR model training. Besides, target dataset distributions are characterized by the learned orthogonal bases, which can be utilized to measure the distributional similarity among face datasets. Our approach yields significant performance gains on unconstrained benchmarks, such as IJB-B, IJB-C, TinyFace and IJB-S (+5.76% Rank1).



Paperid:506
Authors:Manyuan Zhang; Guanglu Song; Yu Liu; Hongsheng Li
Title: Towards Robust Face Recognition with Comprehensive Search
Abstract:
Data cleaning, architecture, and loss function design are important factors contributing to high-performance face recognition. Previously, the research community tries to improve the performance of each single aspect but failed to present a unified solution on the joint search of the optimal designs for all three aspects.In this paper, we for the first time identify that these aspects are tightly coupled to each other. Optimizing the design of each aspect actually greatly limits the performance and biases the algorithmic design.Specifically, we find that the optimal model architecture or loss function is closely coupled with the data cleaning. To eliminate the bias of single-aspect research and provide an overall understanding of the face recognition model design, we first carefully design the search space for each aspect, then a comprehensive search method is introduced to jointly search optimal data cleaning, architecture, and loss function design.In our framework, we make the proposed comprehensive search as flexible as possible, by using an innovative reinforcement learning based approach.Extensive experiments on million-level face recognition benchmarks demonstrate the effectiveness of our newly-designed search space for each aspect and the comprehensive search. We outperform expert algorithms developed for each single research track by large margins. More importantly, we analyze the difference between our searched optimal design and the independent design of the single factors. We point out that strong models tend to optimize with more difficult training datasets and loss functions. Our empirical study can provide guidance in future research towards more robust face recognition systems.



Paperid:507
Authors:Zhiwen Cao; Dongfang Liu; Qifan Wang; Yingjie Chen
Title: Towards Unbiased Label Distribution Learning for Facial Pose Estimation Using Anisotropic Spherical Gaussian
Abstract:
Facial pose estimation refers to the task of predicting face orientation from a single RGB image. It is an important research topic with a wide range of applications in computer vision. Label distribution learning (LDL) based methods have been recently proposed for facial pose estimation, which achieve promising results. However, there are two major issues in existing LDL methods. First, the expectations of label distributions are biased, leading to a biased pose estimation. Second, fixed distribution parameters are applied for all learning samples, severely limiting the model capability. In this paper, we propose an Anisotropic Spherical Gaussian (ASG)-based LDL approach for facial pose estimation. In particular, our approach adopts the spherical Gaussian distribution on a unit sphere which constantly generates unbiased expectation. Meanwhile, we introduce a new loss function that allows the network to learn the distribution parameter for each learning sample flexibly. Extensive experimental results show that our method sets new state-of-the-art records on AFLW2000 and BIWI datasets.



Paperid:508
Authors:Chenyi Kuang; Zijun Cui; Jeffrey O. Kephart; Qiang Ji
Title: AU-Aware 3D Face Reconstruction through Personalized AU-Specific Blendshape Learning
Abstract:
3D face reconstruction and facial action unit (AU) detection have emerged as interesting and challenging tasks in recent years, but are rarely performed in tandem. Image-based 3D face reconstruction, which can represent a dense space of facial motions, is typically accomplished by estimating identity, expression, texture, head pose, and illumination separately via pre-constructed 3D morphable models (3DMMs). Recent 3D reconstruction models can recover high-quality geometric facial details like wrinkles and pores, but are still limited in their ability to recover 3D subtle motions caused by the activation of AUs. We present a multi-stage learning framework that recovers AU-interpretable 3D facial details by learning personalized AU-specific blendshapes from images. Our model explicitly learns 3D expression basis by using AU labels and generic AU relationship prior and then constrains the basis coefficients such that they are semantically mapped to each AU. Our AU-aware 3D reconstruction model generates accurate 3D expressions composed by semantically meaningful AU motion components. Furthermore, the output of the model can be directly applied to generate 3D AU occurrence predictions, which have not been fully explored by prior 3D reconstruction models. We demonstrate the effectiveness of our approach via qualitative and quantitative evaluations.



Paperid:509
Authors:Kai Zhao; Lei Shen; Yingyi Zhang; Chuhan Zhou; Tao Wang; Ruixin Zhang; Shouhong Ding; Wei Jia; Wei Shen
Title: BézierPalm: A Free Lunch for Palmprint Recognition
Abstract:
Palmprints are private and stable information for biometric recognition. In the deep learning era, the development of palmprint recognition is limited by the lack of sufficient training data. In this paper, by observing that palmar creases are the key information to deep-learning-based palmprint recognition, we propose to synthesize training data by manipulating palmar creases. Concretely, we introduce an intuitive geometric model which represents palmar creases with parameterized Bézier curves. By randomly sampling Bézier parameters, we can synthesize massive training samples of diverse identities, which enables us to pretrain large-scale palmprint recognition models Experimental results demonstrate that such synthetically pretrained models have a very strong generalization ability: they can be efficiently transferred to real datasets, leading to significant performance improvements on palm print recognition. For example, under the open-set protocol, our method improves the strong ArcFace baseline by more than 10% in terms of TAR@1e-6. And under the closed-set protocol, our method reduces the equal error rate (EER) by an order of magnitude. The code will be made openly available upon acceptance.



Paperid:510
Authors:Hsin-Ping Huang; Deqing Sun; Yaojie Liu; Wen-Sheng Chu; Taihong Xiao; Jinwei Yuan; Hartwig Adam; Ming-Hsuan Yang
Title: Adaptive Transformers for Robust Few-Shot Cross-Domain Face Anti-Spoofing
Abstract:
While recent face anti-spoofing methods perform well under the intra-domain setups, an effective approach needs to account for much larger appearance variations of images acquired in complex scenes with different sensors for robust performance. In this paper, we present adaptive vision transformers (ViT) for robust cross-domain face anti-spoofing. Specifically, we adopt ViT as a backbone to exploit its strength to account for long-range dependencies among pixels. We further introduce the ensemble adapters module and feature-wise transformation layers in the ViT to adapt to different domains for robust performance with a few samples. Experiments on several benchmark datasets show that the proposed models achieve both robust and competitive performance against the state-of-the-art methods.



Paperid:511
Authors:Kewei Yang; Kang Chen; Daoliang Guo; Song-Hai Zhang; Yuan-Chen Guo; Weidong Zhang
Title: Face2Face$^\rho$: Real-Time High-Resolution One-Shot Face Reenactment
Abstract:
Existing one-shot face reenactment methods either present obvious artifacts in large pose transformations, or cannot well-preserve the identity information in the source images, or fail to meet the requirements of real-time applications due to the intensive amount of computation involved. In this paper, we introduce Face2Face^ρ, the first Real-time High-resolution and One-shot (RHO, ρ) face reenactment framework. To achieve this goal, we designed a new 3DMM-assisted warping-based face reenactment architecture which consists of two fast and efficient sub-networks, i.e., a u-shaped rendering network to reenact faces driven by head poses and facial motion fields, and a hierarchical coarse-to-fine motion network to predict facial motion fields guided by different scales of landmark images. Compared with existing state-of-the-art works, Face2Face^ρ can produce results of equal or better visual quality, yet with significantly less time and memory overhead. We also demonstrate that Face2Face^ρ can achieve real-time performance for face images of 1440×1440 resolution with a desktop GPU and 256×256 resolution with a mobile CPU.



Paperid:512
Authors:Haiwen Feng; Timo Bolkart; Joachim Tesch; Michael J. Black; Victoria Abrevaya
Title: Towards Racially Unbiased Skin Tone Estimation via Scene Disambiguation
Abstract:
Virtual facial avatars will play an increasingly important role in immersive communication, games and the metaverse, and it is therefore critical that they be inclusive. This requires accurate recovery of the albedo, regardless of age, sex, or ethnicity. While significant progress has been made on estimating 3D facial geometry, appearance estimation has received less attention. The task is fundamentally ambiguous because the observed color is a function of albedo and lighting, both of which are unknown. We find that current methods are biased towards light skin tones due to (1) strongly biased priors that prefer lighter pigmentation and (2) algorithmic solutions that disregard the light/albedo ambiguity. To address this, we propose a new evaluation dataset (FAIR) and an algorithm (TRUST) to improve albedo estimation and, hence, fairness. Specifically, we create the first facial albedo evaluation benchmark where subjects are balanced in terms of skin color, and measure accuracy using the Individual Typology Angle (ITA) metric. We then address the light/albedo ambiguity by building on a key observation: the image of the full scene -as opposed to a cropped image of the face- contains important information about lighting that can be used for disambiguation. TRUST regresses facial albedo by conditioning on both the face region and a global illumination signal obtained from the scene image. Our experimental results show significant improvement compared to state-of-the-art methods on albedo estimation, both in terms of accuracy and fairness. The evaluation benchmark and code are available for research purposes at https://trust.is.tue.mpg.de.



Paperid:513
Authors:Shijie Wu; Xun Gong
Title: BoundaryFace: A Mining Framework with Noise Label Self-Correction for Face Recognition
Abstract:
Face recognition has made tremendous progress in recent years due to the advances in loss functions and the explosive growth in training sets size. A properly designed loss is seen as key to extract discriminative features for classification. Several margin-based losses have been proposed as alternatives of softmax loss in face recognition. However, two issues remain to consider: 1) They overlook the importance of hard sample mining for discriminative learning. 2) Label noise ubiquitously exists in large-scale datasets, which can seriously damage the model’s performance. In this paper, starting from the perspective of decision boundary, we propose a novel mining framework that focuses on the relationship between a sample’s ground truth class center and its nearest negative class center. Specifically, a closed-set noise label self-correction module is put forward, making this framework work well on datasets containing a lot of label noise. The proposed method consistently outperforms SOTA methods in various face recognition benchmarks. Training code has been released at https://gitee.com/swjtugx/classmate/tree/master/OurGroup/BoundaryFace.



Paperid:514
Authors:Adrian Bulat; Shiyang Cheng; Jing Yang; Andrew Garbett; Enrique Sanchez; Georgios Tzimiropoulos
Title: Pre-training Strategies and Datasets for Facial Representation Learning
Abstract:
What is the best way to learn a universal face representation? Recent work on Deep Learning in the area of face analysis has focused on supervised learning for specific tasks of interest (e.g. face recognition, facial landmark localization etc.) but has overlooked the overarching question of how to find a facial representation that can be readily adapted to several facial analysis tasks and datasets. To this end, we make the following 4 contributions: (a) we introduce, for the first time, a comprehensive evaluation benchmark for facial representation learning consisting of 5 important face analysis tasks. (b) We systematically investigate two ways of large-scale representation learning applied to faces: supervised and unsupervised pre-training. Importantly, we focus our evaluations on the case of few-shot facial learning. (c) We investigate important properties of the training datasets including their size and quality (labelled, unlabelled or even uncurated). (d) To draw our conclusions, we conducted a very large number of experiments. Our main two findings are: (1) Unsupervised pre-training on completely in-the-wild, uncurated data provides consistent and, in some cases, significant accuracy improvements for all facial tasks considered. (2) Many existing facial video datasets seem to have a large amount of redundancy. We will release code, pre-trained models and data to facilitate future research.



Paperid:515
Authors:Isaac Kasahara; Simon Stent; Hyun Soo Park
Title: Look Both Ways: Self-Supervising Driver Gaze Estimation and Road Scene Saliency
Abstract:
We present a new on-road driving dataset, called “Look Both Ways”, which contains synchronized video of both driver faces and the forward road scene, along with ground truth gaze data registered from eye tracking glasses worn by the drivers. Our dataset supports the study of methods for non-intrusively estimating a driver’s focus of attention while driving - an important application area in road safety. A key challenge is that this task requires accurate gaze estimation, but supervised appearance-based gaze estimation methods often do not transfer well to real driving datasets, and in-domain ground truth to supervise them is difficult to gather. We therefore propose a method for self-supervision of driver gaze, by taking advantage of the geometric consistency between the driver’s gaze direction and the saliency of the scene as observed by the driver. We formulate a 3D geometric learning framework to enforce this consistency, allowing the gaze model to supervise the scene saliency model, and vice versa. We implement a prototype of our method and test it with our dataset, to show that compared to a supervised approach it can yield better gaze estimation and scene saliency estimation with no additional labels.



Paperid:516
Authors:Sanghyeon Na
Title: MFIM: Megapixel Facial Identity Manipulation
Abstract:
Face swapping is a task that changes a facial identity of a given image to that of another person. In this work, we propose a novel face-swapping framework called Megapixel Facial Identity Manipulation (MFIM). The face-swapping model should achieve two goals. First, it should be able to generate a high-quality image. We argue that a model which is proficient in generating a megapixel image can achieve this goal. However, generating a megapixel image is generally difficult without careful model design. Therefore, our model exploits pretrained StyleGAN in the manner of GAN-inversion to effectively generate a megapixel image. Second, it should be able to effectively transform the identity of a given image. Specifically, it should be able to actively transform ID attributes (e.g., face shape and eyes) of a given image into those of another person, while preserving ID-irrelevant attributes (e.g., pose and expression). To achieve this goal, we exploit 3DMM that can capture various facial attributes. Specifically, we explicitly supervise our model to generate a face-swapped image with the desirable attributes using 3DMM. We show that our model achieves state-of-the-art performance through extensive experiments. Furthermore, we propose a new operation called ID mixing, which creates a new identity by semantically mixing the identities of several people. It allows the user to customize the new identity.



Paperid:517
Authors:Erroll Wood; Tadas Baltrušaitis; Charlie Hewitt; Matthew Johnson; Jingjing Shen; Nikola Milosavljević; Daniel Wilde; Stephan Garbin; Toby Sharp; Ivan Stojiljković; Tom Cashman; Julien Valentin
Title: 3D Face Reconstruction with Dense Landmarks
Abstract:
Landmarks often play a key role in face analysis, but many aspects of identity or expression cannot be represented by sparse landmarks alone. Thus, in order to reconstruct faces more accurately, landmarks are often combined with additional signals like depth images or techniques like differentiable rendering. Can we keep things simple by just using more landmarks? In answer, we present the first method that accurately predicts 10x as many landmarks as usual, covering the whole head, including the eyes and teeth. This is accomplished using synthetic training data, which guarantees perfect landmark annotations. By fitting a morphable model to these dense landmarks, we achieve state-of-the-art results for monocular 3D face reconstruction in the wild. We show that dense landmarks are an ideal signal for integrating face shape information across frames by demonstrating accurate and expressive facial performance capture in both monocular and multi-view scenarios. This approach is also highly efficient: we can predict dense landmarks and fit our 3D face model at over 150FPS on a single CPU thread. Please see our website: https://microsoft.github.io/DenseLandmarks/.



Paperid:518
Authors:Daeha Kim; Byung Cheol Song
Title: Emotion-Aware Multi-View Contrastive Learning for Facial Emotion Recognition
Abstract:
When a person recognizes another’s emotion, he or she recognizes the (facial) features associated with emotional expression. So, for a machine to recognize facial emotion(s), the features related to emotional expression must be represented and described properly. However, prior arts based on label supervision not only failed to explicitly capture features related to emotional expression, but also were not interested in learning emotional representations. This paper proposes a novel approach to generate features related to emotional expression through feature transformation and to use them for emotional representation learning. Specifically, the contrast between the generated features and overall facial features is quantified through contrastive representation learning, and then facial emotions are recognized based on understanding of angle and intensity that describe the emotional representation in the polar coordinate, i.e., the Arousal-Valence space. Experimental results show that the proposed method improves the PCC/CCC performance by more than 10% compared to the runner-up method in the wild datasets and is also qualitatively better in terms of neural activation map.



Paperid:519
Authors:Seon-Ho Lee; Chang-Su Kim
Title: Order Learning Using Partially Ordered Data via Chainization
Abstract:
We propose the chainization algorithm for effective order learning when only partially ordered data are available. First, we develop a binary comparator to predict missing ordering relations between instances. Then, by extending the Kahn’s algorithm, we form a chain representing a linear ordering of instances. We fine-tune the comparator over pseudo pairs, which are sampled from the chain, and then re-estimate the linear ordering alternately. As a result, we obtain a more reliable comparator and a more meaningful linear ordering. Experimental results show that the proposed algorithm yields excellent rank estimation performances under various weak supervision scenarios, including semi-supervised learning, domain adaptation, and bipartite cases. The source codes are available at https://github.com/seon92/Chainization"



Paperid:520
Authors:Ron Slossberg; Ibrahim Jubran; Ron Kimmel
Title: Unsupervised High-Fidelity Facial Texture Generation and Reconstruction
Abstract:
Many methods have been proposed over the years to tackle the task of facial 3D geometry and texture recovery from a single image. Such methods often fail to provide high-fidelity texture without relying on 3D facial scans during training. In contrast, the complementary task of 3D facial generation has not received as much attention. As opposed to the 2D texture domain, where GANs have proven to produce highly realistic facial images, the more challenging 3D domain has not yet caught up to the same levels of realism and diversity. In this paper, we propose a novel unified pipeline for both tasks, generation of texture with coupled geometry, and reconstruction of high-fidelity texture. Our texture model is learned, in an unsupervised fashion, from natural images as opposed to scanned textures. To our knowledge, this is the first such unified framework independent of scanned textures. Our novel training pipeline incorporates a pre-trained 2D facial generator coupled with a deep feature manipulation methodology. By applying our two-step geometry fitting process, we seamlessly integrate our modeled textures into synthetically generated background images forming a realistic composition of our textured model with background, hair, teeth, and body. This enables us to apply transfer learning from the 2D image domain, thus leveraging the high-quality results obtained in this domain. We provide a comprehensive study on several recent methods comparing our model in generation and reconstruction tasks. As the extensive qualitative, as well as quantitative analysis, demonstrate, we achieve state-of-the-art results for both tasks.



Paperid:521
Authors:Xiao Guo; Yaojie Liu; Anil Jain; Xiaoming Liu
Title: Multi-Domain Learning for Updating Face Anti-Spoofing Models
Abstract:
In this work, we study multi-domain learning for face anti-spoofing (MD-FAS), where a pre-trained FAS model needs to be updated to perform equally well on both source and target domains while only using target domain data for updating. We present a new model for MD-FAS, which addresses the forgetting issue when learning new domain data, while possessing a high level of adaptability. First, we devise a simple yet effective module, called spoof region estimator (SRE), to identify spoof traces in the spoof image. Such spoof traces reflect the source pre-trained model’s responses that help upgraded models combat catastrophic forgetting during updating. Unlike prior works that estimate spoof traces which generate multiple outputs or a low-resolution binary mask, SRE produces one single, detailed pixel-wise estimate in an unsupervised manner. Secondly, we propose a novel framework, named FAS-wrapper, which transfers knowledge from the pre-trained models and seamlessly integrates with different FAS models. Lastly, to help the community further advance MD-FAS, we construct a new benchmark based on SIW, SIW-Mv2 and Oulu-NPU, and introduce four distinct protocols for evaluation, where source and target domains are different in terms of spoof type, age, ethnicity, and illumination. Our proposed method achieves superior performance on the MD-FAS benchmark than previous methods.



Paperid:522
Authors:Wojciech Zielonka; Timo Bolkart; Justus Thies
Title: Towards Metrical Reconstruction of Human Faces
Abstract:
Face reconstruction and tracking is a building block of numerous applications in AR/VR, human-machine interaction, as well as medical applications. Most of these applications rely on a metrically correct prediction of the shape, especially, when the reconstructed subject is put into a metrical context (i.e., when there is a reference object of known size). A metrical reconstruction is also needed for any application that measures distances and dimensions of the subject (e.g., to virtually fit a glasses frame). State-of-the-art methods for face reconstruction from a single image are trained on large 2D image datasets in a self-supervised fashion. However, due to the nature of a perspective projection they are not able to reconstruct the actual face dimensions, and even predicting the average human face outperforms some of these methods in a metrical sense. To learn the actual shape of a face, we argue for a supervised training scheme. Since there exists no large-scale 3D dataset for this task, we annotated and unified small- and medium-scale databases. The resulting unified dataset is still a medium-scale dataset with more than 2k identities and training purely on it would lead to overfitting. To this end, we take advantage of a face recognition network pretrained on a large-scale 2D image dataset, which provides distinct features for different faces and is robust to expression, illumination, and camera changes. Using these features, we train our face shape estimator in a supervised fashion, inheriting the robustness and generalization of the face recognition network. Our method, which we call \modellong, outperforms the state-of-the-art reconstruction methods by a large margin, both on current non-metric benchmarks as well as on our metric benchmarks (15% and 24% lower average error on NoW, respectively).



Paperid:523
Authors:Zhiheng Li; Anthony Hoogs; Chenliang Xu
Title: Discover and Mitigate Unknown Biases with Debiasing Alternate Networks
Abstract:
Deep image classifiers have been found to learn biases from datasets. To mitigate the biases, most previous methods require labels of protected attributes (e.g., age, skin tone) as full-supervision, which has two limitations: 1) it is infeasible when the labels are unavailable; 2) they are incapable of mitigating unknown biases---biases that humans do not preconceive. To resolve those problems, we propose Debiasing Alternate Networks (DebiAN), which comprises two networks---a Discoverer and a Classifier. By training in an alternate manner, the discoverer tries to find multiple unknown biases of the classifier without any annotations of biases, and the classifier aims at unlearning the biases identified by the discoverer. While previous works evaluate debiasing results in terms of a single bias, we create Multi-Color MNIST dataset to better benchmark mitigation of multiple biases in a multi-bias setting, which not only reveals the problems in previous methods but also demonstrates the advantage of DebiAN in identifying and mitigating multiple biases simultaneously. We further conduct extensive experiments on real-world datasets, showing that the discoverer in DebiAN can identify unknown biases that may be hard to be found by humans. Regarding debiasing, DebiAN achieves strong bias mitigation performance.



Paperid:524
Authors:Alexandra Chouldechova; Siqi Deng; Yongxin Wang; Wei Xia; Pietro Perona
Title: Unsupervised and Semi-Supervised Bias Benchmarking in Face Recognition
Abstract:
We introduce Semi-supervised Performance Evaluation for Face Recognition (SPE-FR). SPE-FR is a statistical method for evaluating the performance and algorithmic bias of face verification systems when identity labels are unavailable or incomplete. The method is based on parametric Bayesian modeling of the face embedding similarity scores. SPE-FR produces point estimates, performance curves, and confidence bands that reflect uncertainty in the estimation procedure. Focusing on the unsupervised setting wherein no identity labels are available, we validate our method through experiments on a wide range of face embedding models and two publicly available evaluation datasets. Experiments show that SPE-FR can accurately assess performance on data with no identity labels, and confidently reveal demographic biases in system performance.



Paperid:525
Authors:Boxi Wu; Jindong Gu; Zhifeng Li; Deng Cai; Xiaofei He; Wei Liu
Title: Towards Efficient Adversarial Training on Vision Transformers
Abstract:
Vision Transformer (ViT), as a powerful alternative to Convolutional Neural Network (CNN), has received much attention. Recent work showed that ViTs are also vulnerable to adversarial examples like CNNs. To build robust ViTs, an intuitive way is to apply adversarial training since it has been shown as one of the most effective ways to accomplish robust CNNs. However, one major limitation of adversarial training is its heavy computational cost. The self-attention mechanism adopted by ViTs is a computationally intense operation whose expense increases quadratically with the number of input patches, making adversarial training on ViTs even more time-consuming. In this work, we first comprehensively study fast adversarial training on a variety of vision transformers and illustrate the relationship between the efficiency and robustness. Then, to expediate adversarial training on ViTs, we propose an efficient Attention Guided Adversarial Training mechanism. Specifically, relying on the specialty of self-attention, we actively remove certain patch embeddings of each layer with an attention-guided dropping strategy during adversarial training. The slimmed self-attention modules accelerate the adversarial training on ViTs significantly. With only 65\% of the fast adversarial training time, we match the state-of-the-art results on the challenging ImageNet benchmark.



Paperid:526
Authors:Pradyumna Chari; Yunhao Ba; Shreeram Athreya; Achuta Kadambi
Title: MIME: Minority Inclusion for Majority Group Enhancement of AI Performance
Abstract:
Several papers have rightly included minority groups in artificial intelligence (AI) training data to improve test inference for minority groups and/or society-at-large. A society-at-large consists of both minority and majority stakeholders. A common misconception is that minority inclusion does not increase performance for majority groups alone. In this paper, we make the surprising finding that including minority samples can improve test error for the majority group. In other words, minority group inclusion leads to majority group enhancements (MIME) in performance. A theoretical existence proof of the MIME effect is presented and found to be consistent with experimental results on six different datasets. Project webpage: https://visual.ee.ucla.edu/mime.htm/.



Paperid:527
Authors:Vongani H. Maluleke; Neerja Thakkar; Tim Brooks; Ethan Weber; Trevor Darrell; Alexei A. Efros; Angjoo Kanazawa; Devin Guillory
Title: Studying Bias in GANs through the Lens of Race
Abstract:
In this work, we study how the performance and evaluation of generative image models are impacted by the racial composition of the datasets upon which these models are trained. By examining and controlling the racial distributions in various training datasets, we are able to observe the impacts of different training distributions on generated image quality and the racial distributions of the generated images. Our results show that the racial compositions of generated images successfully preserve that of the training data. However, we observe that truncation, a technique used to generate higher quality images during inference, exacerbates racial imbalances in the data. Lastly, when examining the relationship between image quality and race, we find that the highest perceived visual quality images of a given race come from a distribution where that race is well-represented, and that annotators consistently prefer generated white faces over Black faces.



Paperid:528
Authors:Ailin Deng; Shen Li; Miao Xiong; Zhirui Chen; Bryan Hooi
Title: "Trust, but Verify: Using Self-Supervised Probing to Improve Trustworthiness"
Abstract:
Trustworthy machine learning is of primary importance to the practical deployment of deep learning models. While state-of-the-art models achieve astonishingly good performance in terms of accuracy, recent literature reveals that their predictive confidence scores unfortunately cannot be trusted: e.g., they are often overconfident when wrong predictions are made, or so even for obvious outliers. In this paper, we introduce a new approach of \emph{self-supervised probing}, which enables us to check and mitigate the overconfidence issue for a trained model, thereby improving its trustworthiness. We provide a simple yet effective framework, which can be flexibly applied to existing trustworthiness-related methods in a plug-and-play manner. Extensive experiments on three trustworthiness-related tasks (misclassification detection, calibration and out-of-distribution detection) across various benchmarks verify the effectiveness of our proposed probing framework.



Paperid:529
Authors:Ayush Chopra; Abhinav Java; Abhishek Singh; Vivek Sharma; Ramesh Raskar
Title: Learning to Censor by Noisy Sampling
Abstract:
Point clouds are an increasingly ubiquitous input modality and the raw signal can be efficiently processed with recent progress in deep learning. This signal may, often inadvertently, capture sensitive information that can leak semantic and geometric properties of the scene which the data owner does not want to share. The goal of this work is to protect sensitive information when learning from point clouds; by censoring signal before the point cloud is released for downstream tasks. Specifically, we focus on preserving utility for perception tasks while mitigating attribute leakage attacks. The key motivating insight is to leverage the localized saliency of perception tasks on point clouds to provide good privacy-utility trade-offs. We realize this through a mechanism called censoring by noisy sampling (CBNS), which is composed of two modules: i) Invariant Sampling: a differentiable point-cloud sampler which learns to remove points invariant to utility and ii) Noise Distortion: which learns to distort sampled points to decouple the sensitive information from utility, and mitigate privacy leakage. We validate the effectiveness of CBNS through extensive comparisons with state-of-the-art baselines and sensitivity analyses of key design choices. Results show that CBNS achieves superior privacy-utility trade-offs.



Paperid:530
Authors:Tong Wang; Yuan Yao; Feng Xu; Shengwei An; Hanghang Tong; Ting Wang
Title: An Invisible Black-Box Backdoor Attack through Frequency Domain
Abstract:
Backdoor attacks have been shown to be a serious threat against deep learning systems such as biometric authentication and autonomous driving. An effective backdoor attack could enforce the model misbehave under certain predefined conditions, i.e., triggers, but behave normally otherwise. The triggers of existing attacks are mainly injected in the pixel space, which tend to be visually identifiable at both training and inference stages and detectable by existing defenses. In this paper, we propose a simple but effective and invisible black-box backdoor attack FTrojan through trojaning the frequency domain. The key intuition is that triggering perturbations in the frequency domain correspond to small pixel-wise perturbations dispersed across the entire image, breaking the underlying assumptions of existing defenses and making the poisoning images visually indistinguishable from clean ones. Extensive experimental evaluations show that FTrojan is highly effective and the poisoning images retain high perceptual quality. Moreover, we show that FTrojan can robustly elude or significantly degenerate the performance of existing defenses.



Paperid:531
Authors:Xiaofeng Lin; Seungbae Kim; Jungseock Joo
Title: FairGRAPE: Fairness-Aware GRAdient Pruning mEthod for Face Attribute Classification
Abstract:
Existing pruning techniques preserve deep neural networks’ overall ability to make correct predictions but could also amplify hidden biases during the compression process. We propose a novel pruning method, Fairness-aware GRAdient Pruning mEthod (FairGRAPE), that minimizes the disproportionate impacts of pruning on different sub-groups. Our method calculates the per-group importance of each model weight and selects a subset of weights that maintain the relative between-group total importance in pruning. The proposed method then prunes network edges with small importance values and repeats the procedure by updating importance values. We demonstrate the effectiveness of our method on four different datasets, FairFace, UTKFace, CelebA, and ImageNet, for the tasks of face attribute classification where our method reduces the disparity in performance degradation by up to 90% compared to the state-of-the-art pruning algorithms. Our method is substantially more effective in a setting with a high pruning rate (99%). The code and dataset used in the experiments are available at https://github.com/Bernardo1998/FairGRAPE"



Paperid:532
Authors:Pravendra Singh; Pratik Mazumder; Mohammed Asad Karim
Title: Attaining Class-Level Forgetting in Pretrained Model Using Few Samples
Abstract:
In order to address real-world problems, deep learning models are jointly trained on many classes. However, in the future, some classes may become restricted due to privacy/ethical concerns, and the restricted class knowledge has to be removed from the models that have been trained on them. The available data may also be limited due to privacy/ethical concerns, and re-training the model will not be possible. We propose a novel approach to address this problem without affecting the model’s prediction power for the remaining classes. Our approach identifies the model parameters that are highly relevant to the restricted classes and removes the knowledge regarding the restricted classes from them using the limited available training data. Our approach is significantly faster and performs similar to the model re-trained on the complete data of the remaining classes.



Paperid:533
Authors:Zihang Zou; Boqing Gong; Liqiang Wang
Title: Anti-Neuron Watermarking: Protecting Personal Data against Unauthorized Neural Networks
Abstract:
We study protecting a user’s data (e.g., images in this work) against a learner’s unauthorized use in training neural networks. It is especially challenging when the user’s data is only a tiny percentage of the learner’s complete training set. We revisit the traditional watermarking under modern deep learning settings to tackle the challenge. We show that when a user watermarks images using a specialized linear color transformation, a neural network classifier will be imprinted with the signature so that a third-party arbitrator can verify the potentially unauthorized usage of the user data by inferring the watermark signature from the neural network. We also discuss what watermarking properties and signature spaces make the arbitrator’s verification convincing. To our best knowledge, this work is the first to protect an individual user’s data ownership from unauthorized use in training neural networks.



Paperid:534
Authors:Francesco Pinto; Philip H. S. Torr; Puneet K. Dokania
Title: An Impartial Take to the CNN vs Transformer Robustness Contest
Abstract:
Following the surge of popularity of Transformers in Computer Vision, several studies have attempted to determine whether they could be more robust to distribution shifts and provide better uncertainty estimates than Convolutional Neural Networks (CNNs). The almost unanimous conclusion is that they are, and it is often conjectured more or less explicitly that the reason of this supposed superiority is to be attributed to the self-attention mechanism. In this paper we perform extensive empirical analyses showing that recent state-of-the-art CNNs (particularly, ConvNeXt) can be as robust and reliable or even sometimes more than the current state-of-the-art Transformers. However, there is no clear winner. Therefore, although it is tempting to state the definitive superiority of one family of architectures over another, they seem to enjoy similar extraordinary performances on a variety of tasks while also suffering from similar vulnerabilities such as texture, background, and simplicity biases.



Paperid:535
Authors:Yanfu Zhang; Shangqian Gao; Heng Huang
Title: Recover Fair Deep Classification Models via Altering Pre-trained Structure
Abstract:
There have been growing interest in algorithmic fairness for biased data. Although various pre-, in-, and post-processing methods are designed to address this problem, new learning paradigms designed for fair deep models are still necessary. Modern computer vision tasks usually involve large generic models and fine-tuning concerning a specific task. Training modern deep models from scratch is expensive considering the enormous training data and the complicated structures. The recently emerged intra-processing methods are designed to debias pre-trained large models. However, existing techniques stress fine-tuning more, but the deep network structure is less leveraged. This paper proposes a novel intra-processing method to improve model fairness by altering the deep network structure. We find that the unfairness of deep models are usually caused by a small portion of sub-modules, which can be uncovered using the proposed differential framework. We can further employ several strategies to modify the corrupted sub-modules inside the unfair pre-trained structure to build a fair counterpart. We experimentally verify our findings and demonstrate that the reconstructed fair models can make fair classification and achieve superior results to the state-of-the-art baselines. We conduct extensive experiments to evaluate the different strategies. The results also show that our method has good scalability when applied to a variety of fairness measures and different data types.



Paperid:536
Authors:Abhishek Singh; Ethan Garza; Ayush Chopra; Praneeth Vepakomma; Vivek Sharma; Ramesh Raskar
Title: Decouple-and-Sample: Protecting Sensitive Information in Task Agnostic Data Release
Abstract:
We propose sanitizer, a framework for secure and task-agnostic data release. While releasing datasets continues to make a big impact in various applications of computer vision, its impact is mostly realized when data sharing is not inhibited by privacy concerns. We alleviate these concerns by sanitizing datasets in a two-stage process. First, we introduce a global decoupling stage for decomposing raw data into sensitive and non-sensitive latent representations. Secondly, we design a local sampling stage to synthetically generate sensitive information with differential privacy and merge it with non-sensitive latent features to create a useful representation while preserving the privacy. This newly formed latent information is a task-agnostic representation of the original dataset with anonymized sensitive information. While most algorithms sanitize data in a task-dependent manner, a few task-agnostic sanitization techniques sanitize data by censoring sensitive information. In this work, we show that a better privacy-utility trade-off is achieved if sensitive information can be synthesized privately. We validate the effectiveness of the sanitizer by outperforming state-of-the-art baselines on the existing benchmark tasks and demonstrating tasks that are not possible using existing techniques.



Paperid:537
Authors:Sudhakar Kumawat; Hajime Nagahara
Title: Privacy-Preserving Action Recognition via Motion Difference Quantization
Abstract:
The widespread use of smart computer vision systems in our personal spaces has led to an increased consciousness about the privacy and security risks that these systems pose. On the one hand, we want these systems to assist in our daily lives by understanding their surroundings, but on the other hand, we want them to do so without capturing any sensitive information. Towards this direction, this paper proposes a simple, yet robust privacy-preserving encoder called BDQ for the task of privacy-preserving human action recognition that is composed of three modules: Blur, Difference, and Quantization. First, the input scene is passed to the Blur module to smoothen the edges. This is followed by the Difference module to apply a pixel-wise intensity subtraction between consecutive frames to highlight motion features and suppress obvious high-level privacy attributes. Finally, the Quantization module is applied to the motion difference frames to remove the low-level privacy attributes. The BDQ parameters are optimized in an end-to-end fashion via adversarial training such that it learns to allow action recognition attributes while inhibiting privacy attributes. Our experiments on three benchmark datasets show that the proposed encoder design can achieve state-of-the-art trade-off when compared with previous works. Furthermore, we show that the trade-off achieved is at par with the DVS sensor-based event cameras. Code available at: https://github.com/suakaw/BDQ_PrivacyAR.



Paperid:538
Authors:Momchil Peychev; Anian Ruoss; Mislav Balunović; Maximilian Baader; Martin Vechev
Title: Latent Space Smoothing for Individually Fair Representations
Abstract:
Fair representation learning transforms user data into a representation that ensures fairness and utility regardless of the downstream application. However, learning individually fair representations, i.e., guaranteeing that similar individuals are treated similarly, remains challenging in high-dimensional settings such as computer vision. In this work, we introduce LASSI, the first representation learning method for certifying individual fairness of high-dimensional data. Our key insight is to leverage recent advances in generative modeling to capture the set of similar individuals in the generative latent space. This enables us to learn individually fair representations that map similar individuals close together by using adversarial training to minimize the distance between their representations. Finally, we employ randomized smoothing to provably map similar individuals close together, in turn ensuring that local robustness verification of the downstream application results in end-to-end fairness certification. Our experimental evaluation on challenging real-world image data demonstrates that our method increases certified individual fairness by up to 90% without significantly affecting task utility.



Paperid:539
Authors:Christian Tomani; Daniel Cremers; Florian Buettner
Title: Parameterized Temperature Scaling for Boosting the Expressive Power in Post-Hoc Uncertainty Calibration
Abstract:
We address the problem of uncertainty calibration and introduce a novel calibration method, Parametrized Temperature Scaling (PTS). Standard deep neural networks typically yield uncalibrated predictions, which can be transformed into calibrated confidence scores using post-hoc calibration methods. In this contribution, we demonstrate that the performance of accuracy-preserving state-of-the-art post-hoc calibrators is limited by their intrinsic expressive power. We generalize temperature scaling by computing prediction-specific temperatures, parameterized by a neural network. We show with extensive experiments that our novel accuracy-preserving approach consistently outperforms existing algorithms across a large number of model architectures, datasets and metrics.



Paperid:540
Authors:Cemre Efe Karakas; Alara Dirik; Eylül Yalçınkaya; Pinar Yanardag
Title: FairStyle: Debiasing StyleGAN2 with Style Channel Manipulations
Abstract:
Recent advances in generative adversarial networks have shown that it is possible to generate high-resolution and hyperrealistic images. However, the images produced by GANs are only as fair and representative as the datasets on which they are trained. In this paper, we propose a method for directly modifying a pre-trained StyleGAN2 model that can be used to generate a balanced set of images with respect to one (e.g., eyeglasses) or more attributes (e.g., gender and eyeglasses). Our method takes advantage of the style space of the StyleGAN2 model to perform disentangled control of the target attributes to be debiased. Our method does not require training additional models and directly debiases the GAN model, paving the way for its use in various downstream applications. Our experiments show that our method successfully debiases the GAN model within a few minutes without compromising the quality of the generated images. To promote fair generative models, we share the code and debiased models at http://catlab-team.github.io/fairstyle.



Paperid:541
Authors:Surgan Jandial; Yash Khasbage; Arghya Pal; Vineeth N Balasubramanian; Balaji Krishnamurthy
Title: Distilling the Undistillable: Learning from a Nasty Teacher
Abstract:
The inadvertent stealing of private/sensitive information using Knowledge Distillation (KD) has been getting significant attention recently and has guided subsequent defense efforts considering its critical nature. Recent work \textit{Nasty Teacher} proposed to develop teachers which can not be distilled or imitated by models attacking it. However, the promise of confidentiality offered by a nasty teacher is not well studied, and as a further step to strengthen against such loopholes, we attempt to bypass its defense and steal (or extract) information in its presence successfully. Specifically, we analyze Nasty Teacher from two different directions and subsequently leverage them carefully to develop simple yet efficient methodologies, named as HTC and SCM, which increase the learning from Nasty Teacher by upto 68.63% on standard datasets. Additionally, we also explore an improvised defense method based on our insights of stealing. Our detailed set of experiments and ablations on diverse models/settings demonstrate the efficacy of our approach.



Paperid:542
Authors:Victor Escorcia; Ricardo Guerrero; Xiatian Zhu; Brais Martinez
Title: SOS! Self-Supervised Learning over Sets of Handled Objects in Egocentric Action Recognition
Abstract:
Learning an egocentric action recognition model from video data is challenging due to distractors in the background, e.g., irrelevant objects. Further integrating object information into an action model is hence beneficial. Existing methods often leverage a generic object detector to identify and represent the objects in the scene. However, several important issues remain. Object class annotations of good quality for the target domain (dataset) are still required for learning good object representation. Moreover, previous methods deeply couple existing action models with object representations, and thus need to retrain them jointly, leading to costly and inflexible integration. To overcome both limitations, we introduce Self-supervised learning Over Sets (SOS), an approach to pre-train a generic Objects In Contact (OIC) representation model from video object regions detected by an off-the-shelf hand-object contact detector. Instead of augmenting object regions individually as in conventional self-supervised learning, we view the action process as a means of natural data transformations with unique spatiotemporal continuity and exploit the inherent relationships among per-video object sets. Extensive experiments on two datasets, EPIC-KITCHENS-100 and EGTEA, show that our OIC significantly boosts the performance of multiple state-of-the-art video classification models.



Paperid:543
Authors:Miao Liu; Lingni Ma; Kiran Somasundaram; Yin Li; Kristen Grauman; James M. Rehg; Chao Li
Title: Egocentric Activity Recognition and Localization on a 3D Map
Abstract:
Given a video captured from a first person perspective and the environment context of where the video is recorded, can we recognize what the person is doing and identify where the action occurs in the 3D space? We address this challenging problem of jointly recognizing and localizing actions of a mobile user on a known 3D map from egocentric videos. To this end, we propose a novel deep probabilistic model. Our model takes the inputs of a Hierarchical Volumetric Representation (HVR) of the 3D environment and an egocentric video, infers the 3D action location as a latent variable, and recognizes the action based on the video and contextual cues surrounding its potential locations. To evaluate our model, we conduct extensive experiments on the subset of Ego4D dataset, in which both human naturalistic actions and photo-realistic 3D environment reconstructions are captured. Our method demonstrates strong results on both action recognition and 3D action localization across seen and unseen environments. We believe our work points to an exciting research direction in the intersection of egocentric vision, and 3D scene understanding.



Paperid:544
Authors:Wenqi Jia; Miao Liu; James M. Rehg
Title: Generative Adversarial Network for Future Hand Segmentation from Egocentric Video
Abstract:
We introduce the novel problem of anticipating a time series of future hand masks from egocentric video. A key challenge is to model the stochasticity of future head motions, which globally impact the head-worn camera video analysis. To this end, we propose a novel deep generative model -- EgoGAN, which uses a 3D Fully Convolutional Network to learn a spatio-temporal video representation for pixel-wise visual anticipation, generates future head motion using Generative Adversarial Network (GAN), and then predicts the future hand masks based on video representation and generated future head motion. We evaluate our method on both the EGTEA Gaze+ and the EPIC-Kitchens datasets. We conduct detailed ablation studies to validate the design choices of our approach. Furthermore, we compare our method with previous state-of-the-art methods on future image segmentation and show that our method can more accurately predict future hand masks.



Paperid:545
Authors:Siddhant Bansal; Chetan Arora; C.V. Jawahar
Title: My View Is the Best View: Procedure Learning from Egocentric Videos
Abstract:
Procedure learning involves identifying the key-steps and determining their logical order to perform a task. Existing approaches commonly use third-person videos for learning the procedure, making the manipulated object small in appearance and often occluded by the actor, leading to significant errors. In contrast, we observe that videos obtained from first-person (egocentric) wearable cameras provide an unobstructed and clear view of the action. However, procedure learning from egocentric videos is challenging because (a) the camera view undergoes extreme changes due to the wearer’s head motion, and (b) the presence of unrelated frames due to the unconstrained nature of the videos. Due to this, current state-of-the-art methods’ assumptions that the actions occur at approximately the same time and are of the same duration, do not hold. Instead, we propose to use the signal provided by the temporal correspondences between key-steps across videos. To this end, we present a novel self-supervised Correspond and Cut (CnC) framework for procedure learning. CnC identifies and utilizes the temporal correspondences between the key-steps across multiple videos to learn the procedure. Our experiments show that CnC outperforms the state-of-the-art on the benchmark ProceL and CrossTask datasets by 5.2% and 6.3%, respectively. Furthermore, for procedure learning using egocentric videos, we propose the EgoProceL dataset consisting of 62 hours of videos captured by 130 subjects performing 16 tasks. The source code and the dataset are available on the project page https://sid2697.github.io/egoprocel/.



Paperid:546
Authors:Yang Zheng; Yanchao Yang; Kaichun Mo; Jiaman Li; Tao Yu; Yebin Liu; Karen Liu; Leonidas J. Guibas
Title: GIMO: Gaze-Informed Human Motion Prediction in Context
Abstract:
Predicting human motion is critical for assistive robots and AR/VR applications, where the interaction with humans needs to be safe and comfortable. Meanwhile, an accurate prediction depends on understanding both the scene context and human intentions. Even though many works study scene-aware human motion prediction, the latter is largely underexplored due to the lack of ego-centric views that disclose human intent and the limited diversity in motion and scenes. To reduce the gap, we propose a large-scale human motion dataset that delivers high-quality body pose sequences, scene scans, as well as ego-centric views with the eye gaze that serves as a surrogate for inferring human intent. By employing inertial sensors for motion capture, our data collection is not tied to specific scenes, which further boosts the motion dynamics observed from our subjects. We perform an extensive study of the benefits of leveraging the eye gaze for ego-centric human motion prediction with various state-of-the-art architectures. Moreover, to realize the full potential of the gaze, we propose a novel network architecture that enables bidirectional communication between the gaze and motion branches. Our network achieves the top performance in human motion prediction on the proposed dataset, thanks to the intent information from eye gaze and the denoised gaze feature modulated by the motion. Code and data can be found at https://github.com/y-zheng18/GIMO.



Paperid:547
Authors:Hila Chefer; Sagie Benaim; Roni Paiss; Lior Wolf
Title: Image-Based CLIP-Guided Essence Transfer
Abstract:
We make the distinction between (i) style transfer, in which a source image is manipulated to match the textures and colors of a target image, and (ii) essence transfer, in which one edits the source image to include high-level semantic attributes from the target. Crucially, the semantic attributes that constitute the essence of an image may differ from image to image. Our blending operator combines the powerful StyleGAN generator and the semantic encoder of CLIP in a novel way that is simultaneously additive in both latent spaces, resulting in a mechanism that guarantees both identity preservation and high-level feature transfer without relying on a facial recognition network. We present two variants of our method. The first is based on optimization, and the second fine-tunes an existing inversion encoder to perform essence extraction. Through extensive experiments, we demonstrate the superiority of our methods for essence transfer over existing methods for style transfer, domain adaptation, and text-based semantic editing.



Paperid:548
Authors:Rui Shao; Tianxing Wu; Ziwei Liu
Title: Detecting and Recovering Sequential DeepFake Manipulation
Abstract:
Since photorealistic faces can be readily generated by facial manipulation technologies nowadays, potential malicious abuse of these technologies has drawn great concerns. Numerous deepfake detection methods are thus proposed. However, existing methods only focus on detecting one-step facial manipulation. As the emergence of easy-accessible facial editing applications, people can easily manipulate facial components using multi-step operations in a sequential manner. This new threat requires us to detect a sequence of facial manipulations, which is vital for both detecting deepfake media and recovering original faces afterwards. Motivated by this observation, we emphasize the need and propose a novel research problem called Detecting Sequential DeepFake Manipulation (Seq-DeepFake). Unlike the existing deepfake detection task only demanding a binary label prediction, detecting Seq-DeepFake manipulation requires correctly predicting a sequential vector of facial manipulation operations. To support a large-scale investigation, we construct the first Seq-DeepFake dataset, where face images are manipulated sequentially with corresponding annotations of sequential facial manipulation vectors. Based on this new dataset, we cast detecting Seq-DeepFake manipulation as a specific image-to-sequence (e.g. image captioning) task and propose a concise yet effective Seq-DeepFake Transformer (SeqFakeFormer). Moreover, we build a comprehensive benchmark and set up rigorous evaluation protocols and metrics for this new research problem. Extensive experiments demonstrate the effectiveness of SeqFakeFormer. Several valuable observations are also revealed to facilitate future research in broader deepfake detection problems.



Paperid:549
Authors:Jhih-Ciang Wu; He-Yen Hsieh; Ding-Jie Chen; Chiou-Shann Fuh; Tyng-Luh Liu
Title: Self-Supervised Sparse Representation for Video Anomaly Detection
Abstract:
Video anomaly detection (VAD) aims at localizing unexpected actions or activities in a video sequence. Existing mainstream VAD techniques are based on either the one-class formulation, which assumes all training data are normal, or weakly-supervised, which requires only video-level normal/anomaly labels. To establish a unified approach to solving the two VAD settings, we introduce a self-supervised sparse representation (S3R) framework that models the concept of anomaly at feature level by exploring the synergy between dictionary-based representation and self-supervised learning. With the learned dictionary, S3R facilitates two coupled modules, en-Normal and de-Normal, to reconstruct snippet-level features and filter out normal-event features. The self-supervised techniques also enable generating samples of pseudo normal/anomaly to train the anomaly detector. We demonstrate with extensive experiments that S3R achieves new state-of-the-art performances on popular benchmark datasets for both one-class and weakly-supervised VAD tasks. Our code is publicly available at https://github.com/louisYen/S3R.



Paperid:550
Authors:Xinwei Liu; Jian Liu; Yang Bai; Jindong Gu; Tao Chen; Xiaojun Jia; Xiaochun Cao
Title: Watermark Vaccine: Adversarial Attacks to Prevent Watermark Removal
Abstract:
As a common security tool, visible watermarking has been widely applied to protect copyrights of digital images. However, recent works have shown that visible watermarks can be removed by DNNs without damaging their host images. Such watermark-removal techniques pose a great threat to the ownership of images. Inspired by the vulnerability of DNNs on adversarial perturbations, we propose a novel defence mechanism by adversarial machine learning for good. From the perspective of the adversary, blind watermark-removal networks can be posed as our target models; then we actually optimize an imperceptible adversarial perturbation on the host images to proactively attack against watermark-removal networks, dubbed Watermark Vaccine. Specifically, two types of vaccines are proposed. Disrupting Watermark Vaccine (DWV) induces to ruin the host image along with watermark after passing through watermark-removal networks. In contrast, Inerasable Watermark Vaccine (IWV) works in another fashion of trying to keep the watermark not removed and still noticeable. Extensive experiments demonstrate the effectiveness of our DWV/IWV in preventing watermark removal, especially on various watermark removal networks. The Code is released in https://github.com/thinwayliu/Watermark-Vaccine.



Paperid:551
Authors:Shichao Dong; Jin Wang; Jiajun Liang; Haoqiang Fan; Renhe Ji
Title: Explaining Deepfake Detection by Analysing Image Matching
Abstract:
This paper aims to interpret how deepfake detection models learn artifact features of images when just supervised by binary labels. To this end, three hypotheses from the perspective of image matching are proposed as follows. 1. Deepfake detection models indicate real/fake images based on visual concepts that are neither source-relevant nor target-relevant, that is, considering such visual concepts as artifact-relevant. 2. Besides the supervision of binary labels, deepfake detection models implicitly learn artifact-relevant visual concepts through the FST-Matching (i.e. the matching fake, source, target images) in the training set. 3. Implicitly learned artifact visual concepts through the FST-Matching in the raw training set are vulnerable to video compression. In experiments, the above hypotheses are verified among various DNNs. Furthermore, based on this understanding, we propose the FST-Matching Deepfake Detection Model to boost the performance of forgery detection on compressed videos. Experiment results show that our method achieves great performance, especially on highly-compressed (e.g. c40) videos.



Paperid:552
Authors:Julia Grabinski; Steffen Jung; Janis Keuper; Margret Keuper
Title: FrequencyLowCut Pooling – Plug \& Play against Catastrophic Overfitting
Abstract:
Over the last years, Convolutional Neural Networks (CNNs) have been the dominating neural architecture in a wide range of computer vision tasks. From an image and signal processing point of view, this success might be a bit surprising as the inherent spatial pyramid design of most CNNs is apparently violating basic signal processing laws, i.e. Sampling Theorem in their down-sampling operations. However, since poor sampling appeared not to affect model accuracy, this issue has been broadly neglected until model robustness started to receive more attention. Recent work [18] in the context of adversarial attacks and distribution shifts, showed after all, that there is a strong correlation between the vulnerability of CNNs and aliasing artifacts induced by poor down-sampling operations. This paper builds on these findings and introduces an aliasing free down-sampling operation which can easily be plugged into any CNN architecture: FrequencyLowCut pooling. Our experiments show, that in combination with simple and Fast Gradient Sign Method (FGSM) adversarial training, our hyper-parameter free operator substantially improves model robustness and avoids catastrophic overfitting. Our code is available at https://github.com/GeJulia/flc_pooling.



Paperid:553
Authors:Shivangi Aneja; Lev Markhasin; Matthias Nießner
Title: TAFIM: Targeted Adversarial Attacks against Facial Image Manipulations
Abstract:
Face manipulation methods can be misused to affect an individual’s privacy or to spread disinformation. To this end, we introduce a novel data-driven approach that produces image-specific perturbations which are embedded in the original images. The key idea is that these protected images prevent face manipulation by causing the manipulation model to produce a predefined manipulation target (uniformly colored output image in our case) instead of the actual manipulation. In addition, we propose to leverage differentiable compression approximation, hence making generated perturbations robust to common image compression. In order to prevent against multiple manipulation methods simultaneously, we further propose a novel attention-based fusion of manipulation-specific perturbations. Compared to traditional adversarial attacks that optimize noise patterns for each image individually, our generalized model only needs a single forward pass, thus running orders of magnitude faster and allowing for easy integration in image processing stacks, even on resource-constrained devices like smartphones"



Paperid:554
Authors:Yonghyun Jeong; Doyeon Kim; Youngmin Ro; Pyounggeon Kim; Jongwon Choi
Title: FingerprintNet: Synthesized Fingerprints for Generated Image Detection
Abstract:
While recent advances in generative models benefit the society, the generated images can be abused for malicious purposes, like fraud, defamation, and false news. To prevent such cases, vigorous research is conducted on distinguishing the generated images from the real ones, but challenges still remain with detecting the unseen generated images outside of the training settings. To overcome this problem, we analyze the distinctive characteristic of the generated images called ‘fingerprints,’ and propose a new framework to reproduce diverse types of fingerprints generated by various generative models. By training the model with the real images only, our framework can avoid data dependency on particular generative models and enhance generalization. With the mathematical derivation that the fingerprint is emphasized at the frequency domain, we design a generated image detector for effective training of the fingerprints. Our framework outperforms the prior state-of-the-art detectors, even though only real images are used for training. We also provide new benchmark datasets to demonstrate the model’s robustness using the images of the latest anti-artifact generative models for reducing the spectral discrepancies.



Paperid:555
Authors:Bo Liu; Fan Yang; Xiuli Bi; Bin Xiao; Weisheng Li; Xinbo Gao
Title: Detecting Generated Images by Real Images
Abstract:
The widespread of generative models have called into question the authenticity of many things on the web. In this situation, the task of image forensics is urgent. The existing methods examine generated images and claim a forgery by detecting visual artifacts or invisible patterns, resulting in generalization issues. We observed that the noise pattern of real images exhibits similar characteristics in the frequency domain, while the generated images are far different. Therefore, we can perform image authentication by checking whether an image follows the patterns of authentic images. The experiments show that a simple classifier using noise patterns can easily detect a wide range of generative models, including GAN and flow-based models. Our method achieves state-of-the-art performance on both low- and high-resolution images from a wide range of generative models and shows superior generalization ability to unseen models.



Paperid:556
Authors:Ke Sun; Hong Liu; Taiping Yao; Xiaoshuai Sun; Shen Chen; Shouhong Ding; Rongrong Ji
Title: An Information Theoretic Approach for Attention-Driven Face Forgery Detection
Abstract:
Recently, Deepfakes arises as a powerful tool to fool the existing real-world face detection systems, which has received wide attention in both academia and society. Most existing forgery face detection methods use heuristic clues to build a binary forgery detector, which mainly takes advantage of the empirical observation based on abnormal texture, blending clues, or high-frequency noise, etc. However, heuristic clues only reflect certain aspects of the forgery, which lead to model bias or sub-optimization. Our key observation is that most of the forgery clues are hidden in the informative region, which can be measured quantitatively by classical information maximization theory. Motivated by this, we make the first attempt to introduce the self-information metric to enhance the forgery feature representation. The metric can be formulated as a plug-and-play block, termed self-information attention (SIA) module, that can be applied to most recent top-performance deep model. The SIA module can explicitly help the model extract high information features and recalibrate channel-wise feature responses, which improves both model’s performance and generalization with few additional parameters. Extensive experiments on several large-scale benchmarks demonstrate the superiority of the proposed method against the state-of-the-art competitors.



Paperid:557
Authors:Jiahao Liang; Huafeng Shi; Weihong Deng
Title: Exploring Disentangled Content Information for Face Forgery Detection
Abstract:
Convolutional neural network based face forgery detection methods have achieved remarkable results during training, but struggled to maintain comparable performance during testing. We observe that the detector is prone to focus more on content information than artifact traces, suggesting that the detector is sensitive to the intrinsic bias of the dataset, which leads to severe overfitting. Motivated by this key observation, we design an easily embeddable disentanglement framework for content information removal, and further propose a Content Consistency Constraint (C2C) and a Global Representation Contrastive Constraint (GRCC) to enhance the independence of disentangled features. Furthermore, we cleverly construct two unbalanced datasets to investigate the impact of the content bias. Extensive visualizations and experiments demonstrate that our framework can not only ignore the interference of content information, but also guide the detector to mine suspicious artifact traces and achieve competitive performance.



Paperid:558
Authors:Tu Bui; Ning Yu; John Collomosse
Title: RepMix: Representation Mixing for Robust Attribution of Synthesized Images
Abstract:
Rapid advances in Generative Adversarial Networks (GANs) raise new challenges for image attribution; detecting whether an image is synthetic and, if so, determining which GAN architecture created it. Uniquely, we present a solution to this task capable of 1) matching images invariant to their semantic content; 2) robust to benign transformations (changes in quality, resolution, shape, etc.) commonly encountered as images are re-shared online. In order to formalize our research, a challenging benchmark, Attribution88, is collected for robust and practical image attribution. We then propose RepMix, our GAN fingerprinting technique based on representation mixing and a novel loss. We validate its capability of tracing the provenance of GAN-generated images invariant to the semantic content of the image and also robust to perturbations. We show our approach improves significantly from existing GAN fingerprinting works on both semantic generalization and robustness.



Paperid:559
Authors:Jingwei Ma; Lucy Chai; Minyoung Huh; Tongzhou Wang; Ser-Nam Lim; Phillip Isola; Antonio Torralba
Title: Totems: Physical Objects for Verifying Visual Integrity
Abstract:
We introduce a new approach to image forensics: placing physical refractive objects, which we call totems, into a scene so as to protect any photograph taken of that scene. Totems bend and redirect light rays, thus providing multiple, albeit distorted, views of the scene within a single image. A defender can use these distorted totem pixels to detect if an image has been manipulated. Our approach unscrambles the light rays passing through the totems by estimating their positions in the scene and using their known geometric and material properties. To verify a totem-protected image, we detect inconsistencies between the scene reconstructed from totem viewpoints and the scene’s appearance from the camera viewpoint. Such an approach makes the adversarial manipulation task more difficult, as the adversary must modify both the totem and image pixels in a geometrically consistent manner without knowing the physical properties of the totem. Unlike prior learning-based approaches, our method does not require training on datasets of specific manipulations, and instead uses physical properties of the scene and camera to solve the forensics problem.



Paperid:560
Authors:Pandeng Li; Hongtao Xie; Jiannan Ge; Lei Zhang; Shaobo Min; Yongdong Zhang
Title: Dual-Stream Knowledge-Preserving Hashing for Unsupervised Video Retrieval
Abstract:
Unsupervised video hashing usually optimizes binary codes by learning to reconstruct input videos. Such reconstruction constraint spends much effort on frame-level temporal context changes without focusing on video-level global semantics that are more useful for retrieval. Hence, we address this problem by decomposing video information into reconstruction-dependent and semantic-dependent information, which disentangles the semantic extraction from reconstruction constraint. Specifically, we first design a simple dual-stream structure, including a temporal layer and a hash layer. Then, with the help of semantic similarity knowledge obtained from self-supervision, the hash layer learns to capture information for semantic retrieval, while the temporal layer learns to capture the information for reconstruction. In this way, the model naturally preserves the disentangled semantics into binary codes. Validated by comprehensive experiments, our method consistently outperforms the state-of-the-arts on three video benchmarks.



Paperid:561
Authors:Kuan Zhu; Haiyun Guo; Tianyi Yan; Yousong Zhu; Jinqiao Wang; Ming Tang
Title: PASS: Part-Aware Self-Supervised Pre-training for Person Re-identification
Abstract:
In person re-identification (ReID), very recent researches have validated pre-training the models on unlabelled person images is much better than on ImageNet. However, these researches directly apply the existing self-supervised learning (SSL) methods designed for image classification to ReID without any adaption in the framework. These SSL methods match the outputs of local views (e.g., red T-shirt, blue shorts) to those of the global views at the same time, losing lots of details. In this paper, we propose a ReID-specific pre-training method, Part-Aware Self-Supervised pre-training (PASS), which can generate part-level features to offer fine-grained information and is more suitable for ReID. PASS divides the images into several local areas, and the local views randomly cropped from each area are assigned with a specific learnable [PART] token. On the other hand, the [PART]s of all local areas are also appended to the global views. PASS learns to match the output of the local views and global views on the same [PART]. That is, the learned [PART] of the local views from a local area is only matched with the corresponding [PART] learned from the global views. As a result, each [PART] can focus on a specific local area of the image and extracts fine-grained information of this area. Experiments show PASS sets the new state-of-the-art performances on Market1501 and MSMT17 on various ReID tasks, e.g., vanilla ViT-S/16 pre-trained by PASS achieves 92.2\%/90.2\%/88.5\% mAP accuracy on Market1501 for supervised/UDA/USL ReID. Our codes are in supplementary materials and will be released.



Paperid:562
Authors:Pengyi Zhang; Huanzhang Dou; Yunlong Yu; Xi Li
Title: Adaptive Cross-Domain Learning for Generalizable Person Re-identification
Abstract:
Domain Generalizable Person Re-Identification (DG-ReID) is a more practical ReID task that is trained from multiple source domains and tested on the unseen target domains. Most existing methods are challenged for dealing with the shared and specific characteristics among different domains, which is called the domain conflict problem. To address this problem, we present an Adaptive Cross-domain Learning (ACL) framework equipped with a CrOss-Domain Embedding Block (CODE-Block) to maintain a common feature space for capturing both the domain-invariant and the domain-specific features, while dynamically mining the relations across different domains. Moreover, our model adaptively adjusts the architecture to focus on learning the corresponding features of a single domain at a time without interference from the biased features of other domains. Specifically, the CODE-Block is composed of two complementary branches, a dynamic branch for extracting domain-adaptive features and a static branch for extracting the domain-invariant features. Extensive experiments demonstrate that the proposed approach achieves state-of-the-art performances on the popular benchmarks. Under Protocol-2, our method outperforms previous SOTA by 7.8% and 7.6% in terms of mAP and rank-1 accuracy.



Paperid:563
Authors:Zeyu Wang; Yu Wu; Karthik Narasimhan; Olga Russakovsky
Title: Multi-Query Video Retrieval
Abstract:
Retrieving target videos based on text descriptions is a task of great practical value and has received increasing attention over the past few years. Despite recent progress, imperfect annotations in existing video retrieval datasets have posed significant challenges on model evaluation and development. In this paper, we tackle this issue by focusing on the less-studied setting of multi-query video retrieval, where multiple descriptions are provided to the model for searching over the video archive. We first show that multi-query retrieval task effectively mitigates the dataset noise introduced by imperfect annotations and better correlates with human judgement on evaluating retrieval abilities of current models. We then investigate several methods which leverage multiple queries at training time, and demonstrate that the multi-query inspired training can lead to superior performance and better generalization. We hope further investigation in this direction can bring new insights on building systems that perform better in real-world video retrieval applications.



Paperid:564
Authors:Elias Ramzi; Nicolas Audebert; Nicolas Thome; Clément Rambour; Xavier Bitot
Title: Hierarchical Average Precision Training for Pertinent Image Retrieval
Abstract:
Image Retrieval is commonly evaluated with Average Precision (AP) or Recall@k. Yet, those metrics, are limited to binary labels and do not take into account errors’ severity. This paper introduces a new hierarchical AP training method for pertinent image retrieval (HAPPIER). HAPPIER is based on a new H-AP metric, which leverages a concept hierarchy to refine AP by integrating errors’ importance and better evaluate rankings. To train deep models with H-AP, we carefully study the problem’s structure and design a smooth lower bound surrogate combined with a clustering loss that ensures consistent ordering. Extensive experiments on 6 datasets show that HAPPIER significantly outperforms state-of-the-art methods for hierarchical retrieval, while being on par with the latest approaches when evaluating fine-grained ranking performances. Finally, we show that HAPPIER leads to better organization of the embedding space, and prevents most severe failure cases of non-hierarchical methods. Our code is publicly available at https://github.com/elias-ramzi/HAPPIER.



Paperid:565
Authors:Shuaiyi Huang; Luyu Yang; Bo He; Songyang Zhang; Xuming He; Abhinav Shrivastava
Title: Learning Semantic Correspondence with Sparse Annotations
Abstract:
Finding dense semantic correspondence is a fundamental problem in computer vision, which remains challenging in complex scenes due to background clutter, extreme intra-class variation, and a severe lack of ground truth. In this paper, we aim to address the challenge of label sparsity in semantic correspondence by enriching supervision signals from sparse keypoint annotations. To this end, we first propose a teacher-student learning paradigm for generating dense pseudo-labels and then develop two novel strategies for denoising pseudo-labels. In particular, we use spatial priors around the sparse annotations to suppress the noisy pseudo-labels. In addition, we introduce a loss-driven dynamic label selection strategy for label denoising. We instantiate our paradigm with two variants of learning strategies: a single offline teacher setting, and a mutual online teachers setting. Our approach achieves notable improvements on three challenging benchmarks for semantic correspondence and establishes the new state-of-the-art. Project page: https://shuaiyihuang.github.io/publications/SCorrSAN.



Paperid:566
Authors:Bingliang Jiao; Lingqiao Liu; Liying Gao; Guosheng Lin; Lu Yang; Shizhou Zhang; Peng Wang; Yanning Zhang
Title: Dynamically Transformed Instance Normalization Network for Generalizable Person Re-identification
Abstract:
Existing person re-identification methods often suffer significant performance degradation on unseen domains, which fuels interest in domain generalizable person re-identification (DG-PReID). As an effective technology to alleviate domain variance, the Instance Normalization (IN) has been widely employed in many existing works. However, IN also suffers from the limitation of eliminating discriminative patterns that might be useful for a particular domain or instance. In this work, we propose a new normalization scheme called Dynamically Transformed Instance Normalization (DTIN) to alleviate the drawback of IN. Our idea is to employ dynamic convolution to allow the unnormalized feature to control the transformation of the normalized features into new representations. In this way, we can ensure the network has sufficient flexibility to strike the right balance between eliminating irrelevant domain-specific features and adapting to individual domains or instances. We further utilize a multi-task learning strategy to train the model, ensuring it can adaptively produce discriminative feature representations for an arbitrary domain. Our results show a great domain generation capability and achieve state-of-the-art performance on three mainstream DG-PReID settings.



Paperid:567
Authors:Junjie Li; Yichao Yan; Guanshuo Wang; Fufu Yu; Qiong Jia; Shouhong Ding
Title: Domain Adaptive Person Search
Abstract:
Person search is a challenging task which aims to achieve joint pedestrian detection and person re-identification (ReID). Previous works have made significant advances under fully and weakly supervised settings. However, existing methods ignore the generalization ability of the person search models. In this paper, we take a further step and present Domain Adaptive Person Search (DAPS), which aims to generalize the model from a labeled source domain to the unlabeled target domain. Two major challenges arises under this new setting: one is how to simultaneously solve the domain misalignment issue for both detection and Re-ID tasks, and the other is how to train the ReID subtask without reliable detection results on the target domain. To address these challenges, we propose a strong baseline framework with two dedicated designs. 1) We design a domain alignment module including image-level and task-sensitive instance-level alignments, to minimize the domain discrepancy. 2) We take full advantage of the unlabeled data with a dynamic clustering strategy, and employ pseudo bounding boxes to support ReID and detection training on the target domain. With the above designs, our framework achieves 34.7% in mAP and 80.6% in top-1 on PRW dataset, surpassing the direct transferring baseline by a large margin. Surprisingly, the performance of our unsupervised DAPS model even surpasses some of the fully and weakly supervised methods. The code is available at https://github.com/caposerenity/DAPS.



Paperid:568
Authors:Yuqi Liu; Pengfei Xiong; Luhui Xu; Shengming Cao; Qin Jin
Title: TS2-Net: Token Shift and Selection Transformer for Text-Video Retrieval
Abstract:
Text-Video retrieval is a task of great practical value and has received increasing attention, among which learning spatial-temporal video representation is one of the research hotspots. The video encoders in the state-of-the-art video retrieval models usually directly adopt the pre-trained vision backbones with the network structure fixed, they therefore can not be further improved to produce the fine-grained spatial-temporal video representation. In this paper, we propose Token Shift and Selection Network (TS2-Net), a novel token shift and selection transformer architecture, which dynamically adjusts the token sequence and selects informative tokens in both temporal and spatial dimensions from input video samples. The token shift module temporally shifts the whole token features back-and-forth across adjacent frames, to preserve the complete token representation and capture subtle movements. Then the token selection module selects tokens that contribute most to local spatial semantics. Based on thorough experiments, the proposed TS2-Net achieves state-of-the-art performance on major text-video retrieval benchmarks, including new records on MSRVTT, VATEX, LSMDC, ActivityNet, and DiDeMo. Code is available at https://github.com/yuqi657/ts2_net.



Paperid:569
Authors:Wen Qian; Hao Luo; Silong Peng; Fan Wang; Chen Chen; Hao Li
Title: Unstructured Feature Decoupling for Vehicle Re-identification
Abstract:
The misalignment of features caused by pose and viewpoint variances is a crucial problem in Vehicle Re-Identification (ReID). Previous methods align the features by structuring the vehicles from pre-defined vehicle parts (such as logos, lights, windows, etc.) or vehicle attributes, which are inefficient because of additional manual annotation. To align the features without requirements of additional annotation, this paper proposes a Unstructured Feature Decoupling Network (UFDN), which consists of a transformer-based feature decomposing head (TDH) and a novel cluster-based decoupling constraint (CDC). Different from the structured knowledge used in previous decoupling methods, we aim to achieve more flexible unstructured decoupled features with diverse discriminative information as shown in Fig. 1. The self-attention mechanism in the decomposing head helps the model preliminarily learn the discriminative decomposed features in a global scope. To further learn diverse but aligned decoupled features, we introduce a cluster-based decoupling constraint consisting of a diversity constraint and an alignment constraint. Furthermore, we improve the alignment constraint into a modulated one to eliminate the negative impact of the outlier features that cannot align the clusters in semantic. Extensive experiments show the proposed UFDN achieves state-of-the-art performance on three popular Vehicle ReID benchmarks with both CNN and Transformer backbones.



Paperid:570
Authors:Young Kyun Jang; Geonmo Gu; Byungsoo Ko; Isaac Kang; Nam Ik Cho
Title: Deep Hash Distillation for Image Retrieval
Abstract:
In hash-based image retrieval systems, degraded or transformed inputs usually generate different codes from the original, deteriorating the retrieval accuracy. To mitigate this issue, data augmentation can be applied during training. However, even if augmented samples of an image are similar in real feature space, the quantization can scatter them far away in Hamming space. This results in representation discrepancies that can impede training and degrade performance. In this work, we propose a novel self-distilled hashing scheme to minimize the discrepancy while exploiting the potential of augmented data. By transferring the hash knowledge of the weakly-transformed samples to the strong ones, we make the hash code insensitive to various transformations. We also introduce hash proxy-based similarity learning and binary cross entropy-based quantization loss to provide fine quality hash codes. Ultimately, we construct a deep hashing framework that not only improves the existing deep hashing approaches, but also achieves the state-of-the-art retrieval results. Extensive experiments are conducted and confirm the effectiveness of our work.



Paperid:571
Authors:Boqiang Xu; Jian Liang; Lingxiao He; Zhenan Sun
Title: Mimic Embedding via Adaptive Aggregation: Learning Generalizable Person Re-identification
Abstract:
Domain generalizable (DG) person re-identification (ReID) aims to test across unseen domains without access to the target domain data at training time, which is a realistic but challenging problem. In contrast to methods assuming an identical model for different domains, Mixture of Experts (MoE) exploits multiple domain-specific networks for leveraging complementary information between domains, obtaining impressive results. However, prior MoE-based DG ReID methods suffer from a large model size with the increase of the number of source domains, and most of them overlook the exploitation of domain-invariant characteristics. To handle the two issues above, this paper presents a new approach called Mimic Embedding via adapTive Aggregation (META) for DG person ReID. To avoid the large model size, experts in META do not adopt a branch network for each source domain but share all the parameters except for the batch normalization layers. Besides multiple experts, META leverages Instance Normalization (IN) and introduces it into a global branch to pursue invariant features across domains. Meanwhile, META considers the relevance of an unseen target sample and source domains via normalization statistics and develops an aggregation module to adaptively integrate multiple experts for mimicking unseen target domain. Benefiting from a proposed consistency loss and an episodic training algorithm, META is expected to mimic embedding for a truly unseen target domain. Extensive experiments verify that META surpasses state-of-the-art DG person ReID methods by a large margin.



Paperid:572
Authors:Jon Almazán; Byungsoo Ko; Geonmo Gu; Diane Larlus; Yannis Kalantidis
Title: Granularity-Aware Adaptation for Image Retrieval over Multiple Tasks
Abstract:
Strong image search models can be learned for a specific domain, ie. set of labels, provided that some labeled images of that domain are available. A practical visual search model, however, should be versatile enough to solve multiple retrieval tasks simultaneously, even if those cover very different specialized domains. Additionally, it should be able to benefit from even unlabeled images from these various retrieval tasks. This is the more practical scenario that we consider in this paper. We address it with the proposed Grappa, an approach that starts from a strong pretrained model, and adapts it to tackle multiple retrieval tasks concurrently, using only unlabeled images from the different task domains. We extend the pretrained model with multiple independently trained sets of adaptors that use pseudo-label sets of different sizes, effectively mimicking different pseudo-granularities. We reconcile all adaptor sets into a single unified model suited for all retrieval tasks by learning fusion layers that we guide by propagating pseudo-granularity attentions across neighbors in the feature space. Results on a benchmark composed of six heterogeneous retrieval tasks show that the unsupervised Grappa model improves the zero-shot performance of a state-of-the-art self-supervised learning model, and in some places reaches or improves over a task label-aware oracle that selects the most fitting pseudo-granularity per task.



Paperid:573
Authors:Arsha Nagrani; Paul Hongsuck Seo; Bryan Seybold; Anja Hauth; Santiago Manen; Chen Sun; Cordelia Schmid
Title: Learning Audio-Video Modalities from Image Captions
Abstract:
There has been a recent explosion of large-scale image-text datasets, as images with alt-text captions can be easily obtained online. Obtaining large-scale, high quality data for video in the form of text-video and text-audio pairs however, is more challenging. To close this gap we propose a new video mining pipeline which involves transferring captions from image captioning datasets to video clips with no additional manual effort. Using this pipeline, we create a new large-scale, weakly labelled audio-video captioning dataset consisting of millions of paired clips and captions. We show that training a multimodal transformed based model on this data achieves competitive performance on video retrieval and video captioning, matching or even outperforming HowTo100M pretraining with 20x fewer clips. We also show that our mined clips are suitable for text-audio pretraining, and achieve state of the art results for the task of audio retrieval.



Paperid:574
Authors:Wei-Ting Chen; I-Hsiang Chen; Chih-Yuan Yeh; Hao-Hsiang Yang; Hua-En Chang; Jian-Jiun Ding; Sy-Yen Kuo
Title: RVSL: Robust Vehicle Similarity Learning in Real Hazy Scenes Based on Semi-Supervised Learning
Abstract:
Recently, vehicle similarity learning, also called re-identification (ReID), has attracted significant attention in computer vision. Several algorithms have been developed and obtained considerable success. However, most existing methods have unpleasant performance in the hazy scenario due to poor visibility. Though some strategies are possible to resolve this problem, they still have room to be improved due to the limited performance in real-world scenarios and the lack of real-world clear ground truth. Thus, to resolve this problem, inspired by CycleGAN, we construct a training paradigm called \textbf{RVSL} which integrates ReID and domain transformation techniques. The network is trained on semi-supervised fashion and does not require to employ the ID labels and the corresponding clear ground truths to learn hazy vehicle ReID mission in the real-world haze scenes. To further constrain the unsupervised learning process effectively, several losses are developed. Experimental results on synthetic and real-world datasets indicate that the proposed method can achieve state-of-the-art performance on hazy vehicle ReID problems. It is worth mentioning that although the proposed method is trained without real-world label information, it can achieve competitive performance compared to existing supervised methods trained on complete label information.



Paperid:575
Authors:Fan Hu; Aozhu Chen; Ziyue Wang; Fangming Zhou; Jianfeng Dong; Xirong Li
Title: Lightweight Attentional Feature Fusion: A New Baseline for Text-to-Video Retrieval
Abstract:
In this paper we revisit feature fusion, an old-fashioned topic, in the new context of text-to-video retrieval. Different from previous research that considers feature fusion only at one end, let it be video or text, we aim for feature fusion for both ends within a unified framework. We hypothesize that optimizing the convex combination of the features is preferred to modeling their correlations by computationally heavy multi-head self attention. We propose Lightweight Attentional Feature Fusion (LAFF). LAFF performs feature fusion at both early and late stages and at both video and text ends, making it a powerful method for exploiting diverse (off-the-shelf) features. The interpretability of LAFF can be used for feature selection. Extensive experiments on five public benchmark sets (MSR-VTT, MSVD, TGIF, VATEX and TRECVID AVS 2016-2020) justify LAFF as a new baseline for text-to-video retrieval.



Paperid:576
Authors:Yiyuan Zhang; Sanyuan Zhao; Yuhao Kang; Jianbing Shen
Title: Modality Synergy Complement Learning with Cascaded Aggregation for Visible-Infrared Person Re-identification
Abstract:
Visible-Infrared Re-Identification (VI-ReID) is challenging in image retrievals. The modality discrepancy will easily make huge intra-class variations. Most existing methods either bridge different modalities through modality-invariance or generate the intermediate modality for better performance. Differently, this paper proposes a novel framework, named Modality Synergy Complement Learning Network (MSCLNet) with Cascaded Aggregation. Its basic idea is to synergize two modalities to construct diverse representations of identity-discriminative semantics and less noise. Then, we complement synergistic representations under the advantages of the two modalities. Furthermore, we propose the Cascaded Aggregation strategy for fine-grained optimization of the feature distribution, which progressively aggregates feature embeddings from the subclass, intra-class, and inter-class. Extensive experiments on SYSU-MM01 and RegDB datasets show that MSCLNet outperforms the state-of-the-art by a large margin. On the large-scale SYSU-MM01 dataset, our model can achieve 76.99% and 71.64% in terms of Rank-1 accuracy and mAP value. Our code will be available at https://github.com/bitreidgroup/VI-ReID-MSCLNet"



Paperid:577
Authors:Kongzhu Jiang; Tianzhu Zhang; Xiang Liu; Bingqiao Qian; Yongdong Zhang; Feng Wu
Title: Cross-Modality Transformer for Visible-Infrared Person Re-identification
Abstract:
Visible-infrared person re-identification (VI-ReID) is a challenging task due to the large cross-modality discrepancies and intra-class variations. Existing works mainly focus on learning modality-shared representations by embedding different modalities into the same feature space. However, these methods usually damage the modality-specific information and identification information contained in the features. To alleviate the above issues, we propose a novel Cross-Modality Transformer (CMT) to jointly explore a modality-level alignment module and an instance-level module for VI-ReID. The proposed CMT enjoys several merits. First, the modality-level alignment module is designed to compensate for the missing modality-specific information via a Transformer encoder-decoder architecture. Second, we propose an instance-level alignment module to adaptively adjust the sample features, which is achieved by a query-adaptive feature modulation. To the best of our knowledge, this is the first work to exploit a cross-modality transformer to achieve the modality compensation for VI-ReID. Extensive experimental results on two standard benchmarks demonstrate that our CMT performs favorably against the state-of-the-art methods.



Paperid:578
Authors:Sangmin Lee; Sungjune Park; Yong Man Ro
Title: Audio-Visual Mismatch-Aware Video Retrieval via Association and Adjustment
Abstract:
Retrieving desired videos using natural language queries has attracted increasing attention in research and industry fields as a huge number of videos appear on the internet. Some existing methods attempted to address this video retrieval problem by exploiting multi-modal information, especially audio-visual data of videos. However, many videos often have mismatched visual and audio cues for several reasons including background music, noise, and even missing sound. Therefore, the naive fusion of such mismatched visual and audio cues can negatively affect the semantic embedding of video scenes. Mismatch condition can be categorized into two cases: (i) Audio itself does not exist, (ii) Audio exists but does not match with visual. To deal with (i), we introduce audio-visual associative memory (AVA-Memory) to associate audio cues even from videos without audio data. The associated audio cues can guide the video embedding feature to be aware of audio information even in the missing audio condition. To address (ii), we propose audio embedding adjustment by considering the degree of matching between visual and audio data. In this procedure, constructed AVA-Memory enables to figure out how well the visual and audio in the video are matched and to adjust the weighting between actual audio and associated audio. Experimental results show that the proposed method outperforms other state-of-the-art video retrieval methods. Further, we validate the effectiveness of the proposed network designs with ablation studies and analyses.



Paperid:579
Authors:Haokui Zhang; Buzhou Tang; Wenze Hu; Xiaoyu Wang
Title: Connecting Compression Spaces with Transformer for Approximate Nearest Neighbor Search
Abstract:
We propose a generic feature compression method for Approximate Nearest Neighbor Search (ANNS) problems, which speeds up existing ANNS methods in a plug-and-play manner. Specifically, based on transformer, we propose a new network structure to compress the feature into a low dimensional space, and an inhomogeneous neighborhood relationship preserving (INRP) loss that aims to maintain high search accuracy. Specifically, we use multiple compression projections to cast the feature into many low dimensional spaces, and then use transformer to globally optimize these projections such that the features are well compressed following the guidance from our loss function. The loss function is designed to assign high weights on point pairs that are close in original feature space, and keep their distances in projected space. Keeping these distances helps maintain the eventual top-k retrieval accuracy, and down weighting others creates room for feature compression. In experiments, we run our compression method on public datasets, and use the compressed features in graph based, product quantization and scalar quantization based ANNS solutions. Experimental results show that our compression method can significantly improve the efficiency of these methods while preserves or even improves search accuracy, suggesting its broad potential impact on real world applications. Source code is available at https://github.com/hkzhang91/CCST"



Paperid:580
Authors:Yang Shen; Xuhao Sun; Xiu-Shen Wei; Qing-Yuan Jiang; Jian Yang
Title: SEMICON: A Learning-to-Hash Solution for Large-Scale Fine-Grained Image Retrieval
Abstract:
In this paper, we propose Suppression-Enhancing Mask based attention and Interactive Channel transformatiON (SEMICON) to learn binary hash codes for dealing with large-scale fine-grained image retrieval tasks. In SEMICON, we first develop a suppression-enhancing mask (SEM) based attention to dynamically localize discriminative image regions. More importantly, different from existing attention mechanism simply erasing previous discriminative regions, our SEM is developed to restrain such regions and then discover other complementary regions by considering the relation between activated regions in a stage-by-stage fashion. In each stage, the interactive channel transformation (ICON) module is afterwards designed to exploit correlations across channels of attended activation tensors. Since channels could generally correspond to the parts of fine-grained objects, the part correlation can be also modeled accordingly, which further improves fine-grained retrieval accuracy. Moreover, to be computational economy, ICON is realized by an efficient two-step process. Finally, the hash learning of our SEMICON consists of both global- and local-level branches for better representing fine-grained objects and then generating binary hash codes explicitly corresponding to multiple levels. Experiments on five benchmark fine-grained datasets show our superiority over competing methods. Codes are available at https://github.com/NJUST-VIPGroup/SEMICON.



Paperid:581
Authors:Jinlin Wu; Lingxiao He; Wu Liu; Yang Yang; Zhen Lei; Tao Mei; Stan Z. Li
Title: CAViT: Contextual Alignment Vision Transformer for Video Object Re-identification
Abstract:
Video object re-identification (reID) aims at re-identifying the same object under non-overlapping cameras by matching the video tracklets with cropped video frames. The key point is how to make full use of spatio-temporal interactions to extract more accurate representation. However, there are dilemmas within existing approaches: (1) 3D solutions model the spatio-temporal interaction but are often troubled with the misalignment of adjacent frames, and (2) 2D solutions adopt a divide-and-conquer strategy against the misalignment but cannot take advantage of the spatio-temporal interactions. To address the above problems, we propose a Contextual Alignment Vision Transformer (\textbf{CAViT}) to the spatio-temporal interaction with a 2D solution. It contains a Multi-shape Patch Embedding (\textbf{MPE}) module and a Temporal Shift Attention (\textbf{TSA}) module. MPE is designed to retain spatial semantic information against the misalignment caused by pose, occlusion, or misdetection. TSA is designed to achieve contextual spatial semantic feature alignment and jointly model spatio-temporal clues. We further propose a Residual Position Embedding (\textbf{RPE}) to guide TSA in focusing on the temporal saliency clues. Experimental results on five video person reID datasets demonstrate the superiority of the proposed CAViT. Additionally, the experiment conducted on VVeRI-901-trial also shows the effectiveness of CAViT for the video vehicle reID. Our code is available on \href{https://github.com/KimWu1994/CAViT}{https://github.com/KimWu1994/CAViT}.



Paperid:582
Authors:Sudipta Paul; Niluthpol Chowdhury Mithun; Amit K. Roy-Chowdhury
Title: Text-Based Temporal Localization of Novel Events
Abstract:
Recent works on text-based localization of moments have shown high accuracy on several benchmark datasets. However, these approaches are trained and evaluated relying on the assumption that the localization system, during testing, will only encounter events that are available in the training set (i.e., seen events). As a result, these models are optimized for a fixed set of seen events and they are unlikely to generalize to the practical requirement of localizing a wider range of events, some of which may be unseen. Moreover, acquiring videos and text comprising all possible scenarios for training is not practical. In this regard, this paper introduces and tackles the problem of text-based temporal localization of novel/unseen events. Our goal is to temporally localize video moments based on text queries, where both the video moments and text queries are not observed/available during training. Towards solving this problem, we formulate the inference task of text-based localization of moments as a relational prediction problem, hypothesizing a conceptual relation between semantically relevant moments, e.g., a temporally relevant moment corresponding to an unseen text query and a moment corresponding to a seen text query may contain shared concepts. The likelihood of a candidate moment to be the correct one based on an unseen text query will depend on its relevance to the moment corresponding to the semantically most relevant seen query. Empirical results on two text-based moment localization datasets show that our proposed approach can reach up to 15% absolute improvement in performance compared to existing localization approaches.



Paperid:583
Authors:Zhaopeng Dou; Zhongdao Wang; Weihua Chen; Yali Li; Shengjin Wang
Title: Reliability-Aware Prediction via Uncertainty Learning for Person Image Retrieval
Abstract:
Current person image retrieval methods have achieved great improvements in accuracy metrics. However, they rarely describe the reliability of the prediction. In this paper, we propose an Uncertainty-Aware Learning (UAL) method to remedy this issue. UAL aims at providing reliability-aware predictions by considering data uncertainty and model uncertainty simultaneously. Data uncertainty captures the “noise” inherent in the sample, while model uncertainty depicts the model’s confidence in the sample’s prediction. Specifically, in UAL, (1) we propose a sampling-free data uncertainty learning method to adaptively assign weights to different samples during training, down-weighting the low-quality ambiguous samples. (2) we leverage the Bayesian framework to model the model uncertainty by assuming the parameters of the network follow a Bernoulli distribution. (3) the data uncertainty and the model uncertainty are jointly learned in a unified network, and they serve as two fundamental criteria for the reliability assessment: if a probe is high-quality (low data uncertainty) and the model is confident in the prediction of the probe (low model uncertainty), the final ranking will be assessed as reliable. Experiments under the risk-controlled settings and the multi-query settings show the proposed reliability assessment is effective. Our method also shows superior performance on three challenging benchmarks under the vanilla single query settings. The code is available at: https://github.com/dcp15/UAL"



Paperid:584
Authors:Zhaoxi Chen; Ziwei Liu
Title: Relighting4D: Neural Relightable Human from Videos
Abstract:
Human relighting is a highly desirable yet challenging task. Existing works either require expensive one-light-at-a-time (OLAT) captured data using light stage or cannot freely change the viewpoints of the rendered body. In this work, we propose a principled framework, Relighting4D, that enables free-viewpoints relighting from only human videos under unknown illuminations. Our key insight is that the space-time varying geometry and reflectance of the human body can be decomposed as a set of neural fields of normal, occlusion, diffuse, and specular maps. These neural fields are further integrated into reflectance-aware physically based rendering, where each vertex in the neural field absorbs and reflects the light from the environment. The whole framework can be learned from videos in a self-supervised manner, with physically informed priors designed for regularization. Extensive experiments on both real and synthetic datasets demonstrate that our framework is capable of relighting dynamic human actors with free-viewpoints. Codes are available at https://github.com/FrozenBurning/Relighting4D.



Paperid:585
Authors:Zhewei Huang; Tianyuan Zhang; Wen Heng; Boxin Shi; Shuchang Zhou
Title: Real-Time Intermediate Flow Estimation for Video Frame Interpolation
Abstract:
Real-time video frame interpolation (VFI) is very useful in video processing, media players, and display devices. We propose RIFE, a Real-time Intermediate Flow Estimation algorithm for VFI. To realize a high-quality flow-based VFI method, RIFE uses a neural network named IFNet that can estimate the intermediate flows end-to-end with much faster speed. A privileged distillation scheme is designed for stable IFNet training and improve the overall performance. RIFE does not rely on pre-trained optical flow models and can support arbitrary-timestep frame interpolation with the temporal encoding input. Experiments demonstrate that RIFE achieves state-of-the-art performance on several public benchmarks. Compared with the popular SuperSlomo and DAIN methods, RIFE is 4--27 times faster and produces better results. Furthermore, RIFE can be extended to wider applications thanks to temporal encoding.



Paperid:586
Authors:Jing He; Yiyi Zhou; Qi Zhang; Jun Peng; Yunhang Shen; Xiaoshuai Sun; Chao Chen; Rongrong Ji
Title: PixelFolder: An Efficient Progressive Pixel Synthesis Network for Image Generation
Abstract:
Pixel synthesis is a promising research paradigm for image generation, which can well exploit pixel-wise prior knowledge for generation. However, existing methods still suffer from excessive memory footprint and computation overhead. In this paper, we propose a progressive pixel synthesis network towards efficient image generation, coined as PixelFolder. Specifically, PixelFolder formulates image generation as a progressive pixel regression problem and synthesizes images via a multi-stage structure, which can greatly reduce the overhead caused by large tensor transformations. In addition, we introduce novel pixel folding operations to further improve model efficiency while maintaining pixel-wise prior knowledge for end-to-end regression. With these innovative designs, we greatly reduce the expenditure of pixel synthesis, e.g., reducing 89% computation and 53% parameters compared with the latest pixel synthesis method CIPS. To validate our approach, we conduct extensive experiments on two benchmark datasets, namely FFHQ and LSUN Church. The experimental results show that with much less expenditure, PixelFolder obtains new state-of-the-art (SOTA) performance on two benchmark datasets, i.e., 3.77 FID and 2.45 FID on FFHQ and LSUN Church, respectively.Meanwhile, PixelFolder is also more efficient than the SOTA methods like StyleGAN2, reducing about 72% computation and 31% parameters, respectively. These results greatly validate the effectiveness of the proposed PixelFolder.



Paperid:587
Authors:Zhiliang Xu; Hang Zhou; Zhibin Hong; Ziwei Liu; Jiaming Liu; Zhizhi Guo; Junyu Han; Jingtuo Liu; Errui Ding; Jingdong Wang
Title: StyleSwap: Style-Based Generator Empowers Robust Face Swapping
Abstract:
Numerous attempts have been made to the task of person-agnostic face swapping given its wide applications. While existing methods mostly rely on tedious network and loss designs, they still struggle in the information balancing between the source and target faces, and tend to produce visible artifacts. In this work, we introduce a concise and effective framework named StyleSwap. Our core idea is to leverage a style-based generator to empower high-fidelity and robust face swapping, thus the generator’s advantage can be adopted for optimizing identity similarity. We identify that with only minimal modifications, a StyleGAN2 architecture can successfully handle the desired information from both source and target. Additionally, inspired by the ToRGB layers, a Swapping-Driven Mask Branch is further devised to improve information blending. Furthermore, the advantage of StyleGAN inversion can be adopted. Particularly, a Swapping-Guided ID Inversion strategy is proposed to optimize identity similarity. Extensive experiments validate that our framework generates high-quality face swapping results that outperform state-of-the-art methods both qualitatively and quantitatively.



Paperid:588
Authors:Jaskirat Singh; Liang Zheng; Cameron Smith; Jose Echevarria
Title: Paint2Pix: Interactive Painting Based Progressive Image Synthesis and Editing
Abstract:
Controllable image synthesis with user scribbles is a topic of keen interest in the computer vision community. In this paper, for the first time we study the problem of photorealistic image synthesis from incomplete and primitive human paintings. In particular, we propose a novel approach paint2pix, which learns to predict (and adapt) “what a user wants to draw” from rudimentary brushstroke inputs, by learning a mapping from the manifold of incomplete human paintings to their realistic renderings. When used in conjunction with recent works in autonomous painting agents, we show that paint2pix can be used for progressive image synthesis from scratch. During this process, paint2pix allows a novice user to progressively synthesize the desired image output, while requiring just few coarse user scribbles to accurately steer the trajectory of the synthesis process. Furthermore, we find that our approach also forms a surprisingly convenient approach for real image editing, and allows the user to perform a diverse range of custom fine-grained edits through the addition of only a few well-placed brushstrokes.



Paperid:589
Authors:Jeongmin Bae; Mingi Kwon; Youngjung Uh
Title: FurryGAN: High Quality Foreground-Aware Image Synthesis
Abstract:
Foreground-aware image synthesis aims to generate images as well as their foreground masks. A common approach is to formulate an image as a masked blending of a foreground image and a background image. It is a challenging problem because it is prone to reach the trivial solution where either image overwhelms the other, i.e., the masks become completely full or empty, and the foreground and background are not meaningfully separated. We present FurryGAN with three key components: 1) imposing both the foreground image and the composite image to be realistic, 2) designing a mask as a combination of coarse and fine masks, and 3) guiding the generator by an auxiliary mask predictor in the discriminator. Our method produces realistic images with remarkably detailed alpha masks which cover hair, fur, and whiskers in a fully unsupervised manner. Project page: \href{https://jeongminb.github.io/FurryGAN/}{https://jeongminb.github.io/FurryGAN/}"



Paperid:590
Authors:Nicolas Dufour; David Picard; Vicky Kalogeiton
Title: SCAM! Transferring Humans between Images with Semantic Cross Attention Modulation
Abstract:
A large body of recent work targets semantically conditioned image generation. Most such methods focus on the narrower task of pose transfer and ignore the more challenging task of subject transfer that consists in not only transferring the pose but also the appearance and background. In this work, we introduce SCAM (Semantic Cross Attention Modulation), a system that encodes rich and diverse information in each semantic region of the image (including foreground and background), thus achieving precise generation with emphasis on fine details. This is enabled by the Semantic Attention Transformer Encoder that extracts multiple latent vectors for each semantic region, and the corresponding generator that exploits these multiple latents by using semantic cross attention modulation. It is trained only using a reconstruction setup, while subject transfer is performed at test time. Our analysis shows that our proposed architecture is successful at encoding the diversity of appearance in each semantic region. Extensive experiments on the iDesigner, CelebAMask-HD and ADE20K datasets show that SCAM outperforms competing approaches; moreover, it sets the new state of the art on subject transfer.



Paperid:591
Authors:Yuedong Chen; Qianyi Wu; Chuanxia Zheng; Tat-Jen Cham; Jianfei Cai
Title: Sem2NeRF: Converting Single-View Semantic Masks to Neural Radiance Fields
Abstract:
Image translation and manipulation have gain increasing attention along with the rapid development of deep generative models. Although existing approaches have brought impressive results, they mainly operated in 2D space. In light of recent advances in NeRF-based 3D-aware generative models, we introduce a new task, Semantic-to-NeRF translation, that aims to reconstruct a 3D scene modelled by NeRF, conditioned on one single-view semantic mask as input. To kick-off this novel task, we propose the Sem2NeRF framework. In particular, Sem2NeRF addresses the highly challenging task by encoding the semantic mask into the latent code that controls the 3D scene representation of a pre-trained decoder. To further improve the accuracy of the mapping, we integrate a new region-aware learning strategy into the design of both the encoder and the decoder. We verify the efficacy of the proposed Sem2NeRF and demonstrate that it outperforms several strong baselines on two benchmark datasets. Code and video are available at https://donydchen.github.io/sem2nerf/"



Paperid:592
Authors:Mengping Yang; Zhe Wang; Ziqiu Chi; Wenyi Feng
Title: WaveGAN: Frequency-Aware GAN for High-Fidelity Few-Shot Image Generation
Abstract:
Existing few-shot image generation approaches typically employ fusion-based strategies, either on the image or the feature level, to produce new images. However, previous approaches struggle to synthesize high-frequency signals with fine details, deteriorating the synthesis quality. To address this, we propose WaveGAN, a frequency-aware model for few-shot image generation. Concretely, we disentangle encoded features into multiple frequency components and perform low-frequency skip connections to preserve outline and structural information. Then we alleviate the generator’s struggles of synthesizing fine details by employing high-frequency skip connections, thus providing informative frequency information to the generator. Moreover, we utilize a frequency L1-loss on the generated and real images to further impede frequency information loss. Extensive experiments demonstrate the effectiveness and advancement of our method on three datasets. Noticeably, we achieve new state-of-the-art with FID 42.17, LPIPS 0.3868, FID 30.35, LPIPS 0.5076, and FID 4.96, LPIPS 0.3822 respectively on Flower, Animal Faces, and VGGFace. GitHub: https://github.com/kobeshegu/ECCV2022_WaveGAN"



Paperid:593
Authors:Andrew Brown; Cheng-Yang Fu; Omkar Parkhi; Tamara L. Berg; Andrea Vedaldi
Title: End-to-End Visual Editing with a Generatively Pre-trained Artist
Abstract:
We consider the targeted image editing problem, namely blending a region in a source image with a driver image that specifies the desired change. Differently from prior works, we solve this problem by learning a conditional probability distribution of the edits, end-to-end in code space. Training such a model requires addressing the lack of example edits for training. To this end, we propose a self-supervised approach that simulates edits by augmenting off-the-shelf images in a target domain. The benefits are remarkable: implemented as a state-of-the-art auto-regressive transformer, our approach is simple, sidesteps difficulties with previous methods based on GAN-like priors, obtains significantly better edits, and is efficient. Furthermore, we show that different blending effects can be learned by an intuitive control of the augmentation process, with no other changes required to the model architecture. We demonstrate the superiority of this approach across several datasets in extensive quantitative and qualitative experiments, including human studies, significantly outperforming prior work.



Paperid:594
Authors:Qingyan Bai; Yinghao Xu; Jiapeng Zhu; Weihao Xia; Yujiu Yang; Yujun Shen
Title: High-Fidelity GAN Inversion with Padding Space
Abstract:
Inverting a Generative Adversarial Network (GAN) facilitates a wide range of image editing tasks using pre-trained generators. Existing methods typically employ the latent space of GANs as the inversion space yet observe the insufficient recovery of spatial details. In this work, we propose to involve the padding space of the generator to complement the latent space with spatial information. Concretely, we replace the constant padding (e.g., usually zeros) used in convolution layers with some instance-aware coefficients. In this way, the inductive bias assumed in the pre-trained model can be appropriately adapted to fit each individual image. Through learning a carefully designed encoder, we manage to improve the inversion quality both qualitatively and quantitatively, outperforming existing alternatives. We then demonstrate that such a space extension barely affects the native GAN manifold, hence we can still reuse the prior knowledge learned by GANs for various downstream applications. Beyond the editing tasks explored in prior arts, our approach allows a more flexible image manipulation, such as the separate control of face contour and facial details, and enables a novel editing manner where users can customize their own manipulations highly efficiently.



Paperid:595
Authors:Chao Xu; Jiangning Zhang; Yue Han; Guanzhong Tian; Xianfang Zeng; Ying Tai; Yabiao Wang; Chengjie Wang; Yong Liu
Title: Designing One Unified Framework for High-Fidelity Face Reenactment and Swapping
Abstract:
Face reenactment and swapping share a similar identity and attribute manipulating pattern, but most methods treat them separately, which is redundant and practical-unfriendly. In this paper, we propose an effective end-to-end unified framework to achieve both tasks. Unlike existing methods that directly utilize pre-estimated structures and do not fully exploit their potential similarity, our model sufficiently transfers identity and attribute based on learned disentangled representations to generate high-fidelity faces. Specifically, Feature Disentanglement first disentangles identity and attribute unsupervisedly. Then the proposed Attribute Transfer (AttrT) employs learned Feature Displacement Fields to transfer the attribute granularly, and Identity Transfer (IdT) explicitly models identity-related feature interaction to adaptively control the identity fusion. We joint AttrT and IdT according to their intrinsic relationship to further facilitate each task, i.e., help improve identity consistency in reenactment and attribute preservation in swapping. Extensive experiments demonstrate the superiority of our method.



Paperid:596
Authors:Wentao Yuan; Qingtian Zhu; Xiangyue Liu; Yikang Ding; Haotian Zhang; Chi Zhang
Title: Sobolev Training for Implicit Neural Representations with Approximated Image Derivatives
Abstract:
Recently, Implicit Neural Representations (INRs) parameterized by neural networks have emerged as a powerful and promising tool to represent different kinds of signals due to its continuous, differentiable properties, showing superiorities to classical discretized representations. However, the training of neural networks for INRs only utilizes input-output pairs, and the derivatives of the target output with respect to the input, which can be accessed in some cases, are usually ignored. In this paper, we propose a training paradigm for INRs whose target output is image pixels, to encode image derivatives in addition to image values in the neural network. Specifically, we use finite differences to approximate image derivatives. We show how the training paradigm can be leveraged to solve typical INRs problems, i.e., image regression and inverse rendering, and demonstrate this training paradigm can improve the data-efficiency and generalization capabilities of INRs. The code of our method is available at https://github.com/megvii-research/Sobolev_INRs.



Paperid:597
Authors:Oran Gafni; Adam Polyak; Oron Ashual; Shelly Sheynin; Devi Parikh; Yaniv Taigman
Title: Make-a-Scene: Scene-Based Text-to-Image Generation with Human Priors
Abstract:
Recent text-to-image generation methods provide a simple yet exciting conversion capability between text and image domains. While these methods have incrementally improved the generated image fidelity and text relevancy, several pivotal gaps remain unanswered, limiting applicability and quality. We propose a novel text-to-image method that addresses these gaps by (i) enabling a simple control mechanism complementary to text in the form of a scene, (ii) introducing elements that substantially improve the tokenization process by employing domain-specific knowledge over key image regions (faces and salient objects), and (iii) adapting classifier-free guidance for the transformer use case. Our model achieves state-of-the-art FID and human evaluation results, unlocking the ability to generate high fidelity images in a resolution of 512×512 pixels, significantly improving visual quality. Through scene controllability, we introduce several new capabilities: (i) Scene editing, (ii) text editing with anchor scenes, (iii) overcoming out-of-distribution text prompts, and (iv) story illustration generation, as demonstrated in \href{https://youtu.be/N4BagnXzPXY}{the story we wrote}.



Paperid:598
Authors:Yuchen Liu; Zhixin Shu; Yijun Li; Zhe Lin; Richard Zhang; S.Y. Kung
Title: 3D-FM GAN: Towards 3D-Controllable Face Manipulation
Abstract:
3D-controllable portrait synthesis has significantly advanced, thanks to breakthroughs in generative adversarial networks (GANs). However, it is still challenging to manipulate existing face images with precise 3D control. While concatenating GAN inversion and a 3D-aware, noise-to-image GAN is a straight-forward solution, it is inefficient and may lead to noticeable drop in editing quality. To fill this gap, we propose 3D-FM GAN, a novel conditional GAN framework designed specifically for 3D-controllable Face Manipulation, and does not require any tuning after the end-to-end learning phase. By carefully encoding both the input face image and a physically-based rendering of 3D edits into a StyleGAN’s latent spaces, our image generator provides high-quality, identity-preserved, 3D-controllable face manipulation. To effectively learn such novel framework, we develop two essential training strategies and a novel multiplicative co-modulation architecture that improves significantly upon naive schemes. With extensive evaluations, we show that our method outperforms the prior arts on various tasks, with better editability, stronger identity preservation, and higher photo-realism. In addition, we demonstrate a better generalizability of our design on large pose editing and out-of-domain images.



Paperid:599
Authors:Yuda Song; Hui Qian; Xin Du
Title: Multi-Curve Translator for High-Resolution Photorealistic Image Translation
Abstract:
The dominant image-to-image translation methods are based on fully convolutional networks, which extract and translate an image’s features and then reconstruct the image. However, they have unacceptable computational costs when working with high-resolution images. To this end, we present the Multi-Curve Translator (MCT), which not only predicts the translated pixels for the corresponding input pixels but also for their neighboring pixels. And if a high-resolution image is downsampled to its low-resolution version, the lost pixels are the remaining pixels’ neighboring pixels. So MCT makes it possible to feed the network only the downsampled image to perform the mapping for the full-resolution image, which can dramatically lower the computational cost. Besides, MCT is a plug-in approach that utilizes existing base models and requires only replacing their output layers. Experiments demonstrate that the MCT variants can process 4K images in real-time and achieve comparable or even better performance than the base models on various photorealistic image-to-image translation tasks.



Paperid:600
Authors:Zhiyang Yu; Yu Zhang; Xujie Xiang; Dongqing Zou; Xijun Chen; Jimmy S. Ren
Title: Deep Bayesian Video Frame Interpolation
Abstract:
We present deep Bayesian video frame interpolation, a novel approach for upsampling a low frame-rate video temporally to its higher frame-rate counterpart. Our approach learns posterior distributions of optical flows and frames to be interpolated, which is optimized via learned gradient descent for fast convergence. Each learned step is a lightweight network manipulating gradients of the log-likelihood of estimated frames and flows. Such gradients, parameterized either explicitly or implicitly, model the fidelity of current estimations when matching real image and flow distributions to explain the input observations. With this approach we show new records on 8 of 10 benchmarks, using an architecture with half the parameters of the state-of-the-art model. Code and models are publicly available at https://github.com/Oceanlib/DBVI.



Paperid:601
Authors:Xinyue Zhou; Mingyu Yin; Xinyuan Chen; Li Sun; Changxin Gao; Qingli Li
Title: Cross Attention Based Style Distribution for Controllable Person Image Synthesis
Abstract:
Controllable person image synthesis task enables a wide range of applications through explicit control over body pose and appearance. In this paper, we propose a cross attention based style distribution module that computes between the source semantic styles and target pose for pose transfer. The module intentionally selects the style represented by each semantic and distributes them according to the target pose. The attention matrix in cross attention expresses the dynamic similarities between the target pose and the source styles for all semantics. Therefore, it can be utilized to route the color and texture from the source image, and is further constrained by the target parsing map to achieve a clearer objective. At the same time, to encode the source appearance accurately, the self attention among different semantic styles is also added. The effectiveness of our model is validated quantitatively and qualitatively on pose transfer and virtual try-on tasks. Codes are available at https://github.com/xyzhouo/CASD.



Paperid:602
Authors:Marko Mihajlovic; Aayush Bansal; Michael Zollhöfer; Siyu Tang; Shunsuke Saito
Title: KeypointNeRF: Generalizing Image-Based Volumetric Avatars Using Relative Spatial Encoding of Keypoints
Abstract:
Image-based volumetric avatars using pixel-aligned features promise generalization to unseen poses and identities. Prior work leverages global spatial encodings and multi-view geometric consistency to reduce spatial ambiguity. However, global encodings often suffer from overfitting to the distribution of the training data, and it is difficult to learn multi-view consistent reconstruction from sparse views. In this work, we investigate common issues with existing spatial encodings and propose a simple yet highly effective approach to modeling high-fidelity volumetric avatars from sparse views. One of the key ideas is to encode relative spatial 3D information via sparse 3D keypoints. This approach is robust to novel view synthesis and the sparsity of viewpoints. Our approach outperforms state-of-the-art methods for head reconstruction. On body reconstruction for unseen subjects, we also achieve performance comparable to prior art that uses a parametric human body model and temporal feature aggregation. Our experiments show that a majority of errors in prior work stem from an inappropriate choice of spatial encoding and thus we suggest a new direction for high-fidelity image-based avatar modeling.



Paperid:603
Authors:Jonáš Kulhánek; Erik Derner; Torsten Sattler; Robert Babuška
Title: ViewFormer: NeRF-Free Neural Rendering from Few Images Using Transformers
Abstract:
Novel view synthesis is a long-standing problem. In this work, we consider a variant of the problem where we are given only a few context views sparsely covering a scene or an object. The goal is to predict novel viewpoints in the scene, which requires learning priors. The current state of the art is based on Neural Radiance Field (NeRF), and while achieving impressive results, the methods suffer from long training times as they require evaluating millions of 3D point samples via a neural network for each image. We propose a 2D-only method that maps multiple context views and a query pose to a new image in a single pass of a neural network. Our model uses a two-stage architecture consisting of a codebook and a transformer model. The codebook is used to embed individual images into a smaller latent space, and the transformer solves the view synthesis task in this more compact space. To train our model efficiently, we introduce a novel branching attention mechanism that allows us to use the same model not only for neural rendering but also for camera pose estimation. Experimental results on real-world scenes show that our approach is competitive compared to NeRF-based methods while not reasoning explicitly in 3D, and it is faster to train.



Paperid:604
Authors:Ziyu Chen; Chenjing Ding; Jianfei Guo; Dongliang Wang; Yikang Li; Xuan Xiao; Wei Wu; Li Song
Title: L-Tracing: Fast Light Visibility Estimation on Neural Surfaces by Sphere Tracing
Abstract:
We introduce a highly efficient light visibility estimation method, called L-Tracing, for reflectance factorization on neural implicit surfaces. Light visibility is indispensable for modeling shadows and specular of high quality on object’s surface. For neural implicit representations, former methods of computing light visibility suffer from efficiency and quality drawbacks. L-Tracing leverages the distance meaning of the Signed Distance Function(SDF), and computes the light visibility of the solid object surface according to binary geometry occlusions. We prove the linear convergence of L-Tracing algorithm and give out the theoretical lower bound of tracing iteration. Based on L-Tracing, we propose a new surface reconstruction and reflectance factorization framework. Experiments show our framework performs nearly 10x speedup on factorization, and achieves competitive albedo and relighting results with existing approaches.



Paperid:605
Authors:Qiqi Hou; Abhijay Ghildyal; Feng Liu
Title: A Perceptual Quality Metric for Video Frame Interpolation
Abstract:
Research on video frame interpolation has made significant progress in recent years. However, existing methods mostly use off-the-shelf metrics to measure the interpolation results with the exception of a few methods that employ user studies, which is time-consuming. As video frame interpolation results often exhibit unique artifacts, existing quality metrics sometimes are not consistent with human perception when measuring the interpolation results. Recent deep learning-based perceptual quality metrics are shown more consistent with human judgments, but their performance on videos is compromised since they do not consider temporal information. In this paper, we present a dedicated perceptual quality metric for measuring video frame interpolation results. Our method learns perceptual features directly from videos instead of individual frames. It compares pyramid features extracted from video frames and employs Swin Transformer blocks-based spatio-temporal modules to extract spatio-temporal information. To train our metric, we collected a new video frame interpolation quality assessment dataset. Our experiments show that our dedicated quality metric outperforms state-of-the-art methods when measuring video frame interpolation results.



Paperid:606
Authors:Mengyu Dai; Haibin Hang; Xiaoyang Guo
Title: Adaptive Feature Interpolation for Low-Shot Image Generation
Abstract:
Training of generative models especially Generative Adversarial Networks can easily diverge in low-data setting. To mitigate this issue, we propose a novel implicit data augmentation approach which facilitates stable training and synthesize high-quality samples without need of label information. Specifically, we view the discriminator as a metric embedding of the real data manifold, which offers proper distances between real data points. We then utilize information in the feature space to develop a fully unsupervised and data-driven augmentation method. Experiments on few-shot generation tasks show the proposed method significantly improve results from strong baselines with hundreds of training samples.



Paperid:607
Authors:Yi Wang; Menghan Xia; Lu Qi; Jing Shao; Yu Qiao
Title: PalGAN: Image Colorization with Palette Generative Adversarial Networks
Abstract:
Multimodal ambiguity and color bleeding remain challenging in colorization. To tackle these problems, we propose a new GAN-based colorization approach PalGAN, integrated with palette estimation and chromatic attention. To circumvent the multimodality issue, we present a new colorization formulation that estimates a probabilistic palette from the input gray image first, then conducts color assignment conditioned on the palette through a generative model. Further, we handle color bleeding with chromatic attention. It studies color affinities by considering both semantic and intensity correlation. In extensive experiments, PalGAN outperforms state-of-the-arts in quantitative evaluation and visual comparison, delivering notable diverse, contrastive, and edge-preserving appearances. With the palette design, our method enables color transfer between images even with irrelevant contexts.



Paperid:608
Authors:Long Zhuo; Guangcong Wang; Shikai Li; Wayne Wu; Ziwei Liu
Title: Fast-Vid2Vid: Spatial-Temporal Compression for Video-to-Video Synthesis
Abstract:
Video-to-Video synthesis (Vid2Vid) has achieved remarkable results on generating a photo-realistic video from a sequence of semantic maps. However, this pipeline suffers from high computational cost and long inference latency, which largely depends on two essential factors: 1) network architecture parameters, 2) sequential data stream. Recently, the parameters of image-based generative models have been significantly compressed via more efficient network architectures. Nevertheless, existing methods mainly focus on slimming network architectures and ignore the size of sequential data stream. Moreover, due to the lack of temporal coherence, image-based compression is not sufficient for the compression of video synthesis with a dynamic time series. In this paper, we present a spatial-temporal compression framework, Fast-Vid2Vid, which focuses on data aspects of generative models. It makes the first attempt at time dimension to reduce computational resources and accelerate inference. Specifically, we compress the input data stream spatially and reduce the temporal redundancy. After the proposed spatial-knowledge distillation, our model can synthesize key-frames using low-resolution data stream. Finally, Fast-Vid2Vid interpolates inter frames by motion compensation with negligible latency. On standard benchmarks, Fast-Vid2Vid achieves around real-time performance as 20 FPS and saves around 8× computational cost on a single V100 GPU.



Paperid:609
Authors:Chenjie Cao; Qiaole Dong; Yanwei Fu
Title: Learning Prior Feature and Attention Enhanced Image Inpainting
Abstract:
Many recent inpainting works have achieved impressive results by leveraging Deep Neural Networks (DNNs) to model various prior information for image restoration. Unfortunately, the performance of these methods is largely limited by the representation ability of vanilla Convolutional Neural Networks (CNNs) backbones. On the other hand, Vision Transformers (ViT) with self-supervised pre-training have shown great potential for many visual recognition and object detection tasks. A natural question is whether the inpainting task can be greatly benefited from the ViT backbone? However, it is nontrivial to directly replace the new backbones in inpainting networks, as the inpainting is an inverse problem fundamentally different from the recognition tasks. To this end, this paper incorporates the pre-training based Masked AutoEncoder (MAE) into the inpainting model, which enjoys richer informative priors to enhance the inpainting process. Moreover, we propose to use attention priors from MAE to make the inpainting model learn more long-distance dependencies between masked and unmasked regions. Sufficient ablations have been discussed about the inpainting and the self-supervised pre-training models in this paper. Besides, experiments on both Places2 and FFHQ demonstrate the effectiveness of our proposed model. Codes and pre-trained models will be released.



Paperid:610
Authors:Wenpeng Xing; Jie Chen
Title: Temporal-MPI: Enabling Multi-Plane Images for Dynamic Scene Modelling via Temporal Basis Learning
Abstract:
Novel view synthesis of static scenes has achieved remarkable advancements in producing photo-realistic results. However, key challenges remain for immersive rendering of dynamic scenes. One of the seminal image-based rendering method, the multi-plane image (MPI), produces high novel-view synthesis quality for static scenes. But modelling dynamic contents by MPI is not studied. In this paper, we propose a novel Temporal-MPI representation which is able to encode the rich 3D and dynamic variation information throughout the entire video as compact temporal basis and coefficients jointly learned. Time-instance MPI for rendering can be generated efficiently using mini-seconds by linear combinations of temporal basis and coefficients from Temporal-MPI. Thus novel-views at arbitrary time-instance will be able to be rendered via Temporal-MPI in real-time with high visual quality. Our method is trained and evaluated on Nvidia Dynamic Scene Dataset. We show that our proposed Temporal-MPI is much faster compared with other state-of-the-art dynamic scene modelling methods using MPI.



Paperid:611
Authors:Jichao Zhang; Enver Sangineto; Hao Tang; Aliaksandr Siarohin; Zhun Zhong; Nicu Sebe; Wei Wang
Title: 3D-Aware Semantic-Guided Generative Model for Human Synthesis
Abstract:
Generative Neural Radiance Field (GNeRF) models, which extract implicit 3D representations from 2D images, have recently been shown to produce realistic images representing rigid/semi-rigid objects, such as human faces or cars. However, they usually struggle to generate high-quality images representing non-rigid objects, such as the human body, which is of a great interest for many computer graphics applications. This paper proposes a 3D-aware Semantic-Guided Generative Model (3D-SGAN) for human image synthesis, which combines a GNeRF with a texture generator. The former learns an implicit 3D representation of the human body and outputs a set of 2D semantic segmentation masks. The latter transforms these semantic masks into a real image, adding a realistic texture to the human appearance. Without requiring additional 3D information, our model can learn 3D human representations with a photo-realistic controllable generation. Our experiments on the DeepFashion dataset show that 3D-SGAN significantly outperforms the most recent baselines.



Paperid:612
Authors:Yiran Xu; Badour AlBahar; Jia-Bin Huang
Title: Temporally Consistent Semantic Video Editing
Abstract:
Generative adversarial networks (GANs) have demonstrated impressive image generation quality and semantic editing capability of real images, e.g., changing object classes, modifying attributes, or transferring styles. However, applying these GAN-based editing to a video independently for each frame inevitably results in temporal flickering artifacts. We present a simple yet effective method to facilitate temporally coherent video editing. Our core idea is to minimize the temporal photometric inconsistency by optimizing both the latent code and the pre-trained generator. We evaluate the quality of our editing on different domains and GAN inversion techniques and show favorable results against the baselines.



Paperid:613
Authors:Jaeyeon Kang; Seoung Wug Oh; Seon Joo Kim
Title: Error Compensation Framework for Flow-Guided Video Inpainting
Abstract:
The key to video inpainting is to use correlation information from as many reference frames as possible. Existing flow-based propagation methods split the video synthesis process into multiple steps: flow completion -> pixel propagation -> synthesis. However, there is a significant drawback that the errors in each step continue to accumulate and amplify in the next step. To this end, we propose an Error Compensation Framework for Flow-guided Video Inpainting (ECFVI), which takes advantage of the flow-based method and offsets its weaknesses. We address the weakness with the newly designed flow completion module and the error compensation network that exploits the error guidance map. Our approach greatly improves the temporal consistency and the visual quality of the completed videos. Experimental results show the superior performance of our proposed method with the speed up of ×6, compared to the state-of-the-art methods. In addition, we present a new benchmark dataset for evaluation by supplementing the weaknesses of existing test datasets.



Paperid:614
Authors:Xueting Li; Xiaolong Wang; Ming-Hsuan Yang; Alexei A. Efros; Sifei Liu
Title: Scraping Textures from Natural Images for Synthesis and Editing
Abstract:
Existing texture synthesis methods focus on generating large texture images given a small texture sample. But such samples are typically assumed to be highly curated: rectangular, clean, and stationary. This paper aims to scrape textures directly from natural images of everyday objects and scenes, build texture models, and employ them for texture synthesis, texture editing, etc. The key idea is to jointly learn image grouping and texture modeling. The image grouping module discovers clean texture segments, each of which is represented as a texture code and a parametric sine wave by the texture modeling module. By enforcing the model to reconstruct the input image from the texture codes and sine waves, our model can be learned via self-supervision on a set of cluttered natural images, without requiring any form of annotation or clean texture images. We show that the learned texture features capture many natural and man-made textures in real images, and can be applied to tasks like texture synthesis, texture editing and texture swapping.



Paperid:615
Authors:Shuai Bai; Huiling Zhou; Zhikang Li; Chang Zhou; Hongxia Yang
Title: Single Stage Virtual Try-On via Deformable Attention Flows
Abstract:
Virtual try-on aims to generate a photo-realistic fitting result given an in-shop garment and a reference person image. Existing methods usually build up multi-stage frameworks to deal with clothes warping and body blending respectively, or rely heavily on intermediate parser-based labels which may be noisy or even inaccurate. To solve the above challenges, we propose a single-stage try-on framework by developing a novel Deformable Attention Flow (DAFlow), which applies the deformable attention scheme to multi-flow estimation. With pose keypoints as the guidance only, the self- and cross-deformable attention flows are estimated for the reference person and the garment images, respectively. By sampling multiple flow fields, the feature-level and pixel-level information from different semantic areas is simultaneously extracted and merged through the attention mechanism. It enables clothes warping and body synthesizing at the same time which leads to photo-realistic results in an end-to-end manner. Extensive experiments on two try-on datasets demonstrate that our proposed method achieves state-of-the-art performance both qualitatively and quantitatively. Furthermore, additional experiments on the other two image editing tasks illustrate the versatility of our method for multi-view synthesis and image animation. Code will be made available at https://github.com/OFA-Sys/DAFlow.



Paperid:616
Authors:Harsh Rangwani; Naman Jaswani; Tejan Karmali; Varun Jampani; R. Venkatesh Babu
Title: Improving GANs for Long-Tailed Data through Group Spectral Regularization
Abstract:
Deep long-tailed learning aims to train useful deep networks on practical, real-world imbalanced distributions, wherein most labels of the tail classes are associated with a few samples. There has been a large body of work to train discriminative models for visual recognition on long-tailed distribution. In contrast, we aim to train conditional Generative Adversarial Networks, a class of image generation models on long-tailed distributions. We find that similar to recognition, state-of-the-art methods for image generation also suffer from performance degradation on tail classes. The performance degradation is mainly due to class-specific mode collapse for tail classes, which we observe to be correlated with the spectral explosion of the conditioning parameter matrix. We propose a novel group Spectral Regularizer (gSR) that prevents the spectral explosion alleviating mode collapse, which results in diverse and plausible image generation even for tail classes. We find that gSR effectively combines with existing augmentation and regularization techniques, leading to state-of-the-art image generation performance on long-tailed data. Extensive experiments demonstrate the efficacy of our regularizer on long-tailed datasets with different degrees of imbalance. Project Page: https://sites.google.com/view/gsr-eccv22.



Paperid:617
Authors:Tejan Karmali; Rishubh Parihar; Susmit Agrawal; Harsh Rangwani; Varun Jampani; Maneesh Singh; R. Venkatesh Babu
Title: Hierarchical Semantic Regularization of Latent Spaces in StyleGANs
Abstract:
Progress in GANs has enabled the generation of high-resolution photorealistic images of astonishing quality. StyleGANs allow for compelling attribute modification on such images via mathematical operations on the latent style vectors in the W/W+ space that effectively modulate the rich hierarchical representations of the generator. Such operations have recently been generalized beyond mere attribute swapping in the original StyleGAN paper to include interpolations. In spite of many significant improvements in StyleGANs, they are still seen to generate unnatural images. The quality of the generated images is predicated on two assumptions; (a) The richness of the hierarchical representations learnt by the generator, and, (b) The linearity and smoothness of the style spaces. In this work, we propose a Hierarchical Semantic Regularizer (HSR) which aligns the hierarchical representations learnt by the generator to corresponding powerful features learnt by pretrained networks on large amounts of data. HSR is shown to not only improve generator representations but also the linearity and smoothness of the latent style spaces, leading to the generation of more natural-looking style-edited images. To demonstrate improved linearity, we propose a novel metric - Attribute Linearity Score (ALS). A significant reduction in the generation of unnatural images is corroborated by improvement in the Perceptual Path Length (PPL) metric by 16.19% averaged across different standard datasets while simultaneously improving the linearity of attribute-change in the attribute editing tasks.



Paperid:618
Authors:Seung-Jun Moon; Gyeong-Moon Park
Title: IntereStyle: Encoding an Interest Region for Robust StyleGAN Inversion
Abstract:
Recently, manipulation of real-world images has been highly elaborated along with the development of Generative Adversarial Networks (GANs) and corresponding encoders, which embed real-world images into the latent space. However, designing encoders of GAN still remains a challenging task due to the trade-off between distortion and perception. In this paper, we point out that the existing encoders try to lower the distortion not only on the interest region, e.g., human facial region but also on the uninterest region, e.g., background patterns and obstacles. However, most uninterest regions in real-world images are located at out-of-distribution (OOD), which are infeasible to be ideally reconstructed by generative models. Moreover, we empirically find that the uninterest region overlapped with the interest region can mangle the original feature of the interest region, e.g., a microphone overlapped with a facial region is inverted into the white beard. As a result, lowering the distortion of the whole image while maintaining the perceptual quality is very challenging. To overcome this trade-off, we propose a simple yet effective encoder training scheme, coined IntereStyle, which facilitates encoding by focusing on the interest region. IntereStyle steers the encoder to disentangle the encodings of the interest and uninterest regions. To this end, we filter the information of the uninterest region iteratively to regulate the negative impact of the uninterest region. We demonstrate that IntereStyle achieves both lower distortion and higher perceptual quality compared to the existing state-of-the-art encoders. Especially, our model robustly conserves features of the original images, which shows the robust image editing and style mixing results. We will release our code with the pre-trained model after the review.



Paperid:619
Authors:Guangcong Wang; Yinuo Yang; Chen Change Loy; Ziwei Liu
Title: StyleLight: HDR Panorama Generation for Lighting Estimation and Editing
Abstract:
We present a new lighting estimation and editing framework to generate high-dynamic-range (HDR) indoor panorama lighting from a single limited field-of-view (LFOV) image captured by low-dynamic-range (LDR) cameras. Existing lighting estimation methods either directly regress lighting representation parameters or decompose this problem into LFOV-to-panorama and LDR-to-HDR lighting generation sub-tasks. However, due to the partial observation, the high-dynamic-range lighting, and the intrinsic ambiguity of a scene, lighting estimation remains a challenging task. To tackle this problem, we propose a coupled dual-StyleGAN panorama synthesis network (StyleLight) that integrates LDR and HDR panorama synthesis into a unified framework. The LDR and HDR panorama synthesis share a similar generator but have separate discriminators. During inference, given an LDR LFOV image, we propose a focal-masked GAN inversion method to find its latent code by the LDR panorama synthesis branch and then synthesize the HDR panorama by the HDR panorama synthesis branch. StyleLight takes LFOV-to-panorama and LDR-to-HDR lighting generation into a unified framework and thus greatly improves lighting estimation. Extensive experiments demonstrate that our framework achieves superior performance over state-of-the-art methods on indoor lighting estimation. Notably, StyleLight also enables intuitive lighting editing on indoor HDR panoramas, which is suitable for real-world applications. Code is available at https://style-light.github.io/.



Paperid:620
Authors:Kun Lu; Rongpeng Li; Honggang Zhang
Title: Contrastive Monotonic Pixel-Level Modulation
Abstract:
Continuous one-to-many mapping is a less investigated yet important task in both low-level visions and neural image translation. In this paper, we present a new formulation called MonoPix, an unsupervised and contrastive continuous modulation model, and take a step further to enable a pixel-level spatial control which is critical but can not be properly handled previously. The key feature of this work is to model the monotonicity between controlling signals and the domain discriminator with a novel contrastive modulation framework and corresponding monotonicity constraints. We have also introduced a selective inference strategy with logarithmic approximation complexity and support fast domain adaptations. The state-of-the-art performance is validated on a variety of continuous mapping tasks, including AFHQ cat-dog and Yosemite summer-winter translation. The introduced approach also helps to provide a new solution for many low-level tasks like low-light enhancement and natural noise generation, which is beyond the long-established practice of one-to-one training and inference. Code is available at https://github.com/lukun199/MonoPix.



Paperid:621
Authors:Wentao Shangguan; Yu Sun; Weijie Gan; Ulugbek S. Kamilov
Title: Learning Cross-Video Neural Representations for High-Quality Frame Interpolation
Abstract:
This paper considers the problem of temporal video interpolation, where the goal is to synthesize a new video frame given its two neighbors. We propose Cross-Video Neural Representation (CURE) as the first video interpolation method based on neural fields (NF). NF refers to the recent class of methods for the neural representation of complex 3D scenes that has seen widespread success and application across computer vision. CURE represents the video as a continuous function parameterized by a coordinate-based neural network, whose inputs are the spatiotemporal coordinates and outputs are the corresponding RGB values. CURE introduces a new architecture that conditions the neural network on the input frames for imposing space-time consistency in the synthesized video. This not only improves the final interpolation quality, but also enables CURE to learn a prior across multiple videos. Experimental evaluations show that CURE achieves the state-of-the-art performance on video interpolation on several benchmark datasets.



Paperid:622
Authors:Bowei Chen; Tiancheng Zhi; Martial Hebert; Srinivasa G. Narasimhan
Title: Learning Continuous Implicit Representation for Near-Periodic Patterns
Abstract:
Near-Periodic Patterns (NPP) are ubiquitous in man-made scenes and are composed of tiled motifs with appearance differences caused by lighting, defects, or design elements. A good NPP representation is useful for many applications including image completion, segmentation, and geometric remapping. But representing NPP is challenging because it needs to maintain global consistency (tiled motifs layout) while preserving local variations (appearance differences). Methods trained on general scenes using a large dataset or single-image optimization struggle to satisfy these constraints, while methods that explicitly model periodicity are not robust to periodicity detection errors. To address these challenges, we learn a neural implicit representation using a coordinate-based MLP with single image optimization. We design an input feature warping module and a periodicity-guided patch loss to handle both global consistency and local variations. To further improve the robustness, we introduce a periodicity proposal module to search and use multiple candidate periodicities in our pipeline. We demonstrate the effectiveness of our method on more than 500 images of building facades, friezes, wallpapers, ground, and Mondrian patterns on single and multi-planar scenes.



Paperid:623
Authors:Jiachen Liu; Yuan Xue; Jose Duarte; Krishnendra Shekhawat; Zihan Zhou; Xiaolei Huang
Title: End-to-End Graph-Constrained Vectorized Floorplan Generation with Panoptic Refinement
Abstract:
The automatic generation of floorplans given user inputs has great potential in architectural design and has recently been explored in the computer vision community. However, the majority of existing methods synthesize floorplans in the format of rasterized images, which are difficult to edit or customize. In this paper, we aim to synthesize floorplans as sequences of 1-D vectors, which eases user interaction and design customization. To generate high fidelity vectorized floorplans, we propose a novel two-stage framework, including a draft stage and a multi-round refining stage. In the first stage, we encode the room connectivity graph input by users with a graph convolutional network (GCN), then apply an autoregressive transformer network to generate an initial floorplan sequence. To polish the initial design and generate more visually appealing floorplans, we further propose a novel panoptic refinement network(PRN) composed of a GCN and a transformer network. The PRN takes the initial generated sequence as input and refines the floorplan design while encouraging the correct room connectivity with our proposed geometric loss. We have conducted extensive experiments on a real-world floorplan dataset, and the results show that our method achieves state-of-the-art performance under different settings and evaluation metrics.



Paperid:624
Authors:Chaerin Kong; Jeesoo Kim; Donghoon Han; Nojun Kwak
Title: Few-Shot Image Generation with Mixup-Based Distance Learning
Abstract:
Producing diverse and realistic images with generative models such as GANs typically requires large scale training with vast amount of images. GANs trained with limited data can easily memorize few training samples and display undesirable properties like ""stairlike"" latent space where interpolation in the latent space yields discontinuous transitions in the output space. In this work, we consider a challenging task of pretraining-free few-shot image synthesis, and seek to train existing generative models with minimal overfitting and mode collapse. We propose mixup-based distance regularization on the feature space of both a generator and the counterpart discriminator that encourages the two players to reason not only about the scarce observed data points but the relative distances in the feature space they reside. Qualitative and quantitative evaluation on diverse datasets demonstrates that our method is generally applicable to existing models to enhance both fidelity and diversity under few-shot setting. Codes are available.



Paperid:625
Authors:Xu Yao; Alasdair Newson; Yann Gousseau; Pierre Hellier
Title: A Style-Based GAN Encoder for High Fidelity Reconstruction of Images and Videos
Abstract:
We present a new encoder architecture for GAN inversion. The task is to reconstruct a real image from the latent space of a pre-trained Generative Adversarial Network (GAN). Unlike previous encoder-based methods which predict only a latent code from a real image, the proposed encoder maps the given image to both a latent code and a feature tensor, simultaneously. The feature tensor is key for accurate inversion, which helps to obtain better perceptual quality and lower reconstruction error. We conduct extensive experiments for several style-based generators pre-trained on different data domains. Our method is the first feed-forward encoder to include the feature tensor in the inversion, outperforming the state-of-the-art encoder-based methods for GAN inversion. Additionally, experiments on video inversion show that our method yields a more accurate and stable inversion for videos. This offers the possibility to perform real-time editing in videos.



Paperid:626
Authors:Ziqiang Li; Chaoyue Wang; Heliang Zheng; Jing Zhang; Bin Li
Title: FakeCLR: Exploring Contrastive Learning for Solving Latent Discontinuity in Data-Efficient GANs
Abstract:
Data-Efficient GANs (DE-GANs), which aim to learn generative models with a limited amount of training data, encounter several challenges for generating high-quality samples. Since data augmentation strategies have largely alleviated the training instability, how to further improve the generative performance of DE-GANs becomes a hotspot. Recently, contrastive learning has shown the great potential of increasing the synthesis quality of DE-GANs, yet related principles are not well explored. In this paper, we revisit and compare different contrastive learning strategies in DE-GANs, and identify (i) the current bottleneck of generative performance is the discontinuity of latent space; (ii) compared to other contrastive learning strategies, Instance-perturbation works towards latent space continuity, which brings the major improvement to DE-GANs. Based on these observations, we propose FakeCLR, which only applies contrastive learning on perturbed fake samples, and devises three related training techniques: Noise-related Latent Augmentation, Diversity-aware Queue, and Forgetting Factor of Queue. Our experimental results manifest the new state of the arts on both few-shot generation and limited-data generation. On multiple datasets, FakeCLR acquires more than 15% FID improvement compared to existing DE-GANs. Code is available at https://github.com/iceli1007/FakeCLR.



Paperid:627
Authors:Dave Epstein; Taesung Park; Richard Zhang; Eli Shechtman; Alexei A. Efros
Title: BlobGAN: Spatially Disentangled Scene Representations
Abstract:
We propose an unsupervised, mid-level representation for a generative model of scenes. The representation is mid-level in that it is neither per-pixel nor per-image; rather, scenes are modeled as a collection of spatial, depth-ordered ""blobs"" of features. Blobs are differentiably placed onto a feature grid that is decoded into an image by a generative adversarial network. Due to the spatial uniformity of blobs and the locality inherent to convolution, our network learns to associate different blobs with different entities in a scene and to arrange these blobs to capture scene layout. We demonstrate this emergent behavior by showing that, despite training without any supervision, our method enables applications such as easy manipulation of objects within a scene (e.g., moving, removing, and restyling furniture), creation of feasible scenes given constraints (e.g., plausible rooms with drawers at a particular location), and parsing of real-world images into constituent parts. On a challenging multi-category dataset of indoor scenes, BlobGAN outperforms StyleGAN2 in image quality as measured by FID. See our project page for video results and interactive demo: https://www.dave.ml/blobgan.



Paperid:628
Authors:Zhiwen Fan; Yifan Jiang; Peihao Wang; Xinyu Gong; Dejia Xu; Zhangyang Wang
Title: Unified Implicit Neural Stylization
Abstract:
Representing visual signals by implicit neural representation (INR) has prevailed among many vision tasks. Its potential for editing/processing given signals remains less explored. This work explores a new intriguing direction: training a stylized implicit representation, using a generalized approach that can apply to various 2D and 3D scenarios. We conduct a pilot study on a variety of INRs, including 2D coordinate-based representation, signed distance function, and neural radiance field. Our solution is a Unified Implicit Neural Stylization framework, dubbed INS. In contrary to vanilla INR, INS decouples the ordinary implicit function into a style implicit module and a content implicit module, in order to separately encode the representations from the style image and input scenes. An amalgamation module is then applied to aggregate these information and synthesize the stylized output. To regularize the geometry in 3D scenes, we propose a novel self-distillation geometry consistency loss which preserves the geometry fidelity of the stylized scenes. Comprehensive experiments are conducted on multiple task settings, including fitting images using MLPs, stylization for implicit surfaces and sylized novel view synthesis using neural radiance. We further demonstrate that the learned representation is continuous not only spatially but also style-wise, leading to effortlessly interpolating between different styles and generating images with new mixed styles. Please refer to the video on our project page for more view synthesis results: https://zhiwenfan.github.io/INS.



Paperid:629
Authors:Xuyang Guo; Meina Kan; Tianle Chen; Shiguang Shan
Title: GAN with Multivariate Disentangling for Controllable Hair Editing
Abstract:
Hair editing is an essential but challenging task in portrait editing considering the complex geometry and material of hair. Existing methods have achieved promising results by editing through a reference photo, user-painted mask, or guiding strokes. However, when a user provides no reference photo or hardly paints a desirable mask, these works fail to edit. Going a further step, we propose an efficiently controllable method that can provide a set of sliding bars to do continuous and fine hair editing. Meanwhile, it also naturally supports discrete editing through a reference photo and user-painted mask. Specifically, we propose a generative adversarial network with a multivariate Gaussian disentangling module. Firstly, an encoder disentangles the hair’s major attributes, including color, texture, and shape, to separate latent representations. The latent representation of each attribute is modeled as a standard multivariate Gaussian distribution, to make each dimension of an attribute be changed continuously and finely. Benefiting from the Gaussian distribution, any manual editing including sliding a bar, providing a reference photo, and painting a mask can be easily made, which is flexible and friendly for users to interact with. Finally, with changed latent representations, the decoder outputs a portrait with the edited hair. Experiments show that our method can edit each attribute’s dimension continuously and separately. Besides, when editing through reference images and painted masks like existing methods, our method achieves comparable results in terms of FID and visualization. Codes can be found at https://github.com/XuyangGuo/CtrlHair.



Paperid:630
Authors:Keshigeyan Chandrasegaran; Ngoc-Trung Tran; Alexander Binder; Ngai-Man Cheung
Title: Discovering Transferable Forensic Features for CNN-Generated Images Detection
Abstract:
Visual counterfeits are increasingly causing an existential conundrum in mainstream media with rapid evolution in neural image synthesis methods. Though detection of such counterfeits has been a taxing problem in the image forensics community, a recent class of forensic detectors -- universal detectors -- are able to surprisingly spot counterfeit images regardless of generator architectures, loss functions, training datasets, and resolutions. This intriguing property suggests the possible existence of transferable forensic features (T-FF) in universal detectors. In this work, we conduct the first analytical study to discover and understand T-FF in universal detectors. Our contributions are 2-fold: 1) We propose a novel forensic feature relevance statistic (FF-RS) to quantify and discover T-FF in universal detectors and, 2) Our qualitative and quantitative investigations uncover an unexpected finding: color is a critical T-FF in universal detectors. Code and models are available at https://keshik6.github.io/transferable-forensic-features/"



Paperid:631
Authors:Zhanghan Ke; Chunyi Sun; Lei Zhu; Ke Xu; Rynson W.H. Lau
Title: Harmonizer: Learning to Perform White-Box Image and Video Harmonization
Abstract:
Recent works on image harmonization solve the problem as a pixel-wise image translation task via large autoencoders. They have unsatisfactory performances and slow inference speeds when dealing with high-resolution images. In this work, we observe that adjusting the input arguments of basic image filters, e.g., brightness and contrast, is sufficient for humans to produce realistic images from the composite ones. Hence, we frame image harmonization as an image-level regression problem to learn the arguments of the filters that humans use for the task. We present a Harmonizer framework for image harmonization. Unlike prior methods that are based on black-box autoencoders, Harmonizer contains a neural network for filter argument prediction and several white-box filters (based on the predicted arguments) for image harmonization. We also introduce a cascade regressor and a dynamic loss strategy for Harmonizer to learn filter arguments more stably and precisely. Since our network only outputs image-level arguments and the filters we used are efficient, Harmonizer is much lighter and faster than existing methods. Comprehensive experiments demonstrate that Harmonizer surpasses existing methods notably, especially with high-resolution inputs. Finally, we apply Harmonizer to video harmonization, which achieves consistent results across frames and 56 fps at 1080P resolution.



Paperid:632
Authors:Omer Bar-Tal; Dolev Ofri-Amar; Rafail Fridman; Yoni Kasten; Tali Dekel
Title: Text2LIVE: Text-Driven Layered Image and Video Editing
Abstract:
We present a method for zero-shot, text-driven editing of natural images and videos. Given an image or a video and a text prompt, our goal is to edit the appearance of existing objects (e.g., texture) or augment the scene with visual effects (e.g., smoke, fire) in a semantic manner. We train a generator on an \emph{internal dataset}, extracted from a single input, while leveraging an \emph{external} pretrained CLIP model to impose our losses. Rather than directly generating the edited output, our key idea is to generate an \emph{edit layer} (color+opacity) that is composited over the input. This allows us to control the generation and maintain high fidelity to the input via novel text-driven losses applied directly to the edit layer. Our method neither relies on a pretrained generator nor requires user-provided masks. We demonstrate localized, semantic edits on high-resolution images and videos across a variety of objects and scenes.



Paperid:633
Authors:Jian Zhang; Jinchi Huang; Bowen Cai; Huan Fu; Mingming Gong; Chaohui Wang; Jiaming Wang; Hongchen Luo; Rongfei Jia; Binqiang Zhao; Xing Tang
Title: Digging into Radiance Grid for Real-Time View Synthesis with Detail Preservation
Abstract:
Neural Radiance Fields (NeRF) [31] series are impressive in representing scenes and synthesizing high-quality novel views. However, most previous works fail to preserve texture details and suffer from slow training speed. A recent method SNeRG [11] demonstrates that baking a trained NeRF as a Sparse Neural Radiance Grid enables real-time view synthesis with slight scarification of rendering quality. In this paper, we dig into the Radiance Grid representation and present a set of improvements, which together result in boosted performance in terms of both speed and quality. First, we propose an HieRarchical Sparse Radiance Grid (HrSRG) representation that has higher voxel resolution for informative spaces and fewer voxels for other spaces. HrSRG leverages a hierarchical voxel grid building process inspired by [30, 55], and can describe a scene at high resolution without excessive memory footprint. Furthermore, we show that directly optimizing the voxel grid leads to surprisingly good texture details in rendered images. This direct optimization is memory-friendly and requires multiple orders of magnitude less time than conventional NeRFs as it only involves a tiny MLP. Finally, we find that a critical factor that prevents fine details restoration is the misaligned 2D pixels among images caused by camera pose errors. We propose to use the perceptual loss to add tolerance to misalignments, leading to the improved visual quality of rendered images.



Paperid:634
Authors:Jianglin Fu; Shikai Li; Yuming Jiang; Kwan-Yee Lin; Chen Qian; Chen Change Loy; Wayne Wu; Ziwei Liu
Title: StyleGAN-Human: A Data-Centric Odyssey of Human Generation
Abstract:
Unconditional human image generation is an important task in vision and graphics, enabling various applications in the creative industry. Existing studies in this field mainly focus on “network engineering” such as designing new components and objective functions. This work takes a data-centric perspective and investigates multiple critical aspects in “data engineering”, which we believe would complement the current practice. To facilitate a comprehensive study, we collect and annotate a large-scale human image dataset with over 230K samples capturing diverse poses and textures. Equipped with this large dataset, we rigorously investigate three essential factors in data engineering for StyleGAN-based human generation, namely data size, data distribution, and data alignment. Extensive experiments reveal several valuable observations w.r.t. these aspects: 1) Large-scale data, more than 40K images, are needed to train a high-fidelity unconditional human generation model with a vanilla StyleGAN. 2) A balanced training set helps improve the generation quality with rare face poses compared to the long-tailed counterpart, whereas simply balancing the clothing texture distribution does not effectively bring an improvement. 3) Human GAN models that employ body centers for alignment outperform models trained using face centers or pelvis points as alignment anchors. In addition, a model zoo and human editing applications are demonstrated to facilitate future research in the community.



Paperid:635
Authors:Xiaozhong Ji; Boyuan Jiang; Donghao Luo; Guangpin Tao; Wenqing Chu; Zhifeng Xie; Chengjie Wang; Ying Tai
Title: ColorFormer: Image Colorization via Color Memory Assisted Hybrid-Attention Transformer
Abstract:
Automatic image colorization is a challenging task that attracts a lot of research interest. Previous methods employing deep neural networks have produced impressive results. However, these colorization images are still unsatisfactory and far from practical applications. The reason is that semantic consistency and color richness are two key elements ignored by existing methods. In this work, we propose an automatic image colorization method via color memory assisted hybrid-attention transformer, namely ColorFormer. Our network consists of a transformer-based encoder and a color memory decoder. The core module of the encoder is our proposed global-local hybrid attention operation, which improves the ability to capture global receptive field dependencies. With the strong power to model contextual semantic information of grayscale image in different scenes, our network can produce semantic-consistent colorization results. In decoder part, we design a color memory module which stores various semantic-color mapping for image-adaptive queries. The queried color priors are used as reference to help the decoder produce more vivid and diverse results. Experimental results show that our method can generate more realistic and semantically matched color images compared with state-of-the-art methods. Moreover, owing to the proposed end-to-end architecture, the inference speed reaches 40FPS on a V100 GPU, which meets the real-time requirement.



Paperid:636
Authors:Guohao Ying; Xin He; Bin Gao; Bo Han; Xiaowen Chu
Title: EAGAN: Efficient Two-Stage Evolutionary Architecture Search for GANs
Abstract:
Generative adversarial networks (GANs) have proven successful in image generation tasks. However, GAN training is inherently unstable. Although many works try to stabilize it by manually modifying GAN architecture, it requires much expertise. Neural architecture search (NAS) has become an attractive solution to search GANs automatically. The early NAS-GANs search only generators to reduce search complexity but lead to a sub-optimal GAN. Some recent works try to search both generator (G) and discriminator (D), but they suffer from the instability of GAN training. To alleviate the instability, we propose an efficient two-stage evolutionary algorithm-based NAS framework to search GANs, namely EAGAN. We decouple the search of G and D into two stages, where stage-1 searches G with a fixed D and adopts the many-to-one training strategy, and stage-2 searches D with the optimal G found in stage-1 and adopts the one-to-one training and weight-resetting strategies to enhance the stability of GAN training. Both stages use the non-dominated sorting method to produce Pareto-front architectures under multiple objectives (e.g., model size, Inception Score (IS), and Fréchet Inception Distance (FID)). EAGAN is applied to the unconditional image generation task and can efficiently finish the search on the CIFAR-10 dataset in 1.2 GPU days. Our searched GANs achieve competitive results (IS=8.81±0.10, FID=9.91) on the CIFAR-10 dataset and surpass prior NAS-GANs on the STL-10 dataset (IS=10.44±0.087, FID=22.18). Source code: https://github.com/marsggbo/EAGAN.



Paperid:637
Authors:Dae-Young Song; Geonsoo Lee; HeeKyung Lee; Gi-Mun Um; Donghyeon Cho
Title: Weakly-Supervised Stitching Network for Real-World Panoramic Image Generation
Abstract:
Recently, there has been growing attention on an end-to-end deep learning-based stitching model. However, the most challenging point in deep learning-based stitching is to obtain pairs of input images with a narrow field of view and ground truth images with a wide field of view captured from real-world scenes. To overcome this difficulty, we develop a weakly-supervised learning mechanism to train the stitching model without requiring genuine ground truth images. In addition, we propose a stitching model that takes multiple real-world fisheye images as inputs and creates a 360$^{\circ}$ output image in an equirectangular projection format. In particular, our model consists of color consistency corrections, warping, and blending, and is trained by perceptual and SSIM losses. The effectiveness of the proposed algorithm is verified on two real-world stitching datasets.



Paperid:638
Authors:Songhua Liu; Jingwen Ye; Sucheng Ren; Xinchao Wang
Title: DynaST: Dynamic Sparse Transformer for Exemplar-Guided Image Generation
Abstract:
One key challenge of exemplar-guided image generation lies in establishing fine-grained correspondences between input and guided images. Prior approaches, despite the promising results, have relied on either estimating dense attention to compute per-point matching, which is limited to only coarse scales due to the quadratic memory cost, or fixing the number of correspondences to achieve linear complexity, which lacks flexibility. In this paper, we propose a dynamic sparse attention based Transformer model, termed Dynamic Sparse Transformer (DynaST), to achieve fine-level matching with favorable efficiency. The heart of our approach is a novel dynamic-attention unit, dedicated to covering the variation on the optimal number of tokens one position should focus on. Specifically, DynaST leverages the multi-layer nature of Transformer structure, and performs the dynamic attention scheme in a cascaded manner to refine matching results and synthesize visually-pleasing outputs. In addition, we introduce a unified training objective for DynaST, making it a versatile reference-based image translation framework for both supervised and unsupervised scenarios. Extensive experiments on three applications, pose-guided person image generation, edge-based face synthesis, and undistorted image style transfer, demonstrate that DynaST achieves superior performance in local details, outperforming the state of the art while reducing the computational cost significantly. Our code is available \href{https://github.com/Huage001/DynaST}{here}.



Paperid:639
Authors:Xun Huang; Arun Mallya; Ting-Chun Wang; Ming-Yu Liu
Title: Multimodal Conditional Image Synthesis with Product-of-Experts GANs
Abstract:
Existing conditional image synthesis frameworks generate images based on user inputs in a single modality, such as text, segmentation, or sketch. They do not allow users to simultaneously use inputs in multiple modalities to control the image synthesis output. This reduces their practicality as multimodal inputs are more expressive and complement each other. To address this limitation, we propose the Product-of-Experts Generative Adversarial Networks (PoE-GAN) framework, which can synthesize images conditioned on multiple input modalities or any subset of them, even the empty set. We achieve this capability with a single trained model. PoE-GAN consists of a product-of-experts generator and a multimodal multiscale projection discriminator. Through our carefully designed training scheme, PoE-GAN learns to synthesize images with high quality and diversity. Besides advancing the state of the art in multimodal conditional image synthesis, PoE-GAN also outperforms the best existing unimodal conditional image synthesis approaches when tested in the unimodal setting. The project website is available at https://deepimagination.github.io/PoE-GAN/.



Paperid:640
Authors:Fangneng Zhan; Yingchen Yu; Rongliang Wu; Jiahui Zhang; Kaiwen Cui; Changgong Zhang; Shijian Lu
Title: Auto-Regressive Image Synthesis with Integrated Quantization
Abstract:
Deep generative models have achieved conspicuous progress in realistic image synthesis with multifarious conditional inputs, while generating diverse yet high-fidelity images remains a grand challenge in conditional image generation. This paper presents a versatile framework for conditional image generation which incorporates the inductive bias of CNNs and powerful sequence modeling of auto-regression that naturally leads to diverse image generation. Instead of independently quantizing the features of multiple domains as in prior research, we design an integrated quantization scheme with a variational regularizer that mingles the feature discretization in multiple domains, and markedly boosts the auto-regressive modeling performance. Notably, the variational regularizer enables to regularize feature distributions in incomparable latent spaces by penalizing the intra-domain variations of distributions. In addition, we design a reliable Gumbel sampling strategy that allows to incorporate distribution uncertainty into the auto-regressive training procedure. The reliable Gumbel sampling substantially mitigates the exposure bias that often incurs misalignment between the training and inference stages and severely impairs the inference performance. Extensive experiments over multiple conditional image generation tasks show that our method achieves superior diverse image generation performance qualitatively and quantitatively as compared with the state-of-the-art.



Paperid:641
Authors:Min Jin Chong; David Forsyth
Title: JoJoGAN: One Shot Face Stylization
Abstract:
A style mapper applies some fixed style to its input images (so, for example, taking faces to cartoons). This paper describes a simple procedure -- JoJoGAN -- to learn a style mapper from a single example of the style. JoJoGAN uses a GAN inversion procedure and StyleGAN’s style-mixing property to produce a substantial paired dataset from a single example style. The paired dataset is then used to fine-tune a StyleGAN. An image can then be style mapped by GAN-inversion followed by the fine-tuned StyleGAN. JoJoGAN needs just one reference and as little as 30 seconds of training time. JoJoGAN can use extreme style references (say, animal faces) successfully. Furthermore, one can control what aspects of the style are used and how much of the style is applied. Qualitative and quantitative evaluation show that JoJoGAN produces high quality high resolution images that vastly outperform the current state-of-the-art.



Paperid:642
Authors:Yusuf Dalva; Said Fahri Altındiş; Aysegul Dundar
Title: VecGAN: Image-to-Image Translation with Interpretable Latent Directions
Abstract:
We propose VecGAN, an image-to-image translation framework for facial attribute editing with interpretable latent directions. Facial attribute editing task faces the challenges of precise attribute editing with controllable strength and preservation of the other attributes of an image. For this goal, we design the attribute editing by latent space factorization and for each attribute, we learn a linear direction that is orthogonal to the others. The other component is the controllable strength of the change, a scalar value. In our framework, this scalar can be either sampled or encoded from a reference image by projection. Our work is inspired by the latent space factorization works of fixed pretrained GANs. However, while those models cannot be trained end-to-end and struggle to edit encoded images precisely, VecGAN is end-to-end trained for image translation task and successful at editing an attribute while preserving the others. Our extensive experiments show that VecGAN achieves significant improvements over state-of-the-arts for both local and global edits.



Paperid:643
Authors:Lucy Chai; Michaël Gharbi; Eli Shechtman; Phillip Isola; Richard Zhang
Title: Any-Resolution Training for High-Resolution Image Synthesis
Abstract:
Generative models operate at fixed resolution, even though natural images come in a variety of sizes. As high-resolution details are downsampled away and low-resolution images are discarded altogether, precious supervision is lost. We argue that every pixel matters and create datasets with variable-size images, collected at their native resolutions. To take advantage of varied-size data, we introduce continuous-scale training, a process that samples patches at random scales to train a new generator with variable output resolutions. First, conditioning the generator on a target scale allows us to generate higher resolution images than previously possible, without adding layers to the model. Second, by conditioning on continuous coordinates, we can sample patches that still obey a consistent global layout, which also allows for scalable training at higher resolutions. Controlled FFHQ experiments show that our method can take advantage of multi-resolution training data better than discrete multi-scale approaches, achieving better FID scores and cleaner high-frequency details. We also train on other natural image domains including churches, mountains, and birds, and demonstrate arbitrary scale synthesis with both coherent global layouts and realistic local details, going beyond 2K resolution in our experiments. Our project page is available at: https://chail.github.io/anyres-gan/.



Paperid:644
Authors:Zijie Wu; Zhen Zhu; Junping Du; Xiang Bai
Title: CCPL: Contrastive Coherence Preserving Loss for Versatile Style Transfer
Abstract:
In this paper, we aim to devise a universally versatile style transfer method capable of performing artistic, photo-realistic, and video style transfer jointly, without seeing videos during training. Previous single-frame methods assume a strong constraint on the whole image to maintain temporal consistency, which could be violated in many cases. Instead, we make a mild and reasonable assumption that global inconsistency is dominated by local inconsistencies and devise a generic Contrastive Coherence Preserving Loss (CCPL) applied to local patches. CCPL can preserve the coherence of the content source during style transfer without degrading stylization. Moreover, it owns a neighbor-regulating mechanism, resulting in a vast reduction of local distortions and considerable visual quality improvement. Aside from its superior performance on versatile style transfer, it can be easily extended to other tasks, such as image-to-image translation. Besides, to better fuse content and style features, we propose Simple Covariance Transformation (SCT) to effectively align second-order statistics of the content feature with the style feature. Experiments demonstrate the effectiveness of the resulting model for versatile style transfer, when armed with CCPL.



Paperid:645
Authors:Yung-Han Ho; Chih-Peng Chang; Peng-Yu Chen; Alessandro Gnutti; Wen-Hsiao Peng
Title: CANF-VC: Conditional Augmented Normalizing Flows for Video Compression
Abstract:
This paper presents an end-to-end learning-based video compression system, termed CANF-VC, based on conditional augmented normalizing flows (CANF). Most learned video compression systems adopt the same hybrid-based coding architecture as the traditional codecs. Recent research on conditional coding has shown the sub-optimality of the hybrid-based coding and opens up opportunities for deep generative models to take a key role in creating new coding frameworks. CANF-VC represents a new attempt that leverages the conditional ANF to learn a video generative model for conditional inter-frame coding. We choose ANF because it is a special type of generative model, which includes variational autoencoder as a special case and is able to achieve better expressiveness. CANF-VC also extends the idea of conditional coding to motion coding, forming a purely conditional coding framework. Extensive experimental results on commonly used datasets confirm the superiority of CANF-VC to the state-of-the-art methods. The source code of CANF-VC is available at https://github.com/NYCU-MAPL/CANF-VC.



Paperid:646
Authors:Fangneng Zhan; Yingchen Yu; Rongliang Wu; Jiahui Zhang; Kaiwen Cui; Aoran Xiao; Shijian Lu; Chunyan Miao
Title: Bi-Level Feature Alignment for Versatile Image Translation and Manipulation
Abstract:
Generative adversarial networks (GANs) have achieved great success in image translation and manipulation. However, high-fidelity image generation with faithful style control remains a grand challenge in computer vision. This paper presents a versatile image translation and manipulation framework that achieves accurate semantic and style guidance in image generation by explicitly building a correspondence. To handle the quadratic complexity incurred by building the dense correspondences, we introduce a bi-level feature alignment strategy that adopts a top-$k$ operation to rank block-wise features followed by dense attention between block features which reduces memory cost substantially. As the top-$k$ operation involves index swapping which precludes the gradient propagation, we approximate the non-differentiable top-$k$ operation with a regularized earth mover’s problem so that its gradient can be effectively back-propagated. In addition, we design a novel semantic position encoding mechanism that builds up coordinate for each individual semantic region to preserve texture structures while building correspondences. Further, we design a novel confidence feature injection module which mitigates mismatch problem by fusing features adaptively according to the reliability of built correspondences. Extensive experiments show that our method achieves superior performance qualitatively and quantitatively as compared with the state-of-the-art.



Paperid:647
Authors:Yongsheng Yu; Libo Zhang; Heng Fan; Tiejian Luo
Title: High-Fidelity Image Inpainting with GAN Inversion
Abstract:
Image inpainting seeks a semantically consistent way to recover the corrupted image in the light of its unmasked content. Previous approaches usually reuse the well-trained GAN as effective prior to generate realistic patches for missing holes with GAN inversion. Nevertheless, the ignorance of hard constraint in these algorithms may yield the gap between GAN inversion and image inpainting. Addressing this problem, in this paper we devise a novel GAN inversion model for image inpainting, dubbed {\it InvertFill}, mainly consisting of an encoder with a pre-modulation module and a GAN generator with F&W+ latent space. Within the encoder, the pre-modulation network leverages multi-scale structures to encode more discriminative semantic into style vectors. In order to bridge the gap between GAN inversion and image inpainting, F&W+ latent space is proposed to eliminate glaring color discrepancy and semantic inconsistency. To reconstruct faithful and photorealistic images, a simple yet effective Soft-update Mean Latent module is designed to capture more diverse in-domain patterns that synthesize high-fidelity textures for large corruptions. Comprehensive experiments on four challenging dataset, including Places2, CelebA-HQ, MetFaces, and Scenery, demonstrate that our InvertFill outperforms the advanced approaches qualitatively and quantitatively and supports the completion of out-of-domain images well. All codes, models and results will be made available upon the acceptance.



Paperid:648
Authors:Yan Hong; Li Niu; Jianfu Zhang; Liqing Zhang
Title: DeltaGAN: Towards Diverse Few-Shot Image Generation with Sample-Specific Delta
Abstract:
Learning to generate new images for a novel category based on only a few images, named as few-shot image generation, has attracted increasing research interest. Several state-of-the-art works have yielded impressive results, but the diversity is still limited. In this work, we propose a novel Delta Generative Adversarial Network (DeltaGAN), which consists of a reconstruction subnetwork and a generation subnetwork. The reconstruction subnetwork captures intra-category transformation, i.e., delta, between same-category pairs. The generation subnetwork generates sample-specific “delta” for an input image, which is combined with this input image to generate a new image within the same category. Besides, an adversarial delta matching loss is designed to link the above two subnetworks together. Extensive experiments on six benchmark datasets demonstrate the effectiveness of our proposed method.



Paperid:649
Authors:Haitian Zheng; Zhe Lin; Jingwan Lu; Scott Cohen; Eli Shechtman; Connelly Barnes; Jianming Zhang; Ning Xu; Sohrab Amirghodsi; Jiebo Luo
Title: Image Inpainting with Cascaded Modulation GAN and Object-Aware Training
Abstract:
Recent image inpainting methods have made great progress but often struggle to generate plausible image structures when dealing with large holes in complex images. This is partially due to the lack of effective network structures that can capture both the long-range dependency and high-level semantics of an image. We propose cascaded modulation GAN (CM-GAN), a new network design consisting of an encoder with Fourier convolution blocks that extract multi-scale feature representations from the input image with holes and a dual-stream decoder with a novel cascaded global-spatial modulation block at each scale level. In each decoder block, global modulation is first applied to perform coarse and semantic-aware structure synthesis, followed by spatial modulation to further adjust the feature map in a spatially adaptive fashion. In addition, we design an object-aware training scheme to prevent the network from hallucinating new objects inside holes, fulfilling the needs of object removal tasks in real-world scenarios. Extensive experiments are conducted to show that our method significantly outperforms existing methods in both quantitative and qualitative evaluation. Please refer to the project page: https://github.com/htzheng/CM-GAN-Inpainting.



Paperid:650
Authors:Yuchen Luo; Junwei Zhu; Keke He; Wenqing Chu; Ying Tai; Chengjie Wang; Junchi Yan
Title: StyleFace: Towards Identity-Disentangled Face Generation on Megapixels
Abstract:
Identity swapping and de-identification are two essential applications of identity-disentangled face image generation. Although sharing a similar problem definition, the two tasks have been long studied separately, and identity-disentangled face generation on megapixels is still under exploration. In this work, we propose StyleFace, a unified framework for 1024^2 resolution high-fidelity identity swapping and de-identification. To encode real identity while supporting virtual identity generation, we represent identity as a latent variable and further utilize contrastive learning for latent space regularization. Besides, we utilize StyleGAN2 to improve the generation quality on megapixels and devise an Adaptive Attribute Extractor, which adaptively preserves the identity-irrelevant attributes in a simple yet effective way. Extensive experiments demonstrate the state-of-the-art performance of StyleFace in high-resolution identity swapping and de-identification, respectively.



Paperid:651
Authors:Yunzhi Zhang; Jiajun Wu
Title: Video Extrapolation in Space and Time
Abstract:
Novel view synthesis (NVS) and video prediction (VP) are typically considered disjoint tasks in computer vision. However, they can both be seen as ways to observe the spatial-temporal world: NVS aims to synthesize a scene from a new point of view, while VP aims to see a scene from a new point of time. These two tasks provide complementary signals to obtain a scene representation, as viewpoint changes from spatial observations inform depth, and temporal observations inform the motion of cameras and individual objects. Inspired by these observations, we propose to study the problem of Video Extrapolation in Space and Time (VEST). We propose a model that tackles this problem and leverages the self-supervision from both tasks, while existing methods are designed to solve one of them. Experiments show that our method achieves performance better than or comparable to several state-of-the-art NVS and VP methods on indoor and outdoor real-world datasets.



Paperid:652
Authors:Yuheng Li; Yijun Li; Jingwan Lu; Eli Shechtman; Yong Jae Lee; Krishna Kumar Singh
Title: Contrastive Learning for Diverse Disentangled Foreground Generation
Abstract:
We introduce a new method for diverse foreground generation with explicit control over various factors. Existing image inpainting based foreground generation methods often struggle to generate diverse results and rarely allow users to explicitly control specific factors of variation (e.g., varying the facial identity or expression for face inpainting results). We leverage contrastive learning with latent codes to generate diverse foreground results for the same masked input. Specifically, we define two sets of latent codes, where one controls a pre-defined factor (“known”), and the other controls the remaining factors (“unknown”). The sampled latent codes from the two sets jointly bi-modulate the convolution kernels to guide the generator to synthesize diverse results. Experiments demonstrate the superiority of our method over state-of-the-arts in result diversity and generation controllability.



Paperid:653
Authors:Changgyoon Oh; Wonjune Cho; Yujeong Chae; Daehee Park; Lin Wang; Kuk-Jin Yoon
Title: BIPS: Bi-modal Indoor Panorama Synthesis via Residual Depth-Aided Adversarial Learning
Abstract:
Providing omnidirectional depth along with RGB information is important for numerous applications. However, as omnidirectional RGB-D data is not always available, synthesizing RGB-D panorama data from limited information of a scene can be useful. Therefore, some prior works tried to synthesize RGB panorama images from perspective RGB images; however, they suffer from limited image quality and can not be directly extended for RGB-D panorama synthesis. In this paper, we study a new problem: RGB-D panorama synthesis under the various configurations of cameras and depth sensors. Accordingly, we propose a novel bi-modal (RGB-D) panorama synthesis (BIPS) framework. Especially, we focus on indoor environments where the RGB-D panorama can provide a complete 3D model for many applications. We design a generator that fuses the bi-modal information and train it via residual depth-aided adversarial learning (RDAL). RDAL allows to synthesize realistic indoor layout structures and interiors by jointly inferring RGB panorama, layout depth, and residual depth. In addition, as there is no tailored evaluation metric for RGB-D panorama synthesis, we propose a novel metric (FAED) to effectively evaluate its perceptual quality. Extensive experiments show that our method synthesizes high-quality indoor RGB-D panoramas and provides more realistic 3D indoor models than prior methods. Code is available at https://github.com/chang9711/BIPS.



Paperid:654
Authors:Cheng-Ju Hsieh; Wei-Hao Chung; Chiou-Ting Hsu
Title: Augmentation of rPPG Benchmark Datasets: Learning to Remove and Embed rPPG Signals via Double Cycle Consistent Learning from Unpaired Facial Videos
Abstract:
Remote estimation of human physiological condition has attracted urgent attention during the pandemic of COVID-19. In this paper, we focus on the estimation of remote photoplethysmography (rPPG) from facial videos and address the deficiency issues of large-scale benchmarking datasets. We propose an end-to-end RErPPG-Net, including a Removal-Net and an Embedding-Net, to augment existing rPPG benchmark datasets. In the proposed augmentation scenario, the Removal-Net will first erase any inherent rPPG signals in the input video and then the Embedding-Net will embed another PPG signal into the video to generate an augmented video carrying the specified PPG signal. To train the model from unpaired videos, we propose a novel double-cycle consistent constraint to enforce the RErPPG-Net to learn to robustly and accurately remove and embed the delicate rPPG signals. The new benchmark ""Aug-rPPG dataset"" is augmented from UBFC-rPPG and PURE datasets and includes 5776 videos from 42 subjects with 76 different rPPG signals. Our experimental results show that existing rPPG estimators indeed benefit from the augmented dataset and achieve significant improvement when fine-tuned on the new benchmark. The code and dataset are available at https://github.com/nthumplab/RErPPGNet.



Paperid:655
Authors:Chaonan Ji; Tao Yu; Kaiwen Guo; Jingxin Liu; Yebin Liu
Title: Geometry-Aware Single-Image Full-Body Human Relighting
Abstract:
Single-image human relighting aims to relight a target human under new lighting conditions by decomposing the input image into albedo, shape and lighting. Although plausible relighting results can be achieved, previous methods suffer from both the entanglement between albedo and lighting and the lack of hard shadows, which significantly decrease the realism. To tackle these two problems, we propose a geometry-aware single-image human relighting framework that leverages single-image geometry reconstruction for joint deployment of traditional graphics rendering and neural rendering techniques. For the de-lighting, we explore the shortcomings of UNet architecture and propose a modified HRNet, achieving better disentanglement between albedo and lighting. For the relighting, we introduce a ray tracing-based per-pixel lighting representation that explicitly models high-frequency shadows and propose a learning-based shading refinement module to restore realistic shadows (including hard cast shadows) from ray-traced shading maps. Our framework is able to generate photo-realistic high-frequency shadows such as cast shadows under challenging lighting conditions. Extensive experiments demonstrate that our proposed method outperforms previous methods on both synthetic and real images.



Paperid:656
Authors:Zifan Shi; Yujun Shen; Jiapeng Zhu; Dit-Yan Yeung; Qifeng Chen
Title: 3D-Aware Indoor Scene Synthesis with Depth Priors
Abstract:
Despite the recent advancement of Generative Adversarial Networks (GANs) in learning 3D-aware image synthesis from 2D data, existing methods fail to model indoor scenes due to the large diversity of room layouts and the objects inside. We argue that indoor scenes do not have a shared intrinsic structure, and hence only using 2D images cannot adequately guide the model with the 3D geometry. In this work, we fill in this gap by introducing depth as a 3D prior. Compared with other 3D data formats, depth better fits the convolution-based generation mechanism and is more easily accessible in practice. Specifically, we propose a dual-path generator, where one path is responsible for depth generation, whose intermediate features are injected into the other path as the condition for appearance rendering. Such a design eases the 3D-aware synthesis with explicit geometry information. Meanwhile, we introduce a switchable discriminator both to differentiate real v.s. fake domains and to predict the depth from a given input. In this way, the discriminator can take the spatial arrangement into account and advise the generator to learn an appropriate depth condition. Extensive experimental results suggest that our approach is capable of synthesizing indoor scenes with impressively good quality and 3D consistency, significantly outperforming state-of-the-art alternatives.



Paperid:657
Authors:Joshua Weir; Junhong Zhao; Andrew Chalmers; Taehyun Rhee
Title: Deep Portrait Delighting
Abstract:
We present a deep neural network for removing undesirable shading features from an unconstrained portrait image, recovering the underlying texture. Our training scheme incorporates three regularization strategies: masked loss, to emphasize high-frequency shading features; soft-shadow loss, which improves sensitivity to subtle changes in lighting; and shading-offset estimation, to supervise separation of shading and texture. Our method demonstrates improved delighting quality and generalization when compared with the state-of-the-art. We further demonstrate how our delighting method can enhance the performance of light-sensitive computer vision tasks such as face relighting and semantic parsing, allowing them to handle extreme lighting conditions.



Paperid:658
Authors:Yu-Jie Chen; Shin-I Cheng; Wei-Chen Chiu; Hung-Yu Tseng; Hsin-Ying Lee
Title: Vector Quantized Image-to-Image Translation
Abstract:
Current image-to-image translation methods formulate the task with conditional generation models, leading to learning only the recolorization or regional changes as being constrained by the rich structural information provided by the conditional contexts. In this work, we propose introducing the vector quantization technique into the image-to-image translation framework. The vector quantized content representation can facilitate not only the translation, but also the unconditional distribution shared among different domains. Meanwhile, along with the disentangled style representation, the proposed method further enables the capability of image extension with flexibility in both intra- and inter-domains. Qualitative and quantitative experiments demonstrate that our framework achieves comparable performance to the state-of-the-art image-to-image translation and image extension methods. Compared to methods for individual tasks, the proposed method, as a unified framework, unleashes applications combining image-to-image translation, unconditional generation, and image extension altogether. For example, it provides style variability for image generation and extension, and equips image-to-image translation with further extension capabilities.



Paperid:659
Authors:Hyeonsu Lee; Chankyu Choi
Title: The Surprisingly Straightforward Scene Text Removal Method with Gated Attention and Region of Interest Generation: A Comprehensive Prominent Model Analysis
Abstract:
Scene text removal (STR), a task of erasing text from natural scene images, has recently attracted attention as an important component of editing text or concealing private information such as ID, telephone, and license plate numbers. While there are a variety of different methods for STR actively being researched, it is difficult to evaluate superiority because previously proposed methods do not use the same standardized training/evaluation dataset. We use the same standardized training/testing dataset to evaluate the performance of several previous methods after standardized re-implementation. We also introduce a simple yet extremely effective Gated Attention (GA) and Region-of-Interest Generation (RoIG) methodology in this paper. GA uses attention to focus on the text stroke as well as the textures and colors of the surrounding regions to remove text from the input image much more precisely. RoIG is applied to focus on only the region with text instead of the entire image to train the model more efficiently. Experimental results on the benchmark dataset show that our method significantly outperforms existing state-of-the-art methods in almost all metrics with remarkably higher-quality results. Furthermore, because our model does not generate a text stroke mask explicitly, there is no need for additional refinement steps or sub-models, making our model extremely fast with fewer parameters. The dataset and code are available at https://github.com/naver/garnet.



Paperid:660
Authors:Phong Nguyen-Ha; Nikolaos Sarafianos; Christoph Lassner; Janne Heikkilä; Tony Tung
Title: Free-Viewpoint RGB-D Human Performance Capture and Rendering
Abstract:
Capturing and faithfully rendering photorealistic humans from novel views is a fundamental problem for AR/VR applications. While prior work has shown impressive performance capture results in laboratory settings, it is non-trivial to achieve casual free-viewpoint human capture and rendering for unseen identities with high fidelity, especially for facial expressions, hands, and clothes. To tackle these challenges we introduce a novel view synthesis framework that generates realistic renders from unseen views of any human captured from a single-view and sparse RGB-D sensor, similar to a low-cost depth camera, and without actor-specific models. We propose an architecture to create dense feature maps in novel views obtained by sphere-based neural rendering, and create complete renders using a global context inpainting model. Additionally, an enhancer network leverages the overall fidelity, even in occluded areas from the original view, producing crisp renders with fine details. We show that our method generates high-quality novel views of synthetic and real human actors given a single-stream, sparse RGB-D input. It generalizes to unseen identities, and new poses and faithfully reconstructs facial expressions. Our approach outperforms prior view synthesis methods and is robust to different levels of depth sparsity.



Paperid:661
Authors:Chih-Jung Tsai; Cheng Sun; Hwann-Tzong Chen
Title: Multiview Regenerative Morphing with Dual Flows
Abstract:
This paper aims to address a new task of image morphing under a multiview setting, which takes two sets of multiview images as the input and generates intermediate renderings that not only exhibit smooth transitions between the two input sets but also ensure visual consistency across different views at any transition state. To achieve this goal, we propose a novel approach called Multiview Regenerative Morphing that formulates the morphing process as an optimization to solve for rigid transformation and optimal-transport interpolation. Given the multiview input images of the source and target scenes, we first learn a volumetric representation that models the geometry and appearance for each scene to enable the rendering of novel views. Then, the morphing between the two scenes is obtained by solving optimal transport between the two volumetric representations in Wasserstein metrics. Our approach does not rely on user-specified correspondences or 2D/3D input meshes, and we do not assume any predefined categories of the source and target scenes. The proposed view-consistent interpolation scheme directly works on multiview images to yield a novel and visually plausible effect of multiview free-form morphing.



Paperid:662
Authors:Tim Brooks; Alexei A. Efros
Title: Hallucinating Pose-Compatible Scenes
Abstract:
What does human pose tell us about a scene? We propose a task to answer this question: given human pose as input, hallucinate a compatible scene. Subtle cues captured by human pose --- action semantics, environment affordances, object interactions --- provide surprising insight into which scenes are compatible. We present a large-scale generative adversarial network for pose-conditioned scene generation. We significantly scale the size and complexity of training data, curating a massive meta-dataset containing over 19 million frames of humans in everyday environments. We double the capacity of our model with respect to StyleGAN2 to handle such complex data, and design a pose conditioning mechanism that drives our model to learn the nuanced relationship between pose and scene. We leverage our trained model for various applications: hallucinating pose-compatible scene(s) with or without humans, visualizing incompatible scenes and poses, placing a person from one generated image into another scene, and animating pose. Our model produces diverse samples and outperforms pose-conditioned StyleGAN2 and Pix2Pix/Pix2PixHD baselines in terms of accurate human placement (percent of correct keypoints) and quality (Fréchet inception distance).



Paperid:663
Authors:Borun Xu; Biao Wang; Jinhong Deng; Jiale Tao; Tiezheng Ge; Yuning Jiang; Wen Li; Lixin Duan
Title: Motion and Appearance Adaptation for Cross-Domain Motion Transfer
Abstract:
Motion transfer aims to transfer the motion of a driving video to a source image. When there are considerable differences between object in the driving video and that in the source image, traditional single domain motion transfer approaches often produce notable artifacts; for example, the synthesized image may fail to preserve the human shape of the source image (cf. Fig. 1 (a)). To address this issue, in the present work, we propose a Motion and Appearance Adaptation (MAA) approach for cross-domain motion transfer, in which we regularize the object in the synthesized image to capture the motion of the object in the driving frame, while still preserving the shape and appearance of the object in the source image. On one hand, considering the object shapes of the synthesized image and the driving frame might be different, we design a shape-invariant motion adaptation module that enforces the consistency of the angles of object parts in two images to capture the motion information. On the other hand, we introduce a structure-guided appearance consistency module designed to regularize the similarity between the corresponding patches of the synthesized image and the source image without affecting the learned motion in the synthesized image. Our proposed MAA model can be trained in an end-to-end manner with a cyclic reconstruction loss, and ultimately produces a satisfactory motion transfer result (cf. Fig. 1 (b)). We conduct extensive experiments on human dancing dataset Mixamo-Video to Fashion-Video and human face dataset Vox-Celeb to Cufs; on both of these, our MAA model outperforms existing methods both quantitatively and qualitatively.



Paperid:664
Authors:Jiahui Huang; Yuhe Jin; Kwang Moo Yi; Leonid Sigal
Title: Layered Controllable Video Generation
Abstract:
We introduce layered controllable video generation, where we, without any supervision, decompose the initial frame of a video into foreground and background layers, with which the user can control the video generation process by simply manipulating the foreground mask. The key challenges are the unsupervised foreground-background separation, which is ambiguous, and ability to anticipate user manipulations with access to only raw video sequences. We address these challenges by proposing a two-stage learning procedure. In the first stage, with the rich set of losses and dynamic foreground size prior, we learn how to separate the frame into foreground and background layers and, conditioned on these layers, how to generate the next frame using VQ-VAE generator. In the second stage, we fine-tune this network to anticipate edits to the mask, by fitting (parameterized) control to the mask from future frame. We demonstrate the effectiveness of this learning and the more granular control mechanism, while illustrating state-of-the-art performance on two benchmark datasets.



Paperid:665
Authors:Guillermo Gomez-Trenado; Stéphane Lathuilière; Pablo Mesejo; Óscar Cordón
Title: Custom Structure Preservation in Face Aging
Abstract:
In this work, we propose a novel architecture for face age editing that can produce structural modifications while maintaining relevant details present in the original image. We disentangle the style and content of the input image and propose a new decoder network that adopts a style-based strategy to combine the style and content representations of the input image while conditioning the output on the target age. Furthermore, we go beyond existing aging methods by allowing the users to adjust the degree of structure preservation in the input image at inference time. To this aim, we introduce a masking mechanism, termed Custom Structure Preservation (CUSP) module, that distinguishes relevant regions in the input image that should be preserved from those where details are irrelevant for the task. This mechanism does not require any additional supervision. Finally, we show that our method outperforms prior art on three datasets and demonstrate the effectiveness of our strategy to adjust structure preservation.



Paperid:666
Authors:Huicong Zhang; Haozhe Xie; Hongxun Yao
Title: Spatio-Temporal Deformable Attention Network for Video Deblurring
Abstract:
The key success factor of the video deblurring methods is to compensate for the blurry pixels of the mid-frame with the sharp pixels of the adjacent video frames. Therefore, mainstream methods align the adjacent frames based on the estimated optical flows and fuse the alignment frames for restoration. However, these methods sometimes generate unsatisfactory results because they rarely consider the blur levels of pixels, which may introduce blurry pixels from video frames. Actually, not all the pixels in the video frames are sharp and beneficial for deblurring. To address this problem, we propose the spatio-temporal deformable attention network (STDANet) for video delurring, which extracts the information of sharp pixels by considering the pixel-wise blur levels of the video frames. Specifically, STDANet is an encoder-decoder network combined with the motion estimator and spatio-temporal deformable attention (STDA) module, where motion estimator predicts coarse optical flows that are used as base offsets to find the corresponding sharp pixels in STDA module. Experimental results indicate that the proposed STDANet performs favorably against state-of-the-art methods on the GoPro, DVD, and BSD datasets.



Paperid:667
Authors:Bangbang Yang; Chong Bao; Junyi Zeng; Hujun Bao; Yinda Zhang; Zhaopeng Cui; Guofeng Zhang
Title: NeuMesh: Learning Disentangled Neural Mesh-Based Implicit Field for Geometry and Texture Editing
Abstract:
Very recently neural implicit rendering techniques have been rapidly evolved and shown great advantages in novel view synthesis and 3D scene reconstruction. However, existing neural rendering methods for editing purposes offer limited functionality, e.g., rigid transformation, or not applicable for fine-grained editing for general objects from daily lives. In this paper, we present a novel mesh-based representation by encoding the neural implicit field with disentangled geometry and texture codes on mesh vertices, which facilitates a set of editing functionalities, including mesh-guided geometry editing, designated texture editing with texture swapping, filling and painting operations. To this end, we develop several techniques including learnable sign indicators to magnify spatial distinguishability of mesh-based representation, distillation and fine-tuning mechanism to make a steady convergence, and the spatial-aware optimization strategy to realize precise texture editing. Extensive experiments and editing examples on both real and synthetic data demonstrate the superiority of our method on representation quality and editing ability. Code is available on the project webpage: https://zju3dv.github.io/neumesh/.



Paperid:668
Authors:Viktor Rudnev; Mohamed Elgharib; William Smith; Lingjie Liu; Vladislav Golyanik; Christian Theobalt
Title: NeRF for Outdoor Scene Relighting
Abstract:
Photorealistic editing of outdoor scenes from photographs requires a profound understanding of the image formation process and an accurate estimation of the scene geometry, reflectance and illumination. A delicate manipulation of the lighting can then be performed while keeping the scene albedo and geometry unaltered. We present NeRF-OSR, i.e., the first approach for outdoor scene relighting based on neural radiance fields. In contrast to the prior art, our technique allows simultaneous editing of both scene illumination and camera viewpoint using only a collection of outdoor photos shot in uncontrolled settings. Moreover, it enables direct control over the scene illumination, as defined through a spherical harmonics model. For evaluation, we collect a new benchmark dataset of several outdoor sites photographed from multiple viewpoints and at different times. For each time, a 360 degree environment map is captured together with a colour-calibration chequerboard to allow accurate numerical evaluations on real data against ground truth. Comparisons against SoTA show that NeRF-OSR enables controllable lighting and viewpoint editing at higher quality and with realistic self-shadowing reproduction.



Paperid:669
Authors:Cusuh Ham; Gemma Canet Tarrés; Tu Bui; James Hays; Zhe Lin; John Collomosse
Title: CoGS: Controllable Generation and Search from Sketch and Style
Abstract:
We present CoGS, a novel method for the style-conditioned, sketch-driven synthesis of images. CoGS enables exploration of diverse appearance possibilities for a given sketched object, enabling decoupled control over the structure and the appearance of the output. Coarse-grained control over object structure and appearance are enabled via an input sketch and an exemplar ""style"" conditioning image to a transformer-based sketch and style encoder to generate a discrete codebook representation. We map the codebook representation into a metric space, enabling fine-grained control over selection and interpolation between multiple synthesis options before generating the image via a vector quantized GAN (VQGAN) decoder. Our framework thereby unifies search and synthesis tasks, in that a sketch and style pair may be used to run an initial synthesis which may be refined via combination with similar results in a search corpus to produce an image more closely matching the user’s intent. We show that our model, trained on the 125 object classes of our newly created Pseudosketches dataset, is capable of producing a diverse gamut of semantic content and appearance styles.



Paperid:670
Authors:Peihao Zhu; Rameen Abdal; John Femiani; Peter Wonka
Title: HairNet: Hairstyle Transfer with Pose Changes
Abstract:
We propose a novel algorithm for automatic hairstyle transfer, specifically targeting complicated inputs that do not match in pose. The input to our algorithm are two images, one for the hairstyle and one for the identity (face). We do not require any additional inputs such as segmentation masks. Our algorithm consists of multiple steps and we contribute three novel components. The first contribution is the idea to include baldification into hairstyle editing pipelines to simplify inpainting of background and face regions covered by hair. The second contribution is a novel embedding algorithm that can handle both pose changes and semantic image blending. The third contribution is the hairnet architecture that semantically blends the hairstyle and identity images, performing multiple tasks jointly, such as baldification of the identity image, transformation estimation between the two images, warping, and hairstyle copying. Our results show a clear improvement over current state of the art methods in both quantitative and qualitative results. Code and data will be released.



Paperid:671
Authors:Yongsheng Yu; Dawei Du; Libo Zhang; Tiejian Luo
Title: Unbiased Multi-Modality Guidance for Image Inpainting
Abstract:
Image inpainting is an ill-posed problem to recover missing or damaged image content based on incomplete images with masks. Previous works usually predict the auxiliary structures (\eg, edges, segmentation and contours) to help fill visually realistic patches in a multi-stage fashion. However, imprecise auxiliary priors may yield biased inpainted results. Besides, it is time-consuming for some methods to be implemented by multiple stages of complex neural networks. To solve this issue, we develop an end-to-end multi-modality guided transformer network, including one inpainting branch and two auxiliary branches for semantic segmentation and edge textures. Within each transformer block, the proposed multi-scale spatial-aware attention module can learn the multi-modal structural features efficiently via auxiliary denormalization. Different from previous methods relying on direct guidance from biased priors, our method enriches semantically consistent context in an image based on discriminative interplay information from multiple modalities. Comprehensive experiments on several challenging image inpainting datasets show that our method achieves state-of-the-art performance to deal with various regular/irregular masks efficiently.



Paperid:672
Authors:Jaskirat Singh; Cameron Smith; Jose Echevarria; Liang Zheng
Title: Intelli-Paint: Towards Developing More Human-Intelligible Painting Agents
Abstract:
Stroke based rendering methods have recently become a popular solution for the generation of stylized paintings. However, the current research in this direction is focused mainly on the improvement of final canvas quality, and thus often fails to consider the intelligibility of the generated painting sequences to actual human users. In this work, we motivate the need to learn more human-intelligible painting sequences in order to facilitate the use of autonomous painting systems in a more interactive context (e.g. as a painting assistant tool for human users or for robotic painting applications). To this end, we propose a novel painting approach which learns to generate output canvases while exhibiting a painting style which is more relatable to human users. The proposed painting pipeline Intelli-Paint consists of 1) a progressive layering strategy which allows the agent to first paint a natural background scene before adding in each of the foreground objects in a progressive fashion. 2) We also introduce a novel sequential brushstroke guidance strategy which helps the painting agent to shift its attention between different image regions in a semantic-aware manner. 3) Finally, we propose a brushstroke regularization strategy which allows for 60-80% reduction in the total number of required brushstrokes without any perceivable differences in the quality of generated canvases. Through both quantitative and qualitative results, we show that the resulting agents not only show enhanced efficiency in output canvas generation but also exhibit a more natural-looking painting style which would better assist human users express their ideas through digital artwork.



Paperid:673
Authors:Jiale Tao; Biao Wang; Tiezheng Ge; Yuning Jiang; Wen Li; Lixin Duan
Title: Motion Transformer for Unsupervised Image Animation
Abstract:
Image animation aims to animate a source image by using motion learned from a driving video. Current state-of-the-art methods typically use convolutional neural networks (CNNs) to predict motion information, such as motion keypoints and corresponding local transformations. However, these CNN based methods do not explicitly model the interactions between motions; as a result, the important underlying motion relationship may be neglected, which can potentially lead to noticeable artifacts being produced in the generated animation video. To this end, we propose a new method, the motion transformer, which is the first attempt to build a motion estimator based on a vision transformer. More specifically, we introduce two types of tokens in our proposed method: i) image tokens formed from patch features and corresponding position encoding; and ii) motion tokens encoded with motion information. Both types of tokens are sent into vision transformers to promote underlying interactions between them through multi-head self attention blocks. By adopting this process, the motion information can be better learned to boost the model performance. The final embedded motion tokens are then used to predict the corresponding motion keypoints and local transformations. Extensive experiments on benchmark datasets show that our proposed method achieves promising results to the state-of-the-art baselines. Our source code will be public available.



Paperid:674
Authors:Chenfei Wu; Jian Liang; Lei Ji; Fan Yang; Yuejian Fang; Daxin Jiang; Nan Duan
Title: NÜWA: Visual Synthesis Pre-training for Neural visUal World creAtion
Abstract:
This paper presents a unified multimodal pre-trained model called NÃœWA that can generate new or manipulate existing visual data (i.e., image and video) for various visual synthesis tasks. To cover language, image, and video at the same time for different scenarios, a 3D transformer encoder-decoder framework is designed, which can not only deal with videos as 3D data but also adapt to texts and images as 1D and 2D data, respectively. A 3D Nearby Attention (3DNA) mechanism is also proposed to consider the nature of the visual data and reduce the computational complexity. We evaluate NÃœWA on 8 downstream tasks and 10 datasets. Compared to several strong baselines, NÃœWA achieves state-of-the-art results on text-to-image generation, text-to-video generation, video predictions, etc. It also shows surprisingly good zero-shot capabilities on text-guided image and video manipulation tasks.



Paperid:675
Authors:Chenyu Yang; Wanrong He; Yingqing Xu; Yang Gao
Title: EleGANt: Exquisite and Locally Editable GAN for Makeup Transfer
Abstract:
Most existing methods view makeup transfer as transferring color distributions of different facial regions and ignore details such as eye shadows and blushes. Besides, they only achieve controllable transfer within predefined fixed regions. This paper emphasizes the transfer of makeup details and steps towards more flexible controls. To this end, we propose Exquisite and locally editable GAN for makeup transfer (EleGANt). It encodes facial attributes into pyramidal feature maps to preserves high-frequency information. It uses attention to extract makeup features from the reference and adapt them to the source face, and we introduce a novel Sow-Attention Module that applies attention within shifted overlapped windows to reduce the computational cost. Moreover, EleGANt is the first to achieve customized local editing within arbitrary areas by corresponding editing on the feature maps. Extensive experiments demonstrate that EleGANt generates realistic makeup faces with exquisite details and achieves state-of-the-art performance. The code is available at https://github.com/Chenyu-Yang-2000/EleGANt.



Paperid:676
Authors:Haorui Song; Yong Du; Tianyi Xiang; Junyu Dong; Jing Qin; Shengfeng He
Title: Editing Out-of-Domain GAN Inversion via Differential Activations
Abstract:
Despite the demonstrated editing capacity in the latent space of a pretrained GAN model, inverting real-world images is stuck in a dilemma that the reconstruction cannot be faithful to the original input. The main reason for this is that the distributions between training and real-world data are misaligned, and because of that, it is unstable of GAN inversion for real image editing. In this paper, we propose a novel GAN prior based editing framework to tackle the out-of-domain inversion problem with a composition-decomposition paradigm. In particular, during the phase of composition, we introduce a differential activation module for detecting semantic changes from a global perspective, \ie, the relative gap between the features of edited and unedited images. With the aid of the generated Diff-CAM mask, a coarse reconstruction can intuitively be composited by the paired original and edited images. In this way, the attribute-irrelevant regions can be survived in almost whole, while the quality of such an intermediate result is still limited by an unavoidable ghosting effect. Consequently, in the decomposition phase, we further present a GAN prior based deghosting network for separating the final fine edited image from the coarse reconstruction. Extensive experiments exhibit superiorities over the state-of-the-art methods, in terms of qualitative and quantitative evaluations. The robustness and flexibility of our method is also validated on both scenarios of single attribute and multi-attribute manipulations. Code is available at https://github.com/HaoruiSong622/Editing-Out-of-Domain.



Paperid:677
Authors:Motasem Alfarra; Juan C. Pérez; Anna Frühstück; Philip H. S. Torr; Peter Wonka; Bernard Ghanem
Title: On the Robustness of Quality Measures for GANs
Abstract:
This work evaluates the robustness of quality measures of generative models such as Inception Score (IS) and Fréchet Inception Distance (FID). Analogous to the vulnerability of deep models against a variety of adversarial attacks, we show that such metrics can also be manipulated by additive pixel perturbations. Our experiments indicate that one can generate a distribution of images with very high scores but low perceptual quality. Conversely, one can optimize for small imperceptible perturbations that, when added to real world images, deteriorate their scores. We further extend our evaluation to generative models themselves, including the state of the art network StyleGANv2. We show the vulnerability of both the generative model and the FID against additive perturbations in the latent space. Finally, we show that the FID can be robustified by simply replacing the standard Inception with a robust Inception. We validate the effectiveness of the robustified metric through extensive experiments, showing it is more robust against manipulation.



Paperid:678
Authors:Seung Hyun Lee; Gyeongrok Oh; Wonmin Byeon; Chanyoung Kim; Won Jeong Ryoo; Sang Ho Yoon; Hyunjun Cho; Jihyun Bae; Jinkyu Kim; Sangpil Kim
Title: Sound-Guided Semantic Video Generation
Abstract:
The recent success in StyleGAN demonstrates that pre-trained StyleGAN latent space is useful for realistic video generation. However, the generated motion in the video is usually not semantically meaningful due to the difficulty of determining the direction and magnitude in the StyleGAN latent space. In this paper, we propose a framework to generate realistic videos by leveraging multimodal (sound-image-text) embedding space. As sound provides the temporal contexts of the scene, our framework learns to generate a video that is semantically consistent with sound. First, our sound inversion module maps the audio directly into the StyleGAN latent space. We then incorporate the CLIP-based multimodal embedding space to further provide the audio-visual relationships. Finally, the proposed frame generator learns to find the trajectory in the latent space which is coherent with the corresponding sound and generates a video in a hierarchical manner. We provide the new high-resolution landscape video dataset (audio-visual pair) for the sound-guided video generation task. The experiments show that our model outperforms the state-of-the-art methods in terms of video quality. We further show several applications including image and video editing to verify the effectiveness of our method. We provide diverse examples in the following anonymized project website: \url{https://anonymous5584.github.io}"



Paperid:679
Authors:Lingzhi Zhang; Connelly Barnes; Kevin Wampler; Sohrab Amirghodsi; Eli Shechtman; Zhe Lin; Jianbo Shi
Title: Inpainting at Modern Camera Resolution by Guided PatchMatch with Auto-Curation
Abstract:
Recently, deep models have established SOTA performance for low-resolution image inpainting, but they lack fidelity at resolutions associated with modern cameras such as 4K or more, and for large holes. We contribute an inpainting benchmark dataset of photos at 4K and above representative of modern sensors. We demonstrate a novel framework that combines deep learning and traditional methods. We use an existing deep inpainting model LaMa [28] to fill the hole plausibly, es- tablish three guide images consisting of structure, segmentation, depth, and apply a multiply-guided PatchMatch [1] to produce eight candidate upsampled inpainted images. Next, we feed all candidate inpaintings through a novel curation module that chooses a good inpainting by column summation on an 8x8 antisymmetric pairwise preference matrix. Our framework’s results are overwhelmingly preferred by users over 8 strong baselines, with improvements of quantitative metrics up to 7.4 times over the best baseline LaMa, and our technique when paired with 4 different SOTA inpainting backbones improves each such that ours is overwhelmingly preferred by users over a strong super-res baseline.



Paperid:680
Authors:Aram Davtyan; Paolo Favaro
Title: Controllable Video Generation through Global and Local Motion Dynamics
Abstract:
We present GLASS, a method for Global and Local Action-driven Sequence Synthesis. GLASS is a generative model that is trained on video sequences in an unsupervised manner and that can animate an input image at test time. The method learns to segment frames into foreground-background layers and to generate transitions of the foregrounds over time through a global and local action representation. Global actions are explicitly related to 2D shifts, while local actions are instead related to (both geometric and photometric) local deformations. GLASS uses a recurrent neural network to transition between frames and is trained through a reconstruction loss. We also introduce W-Sprites (Walking Sprites), a novel synthetic dataset with a predefined action space. We evaluate our method on both W-Sprites and real datasets, and find that GLASS is able to generate realistic video sequences from a single input image and to successfully learn a more advanced action space than in prior work. Further details, the code and example videos are available at https://araachie.github.io/glass/.



Paperid:681
Authors:Fei Yin; Yong Zhang; Xiaodong Cun; Mingdeng Cao; Yanbo Fan; Xuan Wang; Qingyan Bai; Baoyuan Wu; Jue Wang; Yujiu Yang
Title: StyleHEAT: One-Shot High-Resolution Editable Talking Face Generation via Pre-trained StyleGAN
Abstract:
One-shot talking face generation aims at synthesizing a high-quality talking face video from an arbitrary portrait image, driven by a video or an audio segment. In this work, we provide a solution from a novel perspective that differs from existing frameworks. We first investigate the latent feature space of a pre-trained StyleGAN and discover some excellent spatial transformation properties. Upon the observation, we propose a novel unified framework based on a pre-trained StyleGAN that enables a set of powerful functionalities, i.e., high-resolution video generation, disentangled control by driving video or audio, and flexible face editing. Our framework elevates the resolution of the synthesized talking face to 1024×1024 for the first time, even though the training dataset has a lower resolution. Moreover, our framework allows two types of facial editing, i.e., global editing via GAN inversion and intuitive editing via 3D morphable models. Comprehensive experiments show superior video quality and flexible controllability over state-of-the-art methods. Code is available at https://github.com/FeiiYin/StyleHEAT.



Paperid:682
Authors:Songwei Ge; Thomas Hayes; Harry Yang; Xi Yin; Guan Pang; David Jacobs; Jia-Bin Huang; Devi Parikh
Title: Long Video Generation with Time-Agnostic VQGAN and Time-Sensitive Transformer
Abstract:
Videos are created to express emotion, exchange information, and share experiences. Video synthesis has intrigued researchers for a long time. Despite the rapid progress driven by advances in visual synthesis, most existing studies focus on improving the frames’ quality and the transitions between them, while little progress has been made in generating longer videos. In this paper, we present a method that builds on 3D-VQGAN and transformers to generate videos with thousands of frames. Our evaluation shows that our model trained on 16-frame video clips from standard benchmarks such as UCF-101, Sky Time-lapse, and Taichi-HD datasets can generate diverse, coherent, and high-quality long videos. We also showcase conditional extensions of our approach for generating meaningful long videos by incorporating temporal information with text and audio. Videos and code can be found at https://songweige.github.io/projects/tats.



Paperid:683
Authors:Eyal Naor; Itai Antebi; Shai Bagon; Michal Irani
Title: Combining Internal and External Constraints for Unrolling Shutter in Videos
Abstract:
Videos obtained by rolling-shutter (RS) cameras result in spatially-distorted frames. These distortions become significant under fast camera/scene motions. Undoing effects of RS is sometimes addressed as a spatial problem, where objects need to be rectified/displaced in order to generate their correct global shutter (GS) frame. However, the cause of the RS effect is inherently temporal, not spatial. In this paper we propose a space-time solution to the RS problem. We observe that despite the severe differences between their xy-frames, a RS video and its corresponding GS video tend to share the exact same xt-slices - up to a known sub-frame temporal shift. Moreover, they share the same distribution of small 2D xt-patches, despite the strong temporal aliasing within each video. This allows to constrain the GS output video using video-specific constraints imposed by the RS input video. Our algorithm is composed of 3 main components: (i) Dense temporal upsampling between consecutive RS frames using an off-the-shelf method, (which was trained on regular video sequences), from which we extract GS “proposals”. (ii) Learning to correctly merge an ensemble of such GS “proposals” using a dedicated MergeNet. (iii) A video-specific zero-shot optimization which imposes the similarity of xt-patches between the GS output video and the RS input video. Our method obtains state-of-the-art results on benchmark datasets, both numerically and visually, despite being trained on a small synthetic RS/GS dataset. Moreover, it generalizes well to new complex RS videos with motion types outside the distribution of the training set (e.g., complex non-rigid motions) - videos which competing methods trained on much more data cannot handle well. We attribute these generalization capabilities to the combination of external and internal constraints.



Paperid:684
Authors:Winfried Lötzsch; Max Reimann; Martin Büssemeyer; Amir Semmo; Jürgen Döllner; Matthias Trapp
Title: WISE: Whitebox Image Stylization by Example-Based Learning
Abstract:
Image-based artistic rendering can synthesize a variety of expressive styles using algorithmic image filtering. In contrast to deep learning-based methods, these heuristics-based filtering techniques can operate on high-resolution images, are interpretable, and can be parameterized according to various design aspects. However, adapting or extending these techniques to produce new styles is often a tedious and error-prone task that requires expert knowledge. We propose a new paradigm to alleviate this problem: implementing algorithmic image filtering techniques as differentiable operations that can learn parametrizations aligned to certain reference styles. To this end, we present WISE, an example-based image-processing system that can handle a multitude of stylization techniques, such as watercolor, oil, or cartoon stylization, within a common framework. By training parameter prediction networks for global and local filter parameterizations, we can simultaneously adapt effects to reference styles and image content, e.g., to enhance facial features. Our method can be optimized in a style-transfer framework or learned in a generative-adversarial setting for image-to-image translation. We demonstrate that jointly training an xDoG filter and a CNN for postprocessing can achieve comparable results to a state-of-the-art GAN-based method.



Paperid:685
Authors:Linjie Lyu; Ayush Tewari; Thomas Leimkühler; Marc Habermann; Christian Theobalt
Title: Neural Radiance Transfer Fields for Relightable Novel-View Synthesis with Global Illumination
Abstract:
Given a set of images of a scene, the re-rendering of this scene from novel views and lighting conditions is an important and challenging problem in Computer Vision and Graphics. On the one hand, most existing works in Computer Vision usually impose many assumptions regarding the image formation process, e.g. direct illumination and predefined materials, to make scene parameter estimation tractable. On the other hand, mature Computer Graphics tools allow modeling of complex photo-realistic light transport given all the scene parameters. Combining these approaches, we propose a method for scene relighting under novel views by learning a neural precomputed radiance transfer function, which implicitly handles global illumination effects using novel environment maps. Our method can be solely supervised on a set of real images of the scene under a single unknown lighting condition. To disambiguate the task during training, we tightly integrate a differentiable path tracer in the training process and propose a combination of a synthesized OLAT and a real image loss. Results show that the recovered disentanglement of scene parameters improves significantly over the current state of the art and, thus, also our re-rendering results are more realistic and accurate.



Paperid:686
Authors:Yinbo Chen; Xiaolong Wang
Title: Transformers As Meta-Learners for Implicit Neural Representations
Abstract:
Implicit Neural Representations (INRs) have emerged and shown their benefits over discrete representations in recent years. However, fitting an INR to the given observations usually requires optimization with gradient descent from scratch, which is inefficient and does not generalize well with sparse observations. To address this problem, most of the prior works train a hypernetwork that generates a single vector to modulate the INR weights, where the single vector becomes an information bottleneck that limits the reconstruction precision of the output INR. Recent work shows that the whole set of weights in INR can be precisely inferred without the single-vector bottleneck by gradient-based meta-learning. Motivated by a generalized formulation of gradient-based meta-learning, we propose a formulation that uses Transformers as hypernetworks for INRs, where it can directly build the whole set of INR weights with Transformers specialized as set-to-set mapping. We demonstrate the effectiveness of our method for building INRs in different tasks and domains, including 2D image regression and view synthesis for 3D objects. Our work draws connections between the Transformer hypernetworks and gradient-based meta-learning algorithms and we provide further analysis for understanding the generated INRs.



Paperid:687
Authors:Taewoo Kim; Chaeyeon Chung; Yoonseo Kim; Sunghyun Park; Kangyeol Kim; Jaegul Choo
Title: Style Your Hair: Latent Optimization for Pose-Invariant Hairstyle Transfer via Local-Style-Aware Hair Alignment
Abstract:
Editing hairstyle is unique and challenging due to the complexity and delicacy of hairstyle. Although recent approaches significantly improved the hair details, these models often produce undesirable outputs when a pose of a source image is considerably different from that of a target hair image, limiting their real-world applications. HairFIT, a pose-invariant hairstyle transfer model, alleviates this limitation yet still shows unsatisfactory quality in preserving delicate hair textures. To solve these limitations, we propose a high-performing pose-invariant hairstyle transfer model equipped with latent optimization and a newly presented local-style-matching loss. In the StyleGAN2 latent space, we first explore a pose-aligned latent code of a target hair with the detailed textures preserved based on local style matching. Then, our model inpaints the occlusions of the source considering the aligned target hair and blends both images to produce a final output. The experimental results demonstrate that our model has strengths in transferring a hairstyle under larger pose differences and preserving local hairstyle textures. The codes are available at https://github.com/Taeu/Style-Your-Hair.



Paperid:688
Authors:Sangyun Lee; Gyojung Gu; Sunghyun Park; Seunghwan Choi; Jaegul Choo
Title: High-Resolution Virtual Try-On with Misalignment and Occlusion-Handled Conditions
Abstract:
Image-based virtual try-on aims to synthesize an image of a person wearing a given clothing item. To solve the task, the existing methods warp the clothing item to fit the person’s body and generate the segmentation map of the person wearing the item before fusing the item with the person. However, when the warping and the segmentation generation stages operate individually without information exchange, the misalignment between the warped clothes and the segmentation map occurs, which leads to the artifacts in the final image. The information disconnection also causes excessive warping near the clothing regions occluded by the body parts, so-called pixel-squeezing artifacts. To settle the issues, we propose a novel try-on condition generator as a unified module of the two stages (i.e., warping and segmentation generation stages). A newly proposed feature fusion block in the condition generator implements the information exchange, and the condition generator does not create any misalignment or pixel-squeezing artifacts. We also introduce discriminator rejection that filters out the incorrect segmentation map predictions and assures the performance of virtual try-on frameworks. Experiments on a high-resolution dataset demonstrate that our model successfully handles the misalignment and occlusion, and significantly outperforms the baselines. Code is available at https://github.com/sangyun884/HR-VITON"



Paperid:689
Authors:Hengsheng Zhang; Xueyi Zou; Jiaming Guo; Youliang Yan; Rong Xie; Li Song
Title: A Codec Information Assisted Framework for Efficient Compressed Video Super-Resolution
Abstract:
Online processing of compressed videos to increase their resolutions attracts increasing and broad attention. Video Super-Resolution (VSR) using recurrent neural network architecture is a promising solution due to its efficient modeling of long-range temporal dependencies. However, state-of-the-art recurrent VSR models still require significant computation to obtain a good performance, mainly because of the complicated motion estimation for frame/feature alignment and the redundant processing of consecutive video frames. In this paper, considering the characteristics of compressed videos, we propose a Codec Information Assisted Framework (CIAF) to boost and accelerate recurrent VSR models for compressed videos. Firstly, the framework reuses the coded video information of Motion Vectors to model the temporal relationships between adjacent frames. Experiments demonstrate that the models with Motion Vector based alignment can significantly boost the performance with negligible additional computation, even comparable to those using more complex optical flow based alignment. Secondly, by further making use of the coded video information of Residuals, the framework can be informed to skip the computation on redundant pixels. Experiments demonstrate that the proposed framework can save up to 70% of the computation without performance drop on the REDS4 test videos encoded by H.264 when CRF is 23.



Paperid:690
Authors:Jeong-gi Kwak; Yuanming Li; Dongsik Yoon; Donghyeon Kim; David Han; Hanseok Ko
Title: Injecting 3D Perception of Controllable NeRF-GAN into StyleGAN for Editable Portrait Image Synthesis
Abstract:
Over the years, 2D GANs have achieved great successes in photorealistic portrait generation. However, they lack 3D understanding in the generation process, thus they suffer from multi-view inconsistency problem. To alleviate the issue, many 3D-aware GANs have been proposed and shown notable results, but 3D GANs struggle with editing semantic attributes. The controllability and interpretability of 3D GANs have not been much explored. In this work, we propose two solutions to overcome these weaknesses of 2D GANs and 3D-aware GANs. We first introduce a novel 3D-aware GAN, SURF-GAN, which is capable of discovering semantic attributes during training and controlling them in an unsupervised manner. After that, we inject the prior of SURF-GAN into StyleGAN to obtain a high-fidelity 3D-controllable generator. Unlike existing latent-based methods allowing implicit pose control, the proposed 3D-controllable StyleGAN enables explicit pose control over portrait generation. This distillation allows direct compatibility between 3D control and many StyleGAN-based techniques (e.g., inversion and stylization), and also brings an advantage in terms of computational resources. Our codes are available at https://github.com/jgkwak95/SURF-GAN.



Paperid:691
Authors:Andreas Kurz; Thomas Neff; Zhaoyang Lv; Michael Zollhöfer; Markus Steinberger
Title: AdaNeRF: Adaptive Sampling for Real-Time Rendering of Neural Radiance Fields
Abstract:
Novel view synthesis has recently been revolutionized by learning neural radiance fields directly from sparse observations. However, rendering images with this new paradigm is slow due to the fact that an accurate quadrature of the volume rendering equation requires a large number of samples for each ray. Previous work has mainly focused on speeding up the network evaluations that are associated with each sample point, e.g., via caching of radiance values into explicit spatial data structures, but this comes at the expense of model compactness. In this paper, we propose a novel dual-network architecture that takes an orthogonal direction by learning how to best reduce the number of required sample points. To this end, we split our network into a sampling and shading network that are jointly trained. Our training scheme employs fixed sample positions along each ray, and incrementally introduces sparsity throughout training to achieve high quality even at low sample counts. After fine-tuning with the target number of samples, the resulting compact neural representation can be rendered in real-time. Our experiments demonstrate that our approach outperforms concurrent compact neural representations in terms of quality and frame rate and performs on par with highly efficient hybrid representations. Code and supplementary material is available at https://thomasneff.github.io/adanerf.



Paperid:692
Authors:Shuhong Chen; Matthias Zwicker
Title: Improving the Perceptual Quality of 2D Animation Interpolation
Abstract:
Traditional 2D animation is labor-intensive, often requiring animators to manually draw twelve illustrations per second of movement. While automatic frame interpolation may ease this burden, 2D animation poses additional difficulties compared to photorealistic video. In this work, we address challenges unexplored in previous animation interpolation systems, with a focus on improving perceptual quality. Firstly, we propose SoftsplatLite (SSL), a forward-warping interpolation architecture with fewer trainable parameters and better perceptual performance. Secondly, we design a Distance Transform Module (DTM) that leverages line proximity cues to correct aberrations in difficult solid-color regions. Thirdly, we define a Restricted Relative Linear Discrepancy metric (RRLD) to automate the previously manual training data collection process. Lastly, we explore evaluation of 2D animation generation through a user study, and establish that the LPIPS perceptual metric and chamfer line distance (CD) are more appropriate measures of quality than PSNR and SSIM used in prior art.



Paperid:693
Authors:Jou Won Song; Ye-In Park; Kyeongbo Kong; Jaeho Kwak; Suk-Ju Kang
Title: Selective TransHDR: Transformer-Based Selective HDR Imaging Using Ghost Region Mask
Abstract:
The primary issue in high dynamic range (HDR) imaging is the removal of ghost artifacts afforded when merging multi-exposure low dynamic range images. In the weakly misaligned region, ghost artifacts can be suppressed using convolutional neural network (CNN)-based methods. However, in highly misaligned regions, it is necessary to extract features from the global region because the necessary information does not exist in the local region. Therefore, the CNN-based methods specialized for local features extraction cannot obtain satisfactory results. To address this issue, we propose a transformer-based selective HDR image reconstruction network that uses a ghost region mask. The proposed method separates a given image into ghost and non-ghost regions, and then, selectively applies either the CNN or the transformer. The proposed selective transformer module divides an entire image into several regions to effectively extract the features of each region for HDR image reconstruction, thereby extracting the whole information required for HDR reconstruction in the ghost regions from the entire image. Extensive experiments conducted on several benchmark datasets demonstrate the superiority of the proposed method over existing state-of-the-art methods in terms of the mitigation of ghost artifacts.



Paperid:694
Authors:Cheng Ma; Jingyi Zhang; Jie Zhou; Jiwen Lu
Title: Learning Series-Parallel Lookup Tables for Efficient Image Super-Resolution
Abstract:
Lookup table (LUT) has shown its efficacy in low-level vision tasks due to the valuable characteristics of low computational cost and hardware independence. However, recent attempts to address the problem of single image super-resolution (SISR) with lookup tables are highly constrained by the small receptive field size. Besides, their frameworks of single-layer lookup tables limit the extension and generalization capacities of the model. In this paper, we propose a framework of series-parallel lookup tables (SPLUT) to alleviate the above issues and achieve efficient image super-resolution. On the one hand, we cascade multiple lookup tables to enlarge the receptive field of each extracted feature vector. On the other hand, we propose a parallel network which includes two branches of cascaded lookup tables which process different components of the input low-resolution images. By doing so, the two branches collaborate with each other and compensate for the precision loss of discretizing input pixels when establishing lookup tables. Compared to previous lookup table based methods, our framework has stronger representation abilities with more flexible architectures. Furthermore, we no longer need interpolation methods which introduce redundant computations so that our method can achieve faster inference speed. Extensive experimental results on five popular benchmark datasets show that our method obtains superior SISR performance in a more efficient way. The code is available at https://github.com/zhjy2016/SPLUT.



Paperid:695
Authors:Di Chen; Yu Liu; Lianghua Huang; Bin Wang; Pan Pan
Title: GeoAug: Data Augmentation for Few-Shot NeRF with Geometry Constraints
Abstract:
Neural Radiance Fields (NeRF) show remarkable ability to render novel views of a certain scene by learning an implicit volumetric representation with only posed RGB images. Despite its impressiveness and simplicity, NeRF usually converges to sub-optimal solutions with incorrect geometries given few training images. We hereby present GeoAug: a data augmentation method for NeRF, which enriches training data based on multi-view geometric constraint. GeoAug provides random artificial (novel pose, RGB image) pairs for training, where the RGB image is from a nearby training view. The rendering of a novel pose is warped to the nearby training view with depth map and relative pose to match the RGB image supervision. Our method reduces the risk of over-fitting by introducing more data during training, while also provides additional implicit supervision for depth maps. In experiments, our method significantly boosts the performance of neural radiance fields conditioned on few training views.



Paperid:696
Authors:Ankan Kumar Bhunia; Salman Khan; Hisham Cholakkal; Rao Muhammad Anwer; Fahad Shahbaz Khan; Jorma Laaksonen; Michael Felsberg
Title: DoodleFormer: Creative Sketch Drawing with Transformers
Abstract:
Creative sketching or doodling is an expressive activity, where imaginative and previously unseen depictions of everyday visual objects are drawn. Creative sketch image generation is a challenging vision problem, where the task is to generate diverse, yet realistic creative sketches possessing the unseen composition of the visual-world objects. Here, we propose a novel coarse-to-fine two-stage framework, DoodleFormer, that decomposes the creative sketch generation problem into the creation of coarse sketch composition followed by the incorporation of fine-details in the sketch. We introduce graph-aware transformer encoders that effectively capture global dynamic as well as local static structural relations among different body parts. To ensure diversity of the generated creative sketches, we introduce a probabilistic coarse sketch decoder that explicitly models the variations of each sketch body part to be drawn. Experiments are performed on two creative sketch datasets: Creative Birds and Creative Creatures. Our qualitative, quantitative and human-based evaluations show that DoodleFormer outperforms the state-of-the-art on both datasets, yielding realistic and diverse creative sketches. On Creative Creatures, DoodleFormer achieves an absolute gain of 25 in Fr\`echet inception distance (FID) over state-of-the-art. We also demonstrate the effectiveness of DoodleFormer for related applications of text to creative sketch generation, sketch completion and house layout generation.



Paperid:697
Authors:Pablo Cervantes; Yusuke Sekikawa; Ikuro Sato; Koichi Shinoda
Title: Implicit Neural Representations for Variable Length Human Motion Generation
Abstract:
We propose an action-conditional human motion generation method using variational implicit neural representations (INR). The variational formalism enables action-conditional distributions of INRs, from which one can easily sample representations to generate novel human motion sequences. Our method offers variable-length sequence generation by construction because a part of INR is optimized for a whole sequence of arbitrary length with temporal embeddings. In contrast, previous works reported difficulties with modeling variable-length sequences. We confirm that our method with a Transformer decoder outperforms all relevant methods on HumanAct12, NTU-RGBD, and UESTC datasets in terms of realism and diversity of generated motions. Surprisingly, even our method with an MLP decoder consistently outperforms the state-of-the-art Transformer-based auto-encoder. In particular, we show that variable-length motions generated by our method are better than fixed-length motions generated by the state-of-the-art method in terms of realism and diversity. Code at https://github.com/PACerv/ImplicitMotion"



Paperid:698
Authors:Siyuan Zhou; Liu Liu; Li Niu; Liqing Zhang
Title: Learning Object Placement via Dual-Path Graph Completion
Abstract:
Object placement aims to place a foreground object over a background image with a suitable location and size. In this work, we treat object placement as a graph completion problem and propose a novel graph completion module (GCM). The background scene is represented by a graph with multiple nodes at different spatial locations with various receptive fields. The foreground object is encoded as a special node that should be inserted at a reasonable place in this graph. We also design a dual-path framework upon the structure of GCM to fully exploit annotated composite images. With extensive experiments on OPA dataset, our method proves to significantly outperform existing methods in generating plausible object placement without loss of diversity.



Paperid:699
Authors:Chajin Shin; Hyeongmin Lee; Hanbin Son; Sangjin Lee; Dogyoon Lee; Sangyoun Lee
Title: Expanded Adaptive Scaling Normalization for End to End Image Compression
Abstract:
Recently, learning-based image compression methods that utilize convolutional neural layers have been developed rapidly. Rescaling modules such as batch normalization which are often used in convolutional neural networks do not operate adaptively for the various inputs. Therefore, Generalized Divisible Normalization(GDN) has been widely used in image compression to rescale the input features adaptively across both spatial and channel axes. However, the representation power or degree of freedom of GDN is severely limited. Additionally, GDN cannot consider the spatial correlation of an image. To handle the limitations of GDN, we construct an expanded form of the adaptive scaling module, named Expanded Adaptive Scaling Normalization(EASN). First, we exploit the swish function to increase the representation ability. Then, we increase the receptive field to make the adaptive rescaling module consider the spatial correlation. Furthermore, we introduce an input mapping function to give the module a higher degree of freedom. We demonstrate how our EASN works in an image compression network using the visualization results of the feature map, and we conduct extensive experiments to show that our EASN increases the rate-distortion performance remarkably, and even outperforms the VVC intra at a high bit rate.



Paperid:700
Authors:Gayoung Lee; Hyunsu Kim; Junho Kim; Seonghyeon Kim; Jung-Woo Ha; Yunjey Choi
Title: Generator Knows What Discriminator Should Learn in Unconditional GANs
Abstract:
Recent methods for conditional image generation benefit from dense supervision such as segmentation label maps to achieve high-fidelity. However, it is rarely explored to employ dense supervision for unconditional image generation. Here we explore the efficacy of dense supervision in unconditional generation and find generator feature maps can be an alternative of cost-expensive semantic label maps. From our empirical evidences, we propose a new generator-guided discriminator regularization (GGDR) in which the generator feature maps supervise the discriminator to have rich semantic representations in unconditional generation. In specific, we employ an U-Net architecture for discriminator, which is trained to predict the generator feature maps given fake images as inputs. Extensive experiments on mulitple datasets show that our GGDR consistently improves the performance of baseline methods in terms of quantitative and qualitative aspects. Code is available at https://github.com/naver-ai/GGDR.



Paperid:701
Authors:Nan Liu; Shuang Li; Yilun Du; Antonio Torralba; Joshua B. Tenenbaum
Title: Compositional Visual Generation with Composable Diffusion Models
Abstract:
Large text-guided diffusion models, such as DALLE-2, are able to generate stunning photorealistic images given natural language descriptions. While such models are highly flexible, they struggle to understand the composition of certain concepts, such as confusing the attributes of different objects or relations between objects. In this paper, we propose an alternative structured approach for compositional generation using diffusion models. An image is generated by composing a set of diffusion models, with each of them modeling a certain component of the image. To do this, we interpret diffusion models as energy-based models in which the data distributions defined by the energy functions may be explicitly combined. The proposed method can generate scenes at test time that are substantially more complex than those seen in training, composing sentence descriptions, object relations, human facial attributes, and even generalizing to new combinations that are rarely seen in the real world. We further illustrate how our approach may be used to compose pre-trained text-guided diffusion models and generate photorealistic images containing all the details described in the input descriptions, including the binding of certain object attributes that have been shown difficult for DALLE-2. These results point to the effectiveness of the proposed method in promoting structured generalization for visual generation.



Paperid:702
Authors:Fabio Pizzati; Jean-François Lalonde; Raoul de Charette
Title: ManiFest: Manifold Deformation for Few-Shot Image Translation
Abstract:
Most image-to-image translation methods require a large number of training images, which restricts their applicability. We instead propose ManiFest: a framework for few-shot image translation that learns a context-aware representation of a target domain from a few images only. To enforce feature consistency, our framework learns a style manifold between source and additional anchor domains (assumed to be composed of large numbers of images). The learned manifold is interpolated and deformed towards the few-shot target domain via patch-based adversarial and feature statistics alignment losses. All of these components are trained simultaneously during a single end-to-end loop. In addition to the general few-shot translation task, our approach can alternatively be conditioned on a single exemplar image to reproduce its specific style. Extensive experiments demonstrate the efficacy of ManiFest on multiple tasks, outperforming the state-of-the-art on all metrics. Our code is avaliable at https://github.com/cv-rits/ManiFest.



Paperid:703
Authors:Nannan Li; Bryan A. Plummer
Title: Supervised Attribute Information Removal and Reconstruction for Image Manipulation
Abstract:
The goal of attribute manipulation is to control specified attribute(s) in given images. Prior work approaches this problem by learning disentangled representations for each attribute that enables it to manipulate the encoded source attributes to the target attributes. However, encoded attributes are often correlated with relevant image content. Thus, the source attribute information can often be hidden in the disentangled features, leading to unwanted image editing effects. In this paper, we propose an Attribute Information Removal and Reconstruction (AIRR) network that prevents such information hiding by learning how to remove the attribute information entirely, creating attribute excluded features, and then learns to directly inject the desired attributes in a reconstructed image. We evaluate our approach on four diverse datasets with a variety of attributes including DeepFashion Synthesis, DeepFashion Fine-grained Attribute, CelebA and CelebA-HQ, where our model improves attribute manipulation accuracy and top-k retrieval rate by 10% on average over prior work. A user study also reports that AIRR manipulated images are preferred over prior work in up to 76% of cases.



Paperid:704
Authors:Xiang Kong; Lu Jiang; Huiwen Chang; Han Zhang; Yuan Hao; Haifeng Gong; Irfan Essa
Title: BLT: Bidirectional Layout Transformer for Controllable Layout Generation
Abstract:
Creating visual layouts is a critical step in graphic design. Automatic generation of such layouts is essential for scalable and diverse visual designs. To advance conditional layout generation, we introduce BLT, a bidirectional layout transformer. BLT differs from previous work on transformers in adopting non-autoregressive transformers. In training, BLT learns to predict the masked attributes by attending to surrounding attributes in two directions. During inference, BLT first generates a draft layout from the input and then iteratively refines it into a high-quality layout by masking out low-confident attributes. The masks generated in both training and inference are controlled by a new hierarchical sampling policy. We verify the proposed model on six benchmarks of diverse design tasks. Experimental results demonstrate two benefits compared to the state-of-the-art layout transformer models. First, our model empowers layout transformers to fulfill controllable layout generation. Second, it achieves up to 10x speedup in generating a layout at inference time than the layout transformer baseline. Code is released at https://shawnkx.github.io/blt.



Paperid:705
Authors:Niv Haim; Ben Feinstein; Niv Granot; Assaf Shocher; Shai Bagon; Tali Dekel; Michal Irani
Title: Diverse Generation from a Single Video Made Possible
Abstract:
GANs are able to perform generation and manipulation tasks, trained on a single video. However, these single video GANs require unreasonable amount of time to train on a single video, rendering them almost impractical. In this paper we question the necessity of a GAN for generation from a single video, and introduce a non-parametric baseline for a variety of generation and manipulation tasks. We revive classical space-time patches-nearest-neighbors approaches and adapt them to a scalable unconditional generative model, without any learning. This simple baseline surprisingly outperforms single-video GANs in visual quality and realism (confirmed by quantitative and qualitative evaluations), and is disproportionately faster (runtime reduced from several days to seconds). Other than diverse video generation, we demonstrate other applications using the same framework, including video analogies and spatio-temporal retargeting. Our proposed approach is easily scaled to Full-HD videos. These observations show that the classical approaches, if adapted correctly, significantly outperform heavy deep learning machinery for these tasks. This sets a new baseline for single-video generation and manipulation tasks, and no less important -- makes diverse generation from a single video practically possible for the first time.



Paperid:706
Authors:Guha Balakrishnan; Raghudeep Gadde; Aleix Martinez; Pietro Perona
Title: Rayleigh EigenDirections (REDs): Nonlinear GAN Latent Space Traversals for Multidimensional Features
Abstract:
We present a method for finding paths in a deep generative model’s latent space that can maximally vary one set of image features while holding others constant. Crucially, unlike past traversal approaches, ours can manipulate arbitrary multidimensional features of an image such as facial identity and pixels within a specified region. Our method is principled and conceptually simple: optimal traversal directions are chosen by maximizing differential changes to one feature set such that changes to another set are negligible. We show that this problem is nearly equivalent to one of Rayleigh quotient maximization, and provide a closed-form solution to it based on solving a generalized eigenvalue equation. We use repeated computations of the corresponding optimal directions, which we call Rayleigh EigenDirections (REDs), to generate appropriately curved paths in latent space. We empirically evaluate our method using StyleGAN2 and BigGAN on the following image domains: faces, living rooms and ImageNet. We show that our method is capable of controlling various multidimensional features: face identity, geometric and semantic attributes, spatial frequency bands, pixels within a region, and the appearance and position of an object. Our work suggests that a wealth of opportunities lies in the local analysis of the geometry and semantics of latent spaces.



Paperid:707
Authors:Hyejin Lee; Daehee Kim; Daeun Lee; Jinkyu Kim; Jaekoo Lee
Title: Bridging the Domain Gap towards Generalization in Automatic Colorization
Abstract:
We propose a novel automatic colorization technique that learns domain-invariance across multiple source domains and is able to leverage such invariance to colorize grayscale images in unseen target domains. This would be particularly useful for colorizing sketches, line arts, or line drawings, which are generally difficult to colorize due to a lack of data. To address this issue, we first apply existing domain generalization (DG) techniques, which, however, produce less compelling desaturated images due to the network’s over-emphasis on learning domain-invariant contents (or shapes). Thus, we propose a new domain generalizable colorization model, which consists of two modules: (i) a domain-invariant content-biased feature encoder and (ii) a source-domain-specific color generator. To mitigate the issue of insufficient source domain-specific color information in domain-invariant features, we propose a skip connection that can transfer content feature statistics via adaptive instance normalization. Our experiments with publicly available PACS and Office-Home DG benchmarks confirm that our model is indeed able to produce perceptually reasonable colorized images. Further, we conduct a user study where human evaluators are asked to (1) answer whether the generated image looks naturally colored and to (2) choose the best-generated images against alternatives. Our model significantly outperforms the alternatives, confirming the effectiveness of the proposed method. The code is available at \url{https://github.com/Lhyejin/DG-Colorization}.



Paperid:708
Authors:Ariel Elnekave; Yair Weiss
Title: Generating Natural Images with Direct Patch Distributions Matching
Abstract:
Many traditional computer vision algorithms generate realistic images by requiring that each patch in the generated image be similar to a patch in a training image and vice versa. Recently, this classical approach has been replaced by adversarial training with a patch discriminator. The adversarial approach avoids the computational burden of finding nearest neighbors of patches but often requires very long training times and may fail to match the distribution of patches. In this paper we leverage the Sliced Wasserstein Distance to develop an algorithm that explicitly and efficiently minimizes the distance between patch distributions in two images. Our method is conceptually simple, requires no training and can be implemented in a few lines of codes. On a number of image generation tasks we show that our results are often superior to single-image-GANs, and can generate high quality images in a few seconds. Our implementation is publicly available at https://github.com/ariel415el/GPDM.



Paperid:709
Authors:Wuyang Luo; Su Yang; Hong Wang; Bo Long; Weishan Zhang
Title: Context-Consistent Semantic Image Editing with Style-Preserved Modulation
Abstract:
Semantic image editing utilizes local semantic label maps to generate the desired content in the edited region. A recent work borrows SPADE block to achieve semantic image editing. However, it cannot produce pleasing results due to style discrepancy between the edited region and surrounding pixels. We attribute this to the fact that SPADE only uses an image-independent local semantic layout but ignores the image-specific styles included in the known pixels. To address this issue, we propose a style-preserved modulation (SPM) comprising two modulations processes: The first modulation incorporates the contextual style and semantic layout, and then generates two fused modulation parameters. The second modulation employs the fused parameters to modulate feature maps. By using such two modulations, SPM can inject the given semantic layout while preserving the image-specific context style. Moreover, we design a progressive architecture for generating the edited content in a coarse-to-fine manner. The proposed method can obtain context-consistent results and significantly alleviate the unpleasant boundary between the generated regions and the known pixels.



Paperid:710
Authors:Zekun Li; Zhengyang Geng; Zhao Kang; Wenyu Chen; Yibo Yang
Title: Eliminating Gradient Conflict in Reference-Based Line-Art Colorization
Abstract:
Reference-based line-art colorization is a challenging task in computer vision. The color, texture, and shading are rendered based on an abstract sketch, which heavily relies on the precise long-range dependency modeling between the sketch and reference. Popular techniques to bridge the cross-modal information and model the long-range dependency employ the attention mechanism. However, in the context of reference-based line-art colorization, several techniques would intensify the existing training difficulty of attention, for instance, self-supervised training protocol and GAN-based losses. To understand the instability in training, we detect the gradient flow of attention and observe gradient conflict among attention branches. This phenomenon motivates us to alleviate the gradient issue by preserving the dominant gradient branch while removing the conflict ones. We propose a novel attention mechanism using this training strategy, Stop-Gradient Attention (SGA), outperforming the attention baseline by a large margin with better training stability. Compared with state-of-the-art modules in line-art colorization, our approach demonstrates significant improvements in Fréchet Inception Distance (FID, up to 27.21%) and structural similarity index measure (SSIM, up to 25.67%) on several benchmarks. The code of SGA is available at https://github.com/kunkun0w0/SGA.



Paperid:711
Authors:Atsuhiro Noguchi; Xiao Sun; Stephen Lin; Tatsuya Harada
Title: Unsupervised Learning of Efficient Geometry-Aware Neural Articulated Representations
Abstract:
We propose an unsupervised method for 3D geometry-aware representation learning of articulated objects, in which no image-pose pairs or foreground masks are used for training. Though photorealistic images of articulated objects can be rendered with explicit pose control through existing 3D neural representations, these methods require ground truth 3D pose and foreground masks for training, which are expensive to obtain. We obviate this need by learning the representations with GAN training. The generator is trained to produce realistic images of articulated objects from random poses and latent vectors by adversarial training. To avoid a high computational cost for GAN training, we propose an efficient neural representation for articulated objects based on tri-planes and then present a GAN-based framework for its unsupervised training. Experiments demonstrate the efficiency of our method and show that GAN-based training enables the learning of controllable 3D representations without paired supervision.



Paperid:712
Authors:Xi Wang; Xueyang Fu; Yurui Zhu; Zheng-Jun Zha
Title: JPEG Artifacts Removal via Contrastive Representation Learning
Abstract:
To meet the needs of practical applications, current deep learning-based methods focus on using a single model to handle JPEG images with different compression qualities, while few of them consider the auxiliary effects of the compression quality information. Recently, several methods estimate quality factors in a supervised learning manner to guide their network to remove JPEG artifacts. However, they may fail to estimate unseen compression types, affecting the subsequent restoration performance. To remedy this issue, we propose an unsupervised compression quality representation learning strategy for the blind JPEG artifacts removal. Specifically, we utilize contrastive learning to obtain discriminative compression quality representations in the latent feature space. Then, to fully exploit the learned representations, we design a compression-guided blind JPEG artifacts removal network, which integrates the discriminative compression quality representations in an information lossless way. In this way, our single network can flexibly handle various JPEG compression images. Experiments demonstrate that our method can adapt to different compression qualities to obtain discriminative representations and outperform state-of-art methods.



Paperid:713
Authors:Xiang Chen; Zhentao Fan; Pengpeng Li; Longgang Dai; Caihua Kong; Zhuoran Zheng; Yufeng Huang; Yufeng Li
Title: Unpaired Deep Image Dehazing Using Contrastive Disentanglement Learning
Abstract:
We offer a practical unpaired learning based image dehazing network from an unpaired set of clear and hazy images. This paper provides a new perspective to treat image dehazing as a two-class separated factor disentanglement task, i.e, the task-relevant factor of clear image reconstruction and the task-irrelevant factor of haze-relevant distribution. To achieve the disentanglement of these two-class factors in deep feature space, contrastive learning is introduced into a CycleGAN framework to learn disentangled representations by guiding the generated images to be associated with latent factors. With such formulation, the proposed contrastive disentangled dehazing method (CDD-GAN) employs negative generators to cooperate with the encoder network to update alternately, so as to produce a queue of challenging negative adversaries. Then these negative adversaries are trained end-to-end together with the backbone representation network to enhance the discriminative information and promote factor disentanglement performance by maximizing the adversarial contrastive loss. During the training, we further show that hard negative examples can suppress the task-irrelevant factors and unpaired clear exemples can enhance the task-relevant factors, in order to better facilitate haze removal and help image restoration. Extensive experiments on both synthetic and real-world datasets demonstrate that our method performs favorably against existing unpaired dehazing baselines.



Paperid:714
Authors:Xindong Zhang; Hui Zeng; Shi Guo; Lei Zhang
Title: Efficient Long-Range Attention Network for Image Super-Resolution
Abstract:
Recently, transformer-based methods have demonstrated impressive results in various vision tasks, including image super-resolution (SR), by exploiting the self attention (SA) for feature extraction. However, the computation of SA in most existing transformer based models is very expensive, while some employed operations may be redundant for the SR task. This limits the range of SA computation and consequently limits the SR performance. In this work, we propose an efficient long-range attention network (ELAN) for image SR. Specifically, we first employ shift convolution (shift-conv) to effectively extract the image local structural information while maintaining the same level of complexity as 1x1 convolution, then propose a group-wise multi-scale self-attention (GMSA) module, which calculates SA on non-overlapped groups of features using different window sizes to exploit the long-range image dependency. A highly efficient long-range attention block (ELAB) is then built by simply cascading two shift-conv with a GMSA module, which is further accelerated by using a shared attention mechanism. Without bells and whistles, our ELAN follows a fairly simple design by sequentially cascading the ELABs. Extensive experiments demonstrate that ELAN obtains even better results against the transformer-based SR models but with significantly less complexity. The source codes of ELAN can be found at https://github.com/xindongzhang/ELAN.



Paperid:715
Authors:Zhaoyang Huang; Xiaoyu Shi; Chao Zhang; Qiang Wang; Ka Chun Cheung; Hongwei Qin; Jifeng Dai; Hongsheng Li
Title: FlowFormer: A Transformer Architecture for Optical Flow
Abstract:
We introduce optical Flow transFormer, dubbed as FlowFormer, a transformer-based neural network architecture for learning optical flow. FlowFormer tokenizes the 4D cost volume built from an image pair, encodes the cost tokens into a cost memory with alternate-group transformer (AGT) layers in a novel latent space, and decodes the cost memory via a recurrent transformer decoder with dynamic positional cost queries. On the Sintel benchmark, FlowFormer achieves 1.144 and 2.183 average end-ponit-error (AEPE) on the clean and final pass, a 17.6% and 11.6% error reduction from the best published result (1.388 and 2.47). Besides, FlowFormer also achieves strong generalization performance. Without being trained on Sintel, FlowFormer achieves 0.95 AEPE on the Sintel training set clean pass, outperforming the best published result (1.29) by 26.9%.



Paperid:716
Authors:Yuanhao Cai; Jing Lin; Xiaowan Hu; Haoqian Wang; Xin Yuan; Yulun Zhang; Radu Timofte; Luc Van Gool
Title: Coarse-to-Fine Sparse Transformer for Hyperspectral Image Reconstruction
Abstract:
Many learning-based algorithms have been developed to solve the inverse problem of coded aperture snapshot spectral imaging (CASSI). However, CNN-based methods show limitations in capturing long-range dependencies. Previous Transformer-based methods densely sample tokens, some of which are uninformative, and calculate multi-head self-attention (MSA) between some tokens that are unrelated in content. In this paper, we propose a novel Transformer-based method, coarse-to-fine sparse Transformer (CST), firstly embedding HSI sparsity into deep learning for HSI reconstruction. In particular, CST uses our proposed spectra-aware screening mechanism (SASM) for coarse patch selecting. Then the selected patches are fed into our customized spectra-aggregation hashing multi-head self-attention (SAH-MSA) for fine pixel clustering and self-similarity capturing. Comprehensive experiments show that our CST significantly outperforms state-of-the-art methods while requiring cheaper computational costs. https://github.com/caiyuanhao1998/MST"



Paperid:717
Authors:Xinpeng Ding; Jingwen Yang; Xiaowei Hu; Xiaomeng Li
Title: Learning Shadow Correspondence for Video Shadow Detection
Abstract:
Video shadow detection aims to generate consistent shadow predictions among video frames. However, the current approaches suffer from inconsistent shadow predictions across frames, especially when the illumination and background textures change in a video. We make an observation that the inconsistent predictions are caused by the shadow feature inconsistency, i.e., the features of the same shadow regions show dissimilar proprieties among the nearby frames. In this paper, we present a novel Shadow-Consistent Correspondence method (SC-Cor) to enhance pixel-wise similarity of the specific shadow regions across frames for video shadow detection. Our proposed SC-Cor has three main advantages. Firstly, without requiring the dense pixel-to-pixel correspondence labels, SC-Cor can learn the pixel-wise correspondence across frames in a weakly-supervised manner. Secondly, SC-Cor considers intra-shadow separability, which is robust to the variant textures and illuminations in videos. Finally, SC-Cor is a plug-and-play module that can be easily integrated into existing shadow detectors with no extra computational cost. We further design a new evaluation metric to evaluate the temporal stability of the video shadow detection results. Experimental results show that SC-Cor outperforms the prior state-of-the-art method, by 6.51% on IoU and 3.35% on the newly introduced temporal stability metric.



Paperid:718
Authors:Chong Mou; Yanze Wu; Xintao Wang; Chao Dong; Jian Zhang; Ying Shan
Title: Metric Learning Based Interactive Modulation for Real-World Super-Resolution
Abstract:
Interactive image restoration aims to restore images by adjusting several controlling coefficients, which determine the restoration strength. Existing methods are restricted in learning the controllable functions under the supervision of known degradation types and levels. They usually suffer from a severe performance drop when the real degradation is different from their assumptions. Such a limitation is due to the complexity of real-world degradations, which can not provide explicit supervision to the interactive modulation during training. However, how to realize the interactive modulation in real-world super-resolution has not yet been studied. In this work, we present a Metric Learning based Interactive Modulation for Real-World \Super-Resolution (MM-RealSR). Specifically, we propose an unsupervised degradation estimation strategy to estimate the degradation level in real-world scenarios. Instead of using known degradation levels as explicit supervision to the interactive mechanism, we propose a metric learning strategy to map the unquantifiable degradation levels in real-world scenarios to a metric space, which is trained in an unsupervised manner. Moreover, we introduce an anchor point strategy in the metric learning process to normalize the distribution of metric space. Extensive experiments demonstrate that the proposed MM-RealSR achieves excellent modulation and restoration performance in real-world super-resolution. Codes are available at https://github.com/TencentARC/MM-RealSR.



Paperid:719
Authors:Yunshan Zhong; Mingbao Lin; Xunchao Li; Ke Li; Yunhang Shen; Fei Chao; Yongjian Wu; Rongrong Ji
Title: Dynamic Dual Trainable Bounds for Ultra-Low Precision Super-Resolution Networks
Abstract:
Light-weight super-resolution (SR) models have received considerable attention for their serviceability in mobile devices. Many efforts employ network quantization to compress SR models. However, these methods suffer from severe performance degradation when quantizing the SR models to ultra-low precision (e.g., 2-bit and 3-bit) with the low-cost layer-wise quantizer. In this paper, we identify that the performance drop comes from the contradiction between the layer-wise symmetric quantizer and the highly asymmetric activation distribution in SR models. This discrepancy leads to either a waste on the quantization levels or detail loss in reconstructed images. Therefore, we propose a novel activation quantizer, referred to as Dynamic Dual Trainable Bounds (DDTB), to accommodate the asymmetry of the activations. Specifically, DDTB innovates in: 1) A layer-wise quantizer with trainable upper and lower bounds to tackle the highly asymmetric activations. 2) A dynamic gate controller to adaptively adjust the upper and lower bounds at runtime to overcome the drastically varying activation ranges over different samples. To reduce the extra overhead, the dynamic gate controller is quantized to 2-bit and applied to only part of the SR networks according to the introduced dynamic intensity. Extensive experiments demonstrate that our DDTB exhibits significant performance improvements in ultra-low precision. For example, our DDTB achieves a 0.70dB PSNR increase on Urban100 benchmark when quantizing EDSR to 2-bit and scaling up output images to x4.



Paperid:720
Authors:Jialun Pei; Tianyang Cheng; Deng-Ping Fan; He Tang; Chuanbo Chen; Luc Van Gool
Title: OSFormer: One-Stage Camouflaged Instance Segmentation with Transformers
Abstract:
We present OSFormer, the first one-stage transformer framework for camouflaged instance segmentation (CIS). OSFormer is based on two key designs. First, we design a location-sensing transformer (LST) to obtain the location label and instance-aware parameters by introducing the location-guided queries and the blend-convolution feed-forward network. Second, we develop a coarse-to-fine fusion (CFF) to merge diverse context information from the LST encoder and CNN backbone. Coupling these two components enables OSFormer to efficiently blend local features and long-range context dependencies for predicting camouflaged instances. Compared with two-stage frameworks, our OSFormer reaches 41% AP and achieves good convergence efficiency without requiring enormous training data, i.e., only 3,040 samples under 60 epochs. Code link: https://github.com/PJLallen/OSFormer.



Paperid:721
Authors:Xuebin Qin; Hang Dai; Xiaobin Hu; Deng-Ping Fan; Ling Shao; Luc Van Gool
Title: Highly Accurate Dichotomous Image Segmentation
Abstract:
We present a systematic study on a new task called dichotomous image segmentation (DIS), which aims to segment highly accurate objects from natural images. To this end, we collected the first large-scale DIS dataset, called DIS5K, which contains 5,470 high-resolution (e.g., 2K, 4K or larger) images covering camouflaged, salient, or meticulous objects in various backgrounds. DIS is annotated with extremely fine-grained labels. Besides, we introduce a simple intermediate supervision baseline (IS-Net) using both feature-level and mask-level guidance for DIS model training. IS-Net outperforms various cutting-edge baselines on the proposed DIS5K, making it a general self-learned supervision network that can facilitate future research in DIS. Further, we design a new metric called human correction efforts (HCE) which approximates the number of mouse clicking operations required to correct the false positives and false negatives. HCE is utilized to measure the gap between models and real-world applications and thus can complement existing metrics. Finally, we conduct the largest-scale benchmark, evaluating 16 representative segmentation models, providing a more insightful discussion regarding object complexities, and showing several potential applications (e.g., background removal, art design, 3D reconstruction). Hoping these efforts can open up promising directions for both academic and industries. Project page: https://xuebinqin.github.io/dis/index.html.



Paperid:722
Authors:Xingyu Jiang; Hongkun Dou; Chengwei Fu; Bingquan Dai; Tianrun Xu; Yue Deng
Title: Boosting Supervised Dehazing Methods via Bi-Level Patch Reweighting
Abstract:
Natural images can suffer from non-uniform haze distributions in different regions. However, this important fact is hardly considered in existing supervised dehazing methods, in which all training patches are accounted for equally in the loss design. These supervised methods may fail in making promising recoveries on some regions contaminated by heavy hazes. Therefore, for a more reasonable dehazing losses design, the varying importance of different training patches should be taken into account. Such rationale is exactly in line with the process of human learning that difficult concepts always require more practice in learning. To this end, we propose a bi-level dehazing (BILD) framework by designing an internal loop for weighted supervised dehazing and an external loop for training patch reweighting. With simple derivations, we show the gradients of BILD exhibit natural connections with policy gradient and can thus explain the BILD objective by the rewarding mechanism in reinforcement learning. The BILD is not a new dehazing method per se, it is better recognized as a flexible framework that can seamlessly work with general supervised dehazing approaches for their performance boosting.



Paperid:723
Authors:Kaidong Zhang; Jingjing Fu; Dong Liu
Title: Flow-Guided Transformer for Video Inpainting
Abstract:
We propose a flow-guided transformer, which innovatively leverage the motion discrepancy exposed by optical flows to instruct the attention retrieval in transformer for high fidelity video inpainting. More specially, we design a novel flow completion network to complete the corrupted flows by exploiting the relevant flow features in a local temporal window. With the completed flows, we propagate the content across video frames, and adopt the flow-guided transformer to synthesize the rest corrupted regions. We decouple transformers along temporal and spatial dimension, so that we can easily integrate the locally relevant completed flows to instruct spatial attention only. Furthermore, we design a flow-reweight module to precisely control the impact of completed flows on each spatial transformer. For the sake of efficiency, we introduce window partition strategy to both spatial and temporal transformers. Especially in spatial transformer, we design a dual perspective spatial MHSA, which integrates the global tokens to the window-based attention. Extensive experiments demonstrate the effectiveness of the proposed method qualitatively and quantitatively. Codes are available at https://github.com/hitachinsk/FGT.



Paperid:724
Authors:Abhijay Ghildyal; Feng Liu
Title: Shift-tolerant Perceptual Similarity Metric
Abstract:
Existing perceptual similarity metrics assume an image and its reference are well aligned. As a result, these metrics are often sensitive to a small alignment error that is imperceptible to the human eyes. This paper studies the effect of small misalignment, specifically a small shift between the input and reference image, on existing metrics, and accordingly develops a shift-tolerant similarity metric. This paper builds upon LPIPS, a widely used learned perceptual similarity metric, and explores architectural design considerations to make it robust against imperceptible misalignment. Specifically, we study a wide spectrum of neural network elements, such as anti-aliasing filtering, pooling, striding, padding, and skip connection, and discuss their roles in making a robust metric. Based on our studies, we develop a new deep neural network-based perceptual similarity metric. Our experiments show that our metric is tolerant to imperceptible shifts while being consistent with the human similarity judgment. Code is available at https://tinyurl.com/5n85r28r.



Paperid:725
Authors:Yuehan Zhang; Bo Ji; Jia Hao; Angela Yao
Title: Perception-Distortion Balanced ADMM Optimization for Single-Image Super-Resolution
Abstract:
In image super-resolution, both pixel-wise accuracy and perceptual fidelity are desirable. However, most deep learning methods only achieve high performance in one aspect due to the perception-distortion trade-off, and works that successfully balance the trade-off rely on fusing results from separately trained models with ad-hoc post-processing. In this paper, we propose a novel super-resolution model with a low-frequency constraint (LFc-SR), which balances the objective and perceptual quality through a single model and yields super-resolved images with high PSNR and perceptual scores. We further introduce an ADMM-based alternating optimization method for the non-trivial learning of the constrained model. Experiments showed that our method, without cumbersome post-processing procedures, achieved the state-of-the-art performance. The code is available at https://github.com/Yuehan717/PDASR.



Paperid:726
Authors:Yuchao Gu; Xintao Wang; Liangbin Xie; Chao Dong; Gen Li; Ying Shan; Ming-Ming Cheng
Title: VQFR: Blind Face Restoration with Vector-Quantized Dictionary and Parallel Decoder
Abstract:
Although generative facial prior and geometric prior have recently demonstrated high-quality results for blind face restoration, producing fine-grained facial details faithful to inputs remains a challenging problem. Motivated by the classical dictionary-based methods and the recent vector quantization (VQ) technique, we propose a VQ-based face restoration method - VQFR. VQFR takes advantage of high-quality low-level feature banks extracted from high-quality faces and can thus help recover realistic facial details. However, the simple application of the VQ codebook cannot achieve good results with faithful details and identity preservation. Therefore, we further introduce two special network designs. 1). We first investigate the compression patch size in the VQ codebook and find that the VQ codebook designed with a proper compression patch size is crucial to balance the quality and fidelity. 2). To further fuse low-level features from inputs while not “contaminating” the realistic details generated from the VQ codebook, we proposed a parallel decoder consisting of a texture decoder and a main decoder. Those two decoders then interact with a texture warping module with deformable convolution. Equipped with the VQ codebook as a facial detail dictionary and the parallel decoder design, the proposed VQFR can largely enhance the restored quality of facial details while keeping the fidelity to previous methods.



Paperid:727
Authors:Zhenxuan Fang; Weisheng Dong; Xin Li; Jinjian Wu; Leida Li; Guangming Shi
Title: Uncertainty Learning in Kernel Estimation for Multi-stage Blind Image Super-Resolution
Abstract:
Conventional wisdom in blind super-resolution (SR) first estimates the unknown degradation from the low-resolution image and then exploits the degradation information for image reconstruction. Such sequential approaches suffer from two fundamental weaknesses - i.e., the lack of robustness (the performance drops when the estimated degradation is inaccurate) and the lack of transparency (network architectures are heuristic without incorporating domain knowledge). To address these issues, we propose a joint Maximum a Posterior (MAP) approach for estimating the unknown kernel and high-resolution image simultaneously. Our method first introduces uncertainty learning in the latent space when estimating the blur kernel, aiming at improving the robustness to the estimation error. Then we propose a novel SR network by unfolding the joint MAP estimator with a learned Laplacian Scale Mixture (LSM) prior and the estimated kernel. We have also developed a novel approach of estimating both the scale prior coefficient and the local means of the LSM model through a deep convolutional neural network (DCNN). All parameters of the MAP estimation algorithm and the DCNN parameters are jointly optimized through end-to-end training. Extensive experiments on both synthetic and real-world images show that our method achieves state-of-the-art performance for the task of blind image SR.



Paperid:728
Authors:Xiaoyu Xiang; Yapeng Tian; Vijay Rengarajan; Lucas D. Young; Bo Zhu; Rakesh Ranjan
Title: Learning Spatio-Temporal Downsampling for Effective Video Upscaling
Abstract:
Downsampling is one of the most basic image processing operations. Improper spatio-temporal downsampling applied on videos can cause aliasing issues such as moiré patterns in space and the wagon-wheel effect in time. Consequently, the inverse task of upscaling a low-resolution, low frame-rate video in space and time becomes a challenging ill-posed problem due to information loss and aliasing artifacts. In this paper, we aim to solve the space-time aliasing problem by learning a spatio-temporal downsampler. Towards this goal, we propose a neural network framework that jointly learns spatio-temporal downsampling and upsampling. It enables the downsampler to retain the key patterns of the original video and maximizes the reconstruction performance of the upsampler. To make the downsamping results compatible with popular image and video storage formats, the downsampling results are encoded to uint8 with a differentiable quantization layer. To fully utilize the space-time correspondences, we propose two novel modules for explicit temporal propagation and space-time feature rearrangement. Experimental results show that our proposed method significantly boosts the space-time reconstruction quality by preserving spatial textures and motion patterns in both downsampling and upscaling. Moreover, our framework enables a variety of applications, including arbitrary video resampling, blurry frame reconstruction, and efficient video storage.



Paperid:729
Authors:Jaewon Lee; Kwang Pyo Choi; Kyong Hwan Jin
Title: Learning Local Implicit Fourier Representation for Image Warping
Abstract:
Image warping aims to reshape images defined on rectangular grids into arbitrary shapes. Recently, implicit neural functions have shown remarkable performances in representing images in a continuous manner. However, a standalone multi-layer perceptron suffers from learning high-frequency Fourier coefficients. In this paper, we propose a local texture estimator for image warping (LTEW) followed by an implicit neural representation to deform images into continuous shapes. Local textures estimated from a deep super-resolution (SR) backbone are multiplied by locally-varying Jacobian matrices of a coordinate transformation to predict Fourier responses of a warped image. Our LTEW-based neural function outperforms existing warping methods for asymmetric-scale SR and homography transform. Furthermore, our algorithm well generalizes arbitrary coordinate transformations, such as homography transform with a large magnification factor and equirectangular projection (ERP) perspective transform, which are not provided in training.



Paperid:730
Authors:Canqian Yang; Meiguang Jin; Yi Xu; Rui Zhang; Ying Chen; Huaida Liu
Title: SepLUT: Separable Image-Adaptive Lookup Tables for Real-Time Image Enhancement
Abstract:
Image-adaptive lookup tables (LUTs) have achieved great success in real-time image enhancement tasks due to their high efficiency for modeling color transforms. However, they embed the complete transform, including the color component-independent and the component-correlated parts, into only a single type of LUTs, either 1D or 3D, in a coupled manner. This scheme raises a dilemma of improving model expressiveness or efficiency due to two factors. On the one hand, the 1D LUTs provide high computational efficiency but lack the critical capability of color components interaction. On the other, the 3D LUTs present enhanced component-correlated transform capability but suffer from heavy memory footprint, high training difficulty, and limited cell utilization. Inspired by the conventional divide-and-conquer practice in the image signal processor, we present SepLUT (separable image-adaptive lookup table) to tackle the above limitations. Specifically, we separate a single color transform into a cascade of component-independent and component-correlated sub-transforms instantiated as 1D and 3D LUTs, respectively. In this way, the capabilities of two sub-transforms can facilitate each other, where the 3D LUT complements the ability to mix up color components, and the 1D LUT redistributes the input colors to increase the cell utilization of the 3D LUT and thus enable the use of a more lightweight 3D LUT. Experiments demonstrate that the proposed method presents enhanced performance on photo retouching benchmark datasets than the current state-of-the-art and achieves real-time processing on both GPUs and CPUs.



Paperid:731
Authors:Junlin Han; Weihao Li; Pengfei Fang; Chunyi Sun; Jie Hong; Mohammad Ali Armin; Lars Petersson; Hongdong Li
Title: Blind Image Decomposition
Abstract:
We propose and study a novel task named Blind Image Decomposition (BID), which requires separating a superimposed image into constituent underlying images in a blind setting, that is, both the source components involved in mixing as well as the mixing mechanism are unknown. For example, rain may consist of multiple components, such as rain streaks, raindrops, snow, and haze. Rainy images can be treated as an arbitrary combination of these components, some of them or all of them. How to decompose superimposed images, like rainy images, into distinct source components is a crucial step toward real-world vision systems. To facilitate research on this new task, we construct multiple benchmark datasets, including mixed image decomposition across multiple domains, real-scenario deraining, and joint shadow/reflection/watermark removal. Moreover, we propose a simple yet general Blind Image Decomposition Network (BIDeN) to serve as a strong baseline for future work. Experimental results demonstrate the tenability of our benchmarks and the effectiveness of BIDeN.



Paperid:732
Authors:Jiacheng Li; Chang Chen; Zhen Cheng; Zhiwei Xiong
Title: MuLUT: Cooperating Multiple Look-Up Tables for Efficient Image Super-Resolution
Abstract:
The high-resolution screen of edge devices stimulates a strong demand for efficient image super-resolution (SR). An emerging research, SR-LUT, responds to this demand by marrying the look-up table (LUT) with learning-based SR methods. However, the size of a single LUT grows exponentially with the increase of its indexing capacity. Consequently, the receptive field of a single LUT is restricted, resulting in inferior performance. To address this issue, we extend SR-LUT by enabling the cooperation of Multiple LUTs, termed MuLUT. Firstly, we devise two novel complementary indexing patterns and construct multiple LUTs in parallel. Secondly, we propose a re-indexing mechanism to enable the hierarchical indexing between multiple LUTs. In these two ways, the total size of MuLUT is linear to its indexing capacity, yielding a practical method to obtain superior performance. We examine the advantage of MuLUT on five SR benchmarks. MuLUT achieves a significant improvement over SR-LUT, up to 1.1dB PSNR, while preserving its efficiency. Moreover, we extend MuLUT to address demosaicing of Bayer-patterned images, surpassing SR-LUT on two benchmarks by a large margin.



Paperid:733
Authors:Zhongwei Qiu; Huan Yang; Jianlong Fu; Dongmei Fu
Title: Learning Spatiotemporal Frequency-Transformer for Compressed Video Super-Resolution
Abstract:
Compressed video super-resolution (VSR) aims to restore high-resolution frames from compressed low-resolution counterparts. Most recent VSR approaches often enhance an input frame by “borrowing’’ relevant textures from neighboring video frames. Although some progress has been made, there are grand challenges to effectively extract and transfer high-quality textures from compressed videos where most frames are usually highly degraded. In this paper, we propose a novel Frequency-Transformer for compressed video super-resolution (FTVSR) that conducts self-attention over a joint space-time-frequency domain. First, we divide a video frame into patches, and transform each patch into DCT spectral maps in which each channel represents a frequency band. Such a design enables a fine-grained level self-attention on each frequency band, so that real visual texture can be distinguished from artifacts, and further utilized for video frame restoration. Second, we study different self-attention schemes, and discover that a “divided attention” which conducts a joint space-frequency attention before applying temporal attention on each frequency band, leads to the best video enhancement quality. Experimental results on two widely-used video super-resolution benchmarks show that FTVSR outperforms state-of-the-art approaches on both uncompressed and compressed videos with clear visual margins. Code are available at https://github.com/researchmm/FTVSR.



Paperid:734
Authors:Man Zhou; Jie Huang; Keyu Yan; Hu Yu; Xueyang Fu; Aiping Liu; Xian Wei; Feng Zhao
Title: Spatial-Frequency Domain Information Integration for Pan-Sharpening
Abstract:
Pan-sharpening aims to generate the high-resolution multi-spectral (MS) images by fusing PAN images and low-resolution MS images. Despite the great advances, most existing pan-sharpening methods only work in the spatial domain and rarely explore the potential solution in frequency domain. In this paper, we first attempt to address pan-sharpening in both spatial-frequency domain and propose a Spatial-Frequency Information Integration Network, dubbed as SFIIN. To implement SFIIN, we devise a core building module tailored with pan-sharpening, consisting of three key components: spatial-domain information branch, frequency-domain information one and dual domain interaction. To be specific, the first employs the standard convolution to integrate the local information of two modalities of PAN and MS images in the spatial domain while the second adopts deep Fourier transformation to achieve the image-wide receptive field for exploring global contextual information. Followed by, the third is responsible for facilitating the information flow and learning the complementary representation. We conduct extensive experiments to analyze the effectiveness of the proposed network and demonstrate the favorable performance against state-of-the-art methods.



Paperid:735
Authors:Shizun Wang; Jiaming Liu; Kaixin Chen; Xiaoqi Li; Ming Lu; Yandong Guo
Title: Adaptive Patch Exiting for Scalable Single Image Super-Resolution
Abstract:
Since the future of computing is heterogeneous, scalability is a crucial problem for single image super-resolution. Recent works try to train one network, which can be deployed on platforms with different capacities. However, they rely on the pixel-wise sparse convolution, which is not hardware-friendly and achieves limited practical speedup. As image can be divided into patches, which have various restoration difficulties, we present a scalable method based on Adaptive Patch Exiting (APE) to achieve more practical speedup. Specifically, we propose to train a regressor to predict the incremental capacity of each layer for the patch. Once the incremental capacity is below the threshold, the patch can exit at the specific layer. Our method can easily adjust the trade-off between performance and efficiency by changing the threshold of incremental capacity. Furthermore, we propose a novel strategy to enable the network training of our method. We conduct extensive experiments across various backbones, datasets and scaling factors to demonstrate the advantages of our method. Code is available at https://github.com/littlepure2333/APE"



Paperid:736
Authors:Xiaoqi Li; Jiaming Liu; Shizun Wang; Cheng Lyu; Ming Lu; Yurong Chen; Anbang Yao; Yandong Guo; Shanghang Zhang
Title: Efficient Meta-Tuning for Content-Aware Neural Video Delivery
Abstract:
Recently, Deep Neural Networks (DNNs) are utilized to reduce the bandwidth and improve the quality of Internet video delivery. Existing methods train corresponding content-aware super-resolution (SR) model for each video chunk on the server, and stream low-resolution (LR) video chunks along with SR models to the client. Although they achieve promising results, the huge computational cost of network training limits their practical applications. In this paper, we present a method named Efficient Meta-Tuning (EMT) to reduce the computational cost. Instead of training from scratch, EMT adapts a meta-learned model to the first chunk of the input video. As for the following chunks, it fine-tunes the partial parameters selected by gradient masking of previous adapted model. In order to achieve further speedup for EMT, we propose a novel sampling strategy to extract the most challenging patches from video frames. The proposed strategy is highly efficient and brings negligible additional cost. Our method significantly reduces the computational cost and achieves even better performance, paving the way for applying neural video delivery techniques to practical applications. We conduct extensive experiments based on various efficient SR architectures, including ESPCN, SRCNN, FSRCNN and EDSR-1, demonstrating the generalization ability of our work. The code is released at https://github.com/Neural-video-delivery/EMT-Pytorch-ECCV2022.



Paperid:737
Authors:Jiezhang Cao; Jingyun Liang; Kai Zhang; Yawei Li; Yulun Zhang; Wenguan Wang; Luc Van Gool
Title: Reference-Based Image Super-Resolution with Deformable Attention Transformer
Abstract:
Reference-based image super-resolution (RefSR) aims to exploit auxiliary reference (Ref) images to super-resolve low-resolution (LR) images. Recently, RefSR has been attracting great attention as it provides an alternative way to surpass single image SR. However, addressing the RefSR problem has two critical challenges: (i) It is difficult to match the correspondence between LR and Ref images when they are significantly different; (ii) How to transfer the relevant texture from Ref images to compensate the details for LR images is very challenging. To address these issues of RefSR, this paper proposes a deformable attention Transformer, namely DATSR, with multiple scales, each of which consists of a texture feature encoder (TFE) module, a reference-based deformable attention (RDA) module and a residual feature aggregation (RFA) module. Specifically, TFE first extracts image transformation (e.g., brightness) insensitive features for LR and Ref images, RDA then can exploit multiple relevant textures to compensate more information for LR features, and RFA lastly aggregates LR features and relevant textures to get a more visually pleasant result. Extensive experiments demonstrate that our DATSR achieves state-of-the-art performance on benchmark datasets quantitatively and qualitatively.



Paperid:738
Authors:Haoyuan Wang; Ke Xu; Rynson W.H. Lau
Title: Local Color Distributions Prior for Image Enhancement
Abstract:
Existing image enhancement methods are typically designed to address either the over- or under-exposure problem in the input image. When the illumination of the input image contains both over- and under-exposure problems, these existing methods may not work well. We observe from the image statistics that the local color distributions (LCDs) of an image suffering from both problems tend to vary across different regions of the image, depending on the local illuminations. Based on this observation, we propose in this paper to exploit these LCDs as a prior for locating and enhancing the two types of regions (i.e., over-/under-exposed regions). First, we leverage the LCDs to represent these regions, and propose a novel local color distribution embedded (LCDE) module to formulate LCDs in multi-scales to model the correlations across different regions. Second, we propose a dual-illumination learning mechanism to enhance the two types of regions. Third, we construct a new dataset to facilitate the learning process, by following the camera image signal processing (ISP) pipeline to render standard RGB images with both under-/over-exposures from raw data. Extensive experiments demonstrate that the proposed method outperforms existing state-of-the-art methods quantitatively and qualitatively. Codes and dataset are in https://hywang99.github.io/lcdpnet/.



Paperid:739
Authors:Zheng Chang; Shuchen Weng; Yu Li; Si Li; Boxin Shi
Title: L-CoDer: Language-Based Colorization with Color-Object Decoupling Transformer
Abstract:
Language-based colorization requires the colorized image to be consistent with the the user-provided language caption. A most recent work proposes to decouple the language into color and object conditions in solving the problem. Though decent progress has been made, its performance is limited by three key issues. (i) The large gap between vision and language modalities using independent feature extractors makes it difficult to fully understand the language. (ii) The inaccurate language features are never refined by the image features such that the language may fail to colorize the image precisely. (iii) The local region does not perceive the whole image, producing global inconsistent colors. In this work, we introduce transformer into language-based colorization to tackle the aforementioned issues while keeping the language decoupling property. Our method unifies the modalities of image and language, and further performs color conditions evolving with image features in a coarse-to-fine manner. In addition, thanks to the global receptive field, our method is robust to the strong local variation. Extensive experiments demonstrate our method is able to produce realistic colorization and outperforms prior arts in terms of consistency with the caption.



Paperid:740
Authors:Xiaoming Li; Chaofeng Chen; Xianhui Lin; Wangmeng Zuo; Lei Zhang
Title: From Face to Natural Image: Learning Real Degradation for Blind Image Super-Resolution
Abstract:
How to design proper training pairs is critical for super-resolving real-world low-quality (LQ) images, which suffers from the difficulties in either acquiring paired ground-truth high-quality (HQ) images or synthesizing photo-realistic degraded LQ observations. Recent works mainly focus on modeling the degradation with handcrafted or estimated degradation parameters, which are however incapable to model complicated real-world degradation types, resulting in limited quality improvement. Notably, LQ face images, which may have the same degradation process as natural images, can be robustly restored with photo-realistic textures by exploiting their strong structural priors. This motivates us to use the real-world LQ face images and their restored HQ counterparts to model the complex real-world degradation (namely ReDegNet), and then transfer it to HQ natural images to synthesize their realistic LQ counterparts. By taking these paired HQ-LQ face images as inputs to explicitly predict the degradation-aware and content-independent representations, we could control the degraded image generation, and subsequently transfer these degradation representations from face to natural images to synthesize the degraded LQ natural images. Experiments show that our ReDegNet can well learn the real degradation process from face images. The restoration network trained with our synthetic pairs performs favorably against SOTAs. More importantly, our method provides a new way to handle the real-world complex scenarios by learning their degradation representations from the facial portions, which can be used to significantly improve the quality of non-facial areas. The source code is available at https://github.com/csxmli2016/ReDegNet.



Paperid:741
Authors:Jiezhang Cao; Jingyun Liang; Kai Zhang; Wenguan Wang; Qin Wang; Yulun Zhang; Hao Tang; Luc Van Gool
Title: Towards Interpretable Video Super-Resolution via Alternating Optimization
Abstract:
In this paper, we study a practical space-time video super-resolution (STVSR) problem which aims at generating a high-framerate high-resolution sharp video from a low-framerate low-resolution blurry video. Such problem often occurs when recording a fast dynamic event with a low-framerate and low-resolution camera, and the captured video would suffer from three typical issues: i) motion blur occurs due to object/camera motions during exposure time; ii) motion aliasing is unavoidable when the event temporal frequency exceeds the Nyquist limit of temporal sampling; iii) high-frequency details are lost because of the low spatial sampling rate. These issues can be alleviated by a cascade of three separate sub-tasks, including video deblurring, frame interpolation, and super-resolution, which, however, would fail to capture the spatial and temporal correlations among video sequences. To address this, we propose an interpretable STVSR framework by leveraging both model-based and learning-based methods. Specifically, we formulate STVSR as a joint video deblurring, frame interpolation, and super-resolution problem, and solve it as two sub-problems in an alternate way. For the first sub-problem, we derive an interpretable analytical solution and use it as a Fourier data transform layer. Then, we propose a recurrent video enhancement layer for the second sub-problem to further recover high-frequency details. Extensive experiments demonstrate the superiority of our method in terms of quantitative metrics and visual quality.



Paperid:742
Authors:Lei Sun; Christos Sakaridis; Jingyun Liang; Qi Jiang; Kailun Yang; Peng Sun; Yaozu Ye; Kaiwei Wang; Luc Van Gool
Title: Event-Based Fusion for Motion Deblurring with Cross-Modal Attention
Abstract:
Traditional frame-based cameras inevitably suffer from motion blur due to long exposure times. As a kind of bio-inspired camera, the event camera records the intensity changes in an asynchronous way with high temporal resolution, providing valid image degradation information within the exposure time. In this paper, we rethink the event-based image deblurring problem and unfold it into an end-to-end two-stage image restoration network. To effectively fuse event and image features, we design an event-image cross-modal attention module applied at multiple levels of our network, which allows to focus on relevant features from the event branch and filter out noise. We also introduce a novel symmetric cumulative event representation specifically for image deblurring as well as an event mask gated connection between the two stages of our network which helps avoid information loss. At the dataset level, to foster event-based motion deblurring and to facilitate evaluation on challenging real-world images, we introduce the Real Event Blur (REBlur) dataset, captured with an event camera in an illumination-controlled optical laboratory. Our Event Fusion Network (EFNet) sets the new state of the art in motion deblurring, surpassing both the prior best-performing image-based method and all event-based methods with public implementations on the GoPro dataset (by up to 2.47dB) and on our REBlur dataset, even in extreme blurry conditions. The code and our REBlur dataset are available at https://ahupujr.github.io/EFNet/"



Paperid:743
Authors:Yifan Jiang; Bartlomiej Wronski; Ben Mildenhall; Jonathan T. Barron; Zhangyang Wang; Tianfan Xue
Title: Fast and High Quality Image Denoising via Malleable Convolution
Abstract:
Most image denoising networks apply a single set of static convolutional kernels across the entire input image. This is sub-optimal for natural images, as they often consist of heterogeneous visual patterns. Dynamic convolution tries to address this issue by using per-pixel convolution kernels, but this greatly increases computational cost. In this work, we present \textbf{Malle}able \textbf{Conv}olution (\textbf{MalleConv}), which performs spatial-varying processing with minimal computational overhead. MalleConv uses a smaller set of spatially-varying convolution kernels, a compromise between static and per-pixel convolution kernels. These spatially-varying kernels are produced by an efficient predictor network running on a downsampled input, making them much more efficient to compute than per-pixel kernels produced by a full-resolution image, and also enlarging the network’s receptive field compared with static kernels. These kernels are then jointly upsampled and applied to a full-resolution feature map through an efficient on-the-fly slicing operator with minimum memory overhead. To demonstrate the effectiveness of MalleConv, we use it to build an efficient denoising network we call \textbf{MalleNet}. MalleNet achieves high quality results without a very deep architecture, e.g., running 8.9$\times$ faster than the best performing denoising algorithms (SwinIR) while maintaining similar quality. We also show that a single MalleConv layer added to a standard convolution-based backbone can contribute significantly to reducing the computational cost or can boost image quality at a similar cost.



Paperid:744
Authors:Lin Liu; Lingxi Xie; Xiaopeng Zhang; Shanxin Yuan; Xiangyu Chen; Wengang Zhou; Houqiang Li; Qi Tian
Title: TAPE: Task-Agnostic Prior Embedding for Image Restoration
Abstract:
Learning a generalized prior for natural image restoration is an important yet challenging task. Early methods mostly involved handcrafted priors including normalized sparsity, â„“0 gradients, dark channel priors, etc.. Recently, deep neural networks have been used to learn various image priors but do not guarantee to generalize. In this paper, we propose a novel approach that embeds a task-agnostic prior into a transformer. Our approach, named Task-Agnostic Prior Embedding (TAPE), consists of two stages, namely, task-agnostic pre-training and task-specific fine-tuning, where the first stage embeds prior knowledge about natural images into the transformer and the second stage extracts the knowledge to assist downstream image restoration. Experiments on various types of degradation validate the effectiveness of TAPE. The image restoration performance in terms of PSNR is improved by as much as 1.45 dB and even outperforms task-specific algorithms. More importantly, TAPE shows the ability of disentangling generalized image priors from degraded images, which enjoys favorable transfer ability to unknown downstream tasks.



Paperid:745
Authors:Zhenqi Fu; Wu Wang; Yue Huang; Xinghao Ding; Kai-Kuang Ma
Title: Uncertainty Inspired Underwater Image Enhancement
Abstract:
A main challenge faced in the deep learning-based Underwater Image Enhancement (UIE) is that the ground truth high-quality image is unavailable. Most of the existing methods first generate approximate reference maps and then train an enhancement network with certainty. This kind of method fails to handle the ambiguity of the reference map. In this paper, we resolve UIE into distribution estimation and consensus process. We present a novel probabilistic network to learn the enhancement distribution of degraded underwater images. Specifically, we combine conditional variational autoencoder with adaptive instance normalization to construct the enhancement distribution. After that, we adopt a consensus process to predict a deterministic result based on a set of samples from the distribution. By learning the enhancement distribution, our method can cope with the bias introduced in the reference map labeling to some extent. Additionally, the consensus process is useful to capture a robust and stable result. We examined the proposed method on two widely used real-world underwater image enhancement datasets. Experimental results demonstrate that our approach enables sampling possible enhancement predictions. Meanwhile, the consensus estimate yields competitive performance compared with state-of-the-art UIE methods.



Paperid:746
Authors:Ye Deng; Siqi Hui; Rongye Meng; Sanping Zhou; Jinjun Wang
Title: Hourglass Attention Network for Image Inpainting
Abstract:
Benefiting from the powerful ability of convolutional neural networks (CNNs) to learn semantic information and texture patterns of images, learning-based image inpainting methods have made noticeable breakthroughs over the years. However, certain inherent defects (e.g. local prior, spatially sharing parameters) of CNNs limit their performance when encountering broken images mixed with invalid information. Compared to convolution, attention has a lower inductive bias, and the output is highly correlated with the input, making it more suitable for processing images with various breakage. Inspired by this, in this paper we propose a novel attention-based network (transformer), called hourglass attention network (HAN) for image inpainting, which builds an hourglass-shaped attention structure to generate appropriate features for complemented images. In addition, we design a novel attention called Laplace attention, which introduces a Laplace distance prior for the vanilla multi-head attention, allowing the feature matching process to consider not only the similarity of features themselves, but also distance between features. With the synergy of hourglass attention structure and Laplace attention, our HAN is able to make full use of hierarchical features to mine effective information for broken images. Experiments on several benchmark datasets demonstrate superior performance by our proposed approach.



Paperid:747
Authors:Hongyi Zheng; Hongwei Yong; Lei Zhang
Title: Unfolded Deep Kernel Estimation for Blind Image Super-Resolution
Abstract:
Blind image super-resolution (BISR) aims to reconstruct a high-resolution image from its low-resolution counterpart degraded by unknown blur kernel and noise. Many deep neural network based methods have been proposed to tackle this challenging problem without considering the image degradation model. However, they largely rely on the training sets and often fail to handle images with unseen blur kernels during inference. Deep unfolding methods have also been proposed to perform BISR by utilizing the degradation model. Nonetheless, the existing deep unfolding methods cannot explicitly solve the data term of the unfolding objective function, limiting their capability in blur kernel estimation. In this work, we propose a novel unfolded deep kernel estimation (UDKE) method, which, for the first time to our best knowledge, explicitly solves the data term with high efficiency. The UDKE based BISR method can jointly learn image and kernel priors in an end-to-end manner, and it can effectively exploit the information in both training data and image degradation model. Experiments on benchmark datasets and real-world data demonstrate that the proposed UDKE method could well predict complex unseen non-Gaussian blur kernels in inference, achieving significantly better BISR performance than state-of-the-art. The source code of UDKE is available at https://github.com/natezhenghy/UDKE.



Paperid:748
Authors:Taewoo Kim; Jeongmin Lee; Lin Wang; Kuk-Jin Yoon
Title: Event-Guided Deblurring of Unknown Exposure Time Videos
Abstract:
Motion deblurring is a highly ill-posed problem due to the loss of motion information in the blur degradation process. Since event cameras can capture apparent motion with a high temporal resolution, several attempts have explored the potential of events for guiding deblurring. These methods generally assume that the exposure time is the same as the reciprocal of the video frame rate. However, this is not true in real situations, and the exposure time might be unknown and dynamically varies depending on the video shooting environment(e.g., illumination condition). In this paper, we address the event-guided motion deblurring assuming dynamically variable unknown exposure time of the frame-based camera. To this end, we first derive a new formulation for event-guided motion deblurring by considering the exposure and readout time in the video frame acquisition process. We then propose a novel end-to-end learning framework for event-guided motion deblurring. In particular, we design a novel Exposure Time-based Event Selection(ETES) module to selectively use event features by estimating the cross-modal correlation between the features from blurred frames and the events. Moreover, we propose a feature fusion module to fuse the selected features from events and blur frames effectively. We conduct extensive experiments on various datasets and demonstrate that our method achieves state-of-the-art performance.



Paperid:749
Authors:Zhanbo Huang; Jinyuan Liu; Xin Fan; Risheng Liu; Wei Zhong; Zhongxuan Luo
Title: ReCoNet: Recurrent Correction Network for Fast and Efficient Multi-Modality Image Fusion
Abstract:
Recent advances in deep networks have gained great attention in infrared and visible image fusion (IVIF). Nevertheless, most existing methods are incapable of dealing with slight misalignment on source images and suffer from high computational and spatial expenses. This paper tackles these two critical issues rarely touched in the community by developing a recurrent correction network for robust and efficient fusion, namely ReCoNet. Concretely, we design a deformation module to explicitly compensate geometrical distortions and an attention mechanism to mitigate ghosting-like artifacts, respectively. Meanwhile, the network consists of a parallel dilated convolutional layer and runs in a recurrent fashion, significantly reducing both spatial and computational complexities. ReCoNet can effectively and efficiently alleviates both structural distortions and textural artifacts brought by slight misalignment. Extensive experiments on two public datasets demonstrate the superior accuracy and efficacy of our ReCoNet against the state-of-the-art IVIF methods. Consequently, we obtain a 16% relative improvement of CC on datasets with misalignment and boost the efficiency by 86%. The source code is available at https://github.com/dlut-dimt/reconet.



Paperid:750
Authors:Guanbo Pan; Guo Lu; Zhihao Hu; Dong Xu
Title: Content Adaptive Latents and Decoder for Neural Image Compression
Abstract:
In recent years, neural image compression (NIC) algorithms have shown powerful coding performance. However, most of them are not adaptive to the image content. Although several content adaptive methods have been proposed by updating the encoder-side components, the adaptability of both latents and the decoder is not well exploited. In this work, we propose a new NIC framework that improves the content adaptability on both latents and the decoder. Specifically, to remove redundancy in the latents, our content adaptive channel dropping (CACD) method automatically selects the optimal quality levels for the latents spatially and drops the redundant channels. Additionally, we propose the content adaptive feature transformation (CAFT) method to improve decoder-side content adaptability by extracting the characteristic information of the image content, which is then used to transform the features in the decoder side. Experimental results demonstrate that our proposed methods with the encoder-side updating algorithm achieve the state-of-the-art performance.



Paperid:751
Authors:Jie Liang; Hui Zeng; Lei Zhang
Title: Efficient and Degradation-Adaptive Network for Real-World Image Super-Resolution
Abstract:
Efficient and effective real-world image super-resolution (Real-ISR) is a challenging task due to the unknown complex degradation of real-world images and the limited computation resources in practical applications. Recent research on Real-ISR has achieved significant progress by modeling the image degradation space; however, these methods largely rely on heavy backbone networks and they are inflexible to handle images of different degradation levels. In this paper, we propose an efficient and effective degradation-adaptive super-resolution (DASR) network, whose parameters are adaptively specified by estimating the degradation of each input image. Specifically, a tiny regression network is employed to predict the degradation parameters of the input image, while several convolutional experts with the same topology are jointly optimized to specify the network parameters via a non-linear mixture of experts. The joint optimization of multiple experts and the degradation-adaptive pipeline significantly extend the model capacity to handle degradations of various levels, while the inference remains efficient since only one adaptively specified network is used for super-resolving the input image. Our extensive experiments demonstrate that DASR is not only much more effective than existing methods on handling real-world images with different degradation levels but also efficient for easy deployment. Codes, models and datasets are available at https://github.com/csjliang/DASR.



Paperid:752
Authors:Junyi Li; Xiaohe Wu; Zhenxing Niu; Wangmeng Zuo
Title: Unidirectional Video Denoising by Mimicking Backward Recurrent Modules with Look-Ahead Forward Ones
Abstract:
While significant progress has been made in deep video denoising, it remains very challenging for exploiting historical and future frames. Bidirectional recurrent networks (BiRNN) have exhibited appealing performance in several video restoration tasks. However, BiRNN is intrinsically offline because it uses backward recurrent modules to propagate from the last to current frames, which causes high latency and large memory consumption. To address the offline issue of BiRNN, we present a novel recurrent network consisting of forward and look-ahead recurrent modules for unidirectional video denoising. Particularly, look-ahead module is an elaborate forward module for leveraging information from near-future frames. When denoising the current frame, the hidden features by forward and look-ahead recurrent modules are combined, thereby making it feasible to exploit both historical and near-future frames. Due to the scene motion between non-neighboring frames, border pixels missing may occur when warping look-ahead feature from near-future frame to current frame, which can be largely alleviated by incorporating forward warping and proposed border enlargement. Experiments show that our method achieves state-of-the-art performance with constant latency and memory consumption. Code is avaliable at https://github.com/nagejacob/FloRNN.



Paperid:753
Authors:Zhilu Zhang; Ruohao Wang; Hongzhi Zhang; Yunjin Chen; Wangmeng Zuo
Title: Self-Supervised Learning for Real-World Super-Resolution from Dual Zoomed Observations
Abstract:
In this paper, we consider two challenging issues in reference-based super-resolution (RefSR), (i) how to choose a proper reference image, and (ii) how to learn real-world RefSR in a self-supervised manner. Particularly, we present a novel self-supervised learning approach for real-world image SR from observations at dual camera zooms (SelfDZSR). Considering the popularity of multiple cameras in modern smartphones, the more zoomed (telephoto) image can be naturally leveraged as the reference to guide the SR of the lesser zoomed (short-focus) image. Furthermore, SelfDZSR learns a deep network to obtain the SR result of short-focus image to have the same resolution as the telephoto image. For this purpose, we take the telephoto image instead of an additional high-resolution image as the supervision information and select a center patch from it as the reference to super-resolve the corresponding short-focus image patch. To mitigate the effect of the misalignment between short-focus low-resolution (LR) image and telephoto ground-truth (GT) image, we design an auxiliary-LR generator and map the GT to an auxiliary-LR while keeping the spatial position unchanged. Then the auxiliary-LR can be utilized to deform the LR features by the proposed adaptive spatial transformer networks (AdaSTN), and match the Ref features to GT. During testing, SelfDZSR can be directly deployed to super-solve the whole short-focus image with the reference of telephoto image. Experiments show that our method achieves better quantitative and qualitative performance against state-of-the-arts. Codes are available at https://github.com/cszhilu1998/SelfDZSR.



Paperid:754
Authors:Shintaro Shiba; Yoshimitsu Aoki; Guillermo Gallego
Title: Secrets of Event-Based Optical Flow
Abstract:
Event cameras respond to scene dynamics and offer advantages to estimate motion. Following recent image-based deep-learning achievements, optical flow estimation methods for event cameras have rushed to combine those image-based methods with event data. However, it requires several adaptations (data conversion, loss function, etc.) as they have very different properties. We develop a principled method to extend the Contrast Maximization framework to estimate optical flow from events alone. We investigate key elements: how to design the objective function to prevent overfitting, how to warp events to deal better with occlusions, and how to improve convergence with multi-scale raw events. With these key elements, our method ranks first among unsupervised methods on the MVSEC benchmark, and is competitive on the DSEC benchmark. Moreover, our method allows us to expose the issues of the ground truth flow in those benchmarks, and produces remarkable results when it is transferred to unsupervised learning settings. Our code is available at https://github.com/tub-rip/event_based_optical_flow"



Paperid:755
Authors:Xin Yu; Peng Dai; Wenbo Li; Lan Ma; Jiajun Shen; Jia Li; Xiaojuan Qi
Title: Towards Efficient and Scale-Robust Ultra-High-Definition Image Demoiréing
Abstract:
With the rapid development of mobile devices, modern widely-used mobile phones typically allow users to capture 4K resolution (i.e., ultra-high-definition) images. However, for image demoiréing, a challenging task in low-level vision, existing works are generally carried out on low-resolution or synthetic images. Hence, the effectiveness of these methods on 4K resolution images is still unknown. In this paper, we explore moiré pattern removal for ultra-high-definition images. To this end, we propose the first ultra-high-definition demoiréing dataset (UHDM), which contains 5,000 real-world 4K resolution image pairs, and conduct a benchmark study on current state-of-the-art methods. Further, we present an efficient baseline model ESDNet for tackling 4K moiré images, wherein we build a semantic-aligned scale-aware module to address the scale variation of moiré patterns. Extensive experiments manifest the effectiveness of our approach, which outperforms state-of-the-art methods by a large margin while being much more lightweight. Code and dataset are available at https://xinyu-andy.github.io/uhdm-page.



Paperid:756
Authors:Bangrui Jiang; Zhihuai Xie; Zhen Xia; Songnan Li; Shan Liu
Title: ERDN: Equivalent Receptive Field Deformable Network for Video Deblurring
Abstract:
Video deblurring aims to restore sharp frames from blurry video sequences. Existing methods usually adopt optical flow to compensate misalignment between reference frame and each neighboring frame. However, inaccurate flow estimation caused by large displacements will lead to artifacts in the warped frames. In this work, we propose an equivalent receptive field deformable network (ERDN) to perform alignment at the feature level without estimating optical flow. The ERDN introduces a dual pyramid alignment module, in which a feature pyramid is constructed to align frames using deformable convolution in a cascaded manner. Specifically, we adopt dilated spatial pyramid blocks to predict offsets for deformable convolutions, so that the theoretical receptive field is equivalent for each feature pyramid layer. To restore the sharp frame, we propose a gradient guided fusion module, which incorporates structure priors into the restoration process. Experimental results demonstrate that the proposed method outperforms previous state-of-the-art methods on multiple benchmark datasets. The code is made available at: https://github.com/TencentCloud/ERDN.



Paperid:757
Authors:Nobuhiko Wakai; Satoshi Sato; Yasunori Ishii; Takayoshi Yamashita
Title: Rethinking Generic Camera Models for Deep Single Image Camera Calibration to Recover Rotation and Fisheye Distortion
Abstract:
Although recent learning-based calibration methods can predict extrinsic and intrinsic camera parameters from a single image, the accuracy of these methods is degraded in fisheye images. This degradation is caused by mismatching between the actual projection and expected projection. To address this problem, we propose a generic camera model that has the potential to address various types of distortion. Our generic camera model is utilized for learning-based methods through a closed-form numerical calculation of the camera projection. Simultaneously to recover rotation and fisheye distortion, we propose a learning-based calibration method that uses the camera model. Furthermore, we propose a loss function that alleviates the bias of the magnitude of errors for four extrinsic and intrinsic camera parameters. Extensive experiments demonstrated that our proposed method outperformed conventional methods on two large-scale datasets and images captured by off-the-shelf fisheye cameras. Moreover, we are the first researchers to analyze the performance of learning-based methods using various types of projection for off-the-shelf cameras.



Paperid:758
Authors:Rajeev Yasarla; Carey E. Priebe; Vishal M. Patel
Title: ART-SS: An Adaptive Rejection Technique for Semi-Supervised Restoration for Adverse Weather-Affected Images
Abstract:
In recent years, convolutional neural network-based single image adverse weather removal methods have achieved significant performance improvements on many benchmark datasets. However, these methods require large amounts of clean-weather degraded image pairs for training, which is often difficult to obtain in practice. Although various weather degradation synthesis methods exist in the literature, the use of synthetically generated weather degraded images often results in sub-optimal performance on the real weatherdegraded images due to the domain gap between synthetic and real world images. To deal with this problem, various semi-supervised restoration (SSR) methods have been proposed for deraining or dehazing which learn to restore clean image using synthetically generated datasets while generalizing better using unlabeled real-world images. The performance of a semi-supervised method is essentially based on the quality of the unlabeled data. In particular, if the unlabeled data characteristics are very different from that of the labeled data, then the performance of a semi-supervised method degrades significantly. We theoretically study the effect of unlabeled data on the performance of an SSR method and develop a technique that rejects the unlabeled images that degrade the performance. Extensive experiments and ablation study show that the proposed sample rejection method increases the performance of existing SSR deraining and dehazing methods significantly. Code is available at :\textit{\href{https://github.com/rajeevyasarla/ART-SS}{https://github.com/rajeevyasarla/ART-SS}}"



Paperid:759
Authors:Pengwei Liang; Junjun Jiang; Xianming Liu; Jiayi Ma
Title: Fusion from Decomposition: A Self-Supervised Decomposition Approach for Image Fusion
Abstract:
Image fusion is famous as an alternative solution to generate one high-quality image from multiple images in addition to image restoration from a single degraded image. The essence of image fusion is to integrate complementary information or best parts from source images. The current fusion methods usually need a large number of paired samples or sophisticated loss functions and fusion rules to train the supervised or unsupervised model. In this paper, we propose a powerful image decomposition model for fusion task via the self-supervised representation learning, dubbed Decomposition for Fusion (DeFusion). Without any paired data or sophisticated loss, DeFusion can decompose the source images into a feature embedding space, where the common and unique features can be separated. Therefore, the image fusion can be achieved within the embedding space through the jointly trained reconstruction (projection) head in the decomposition stage even without any fine-tuning. Thanks to the development of self-supervised learning, we can train the model to learn image decomposition ability with a brute but simple pretext task. The pretrained model allows for learning very effective features that generalize well: the DeFusion is a unified versatile framework that is trained with an image fusion irrelevant dataset and can be directly applied to various image fusion tasks. Extensive experiments demonstrate that the proposed DeFusion can achieve comparable or even better performance compared to state-of-the-art methods (whether supervised or unsupervised) for different image fusion tasks.



Paperid:760
Authors:Dasong Li; Yi Zhang; Ka Chun Cheung; Xiaogang Wang; Hongwei Qin; Hongsheng Li
Title: Learning Degradation Representations for Image Deblurring
Abstract:
In various learning-based image restoration tasks, such as image denoising and image super-resolution, the degradation representations were widely used to model the degradation process and handle complicated degradation patterns. However, they are less explored in learning-based image deblurring as blur kernel estimation cannot perform well in real-world challenging cases. We argue that it is particularly necessary for image deblurring to model degradation representations since blurry patterns typically show much larger variations than noisy patterns or high-frequency textures. In this paper, we propose a framework to learn spatially adaptive degradation representations of blurry images. A novel joint image reblurring and deblurring learning process is presented to improve the expressiveness of degradation representations. To make learned degradation representations effective in reblurring and deblurring, we propose a Multi-Scale Degradation Injection Network (MSDI-Net) to integrate them into the neural networks. With the integration, MSDI-Net can handle various and complicated blurry patterns adaptively. Experiments on the GoPro and RealBlur datasets demonstrate that our proposed deblurring framework with the learned degradation representations outperforms state-of-the-art methods with appealing improvements. The code is released at https://github.com/dasongli1/Learning_degradation.



Paperid:761
Authors:Xiaoyu Dong; Naoto Yokoya; Longguang Wang; Tatsumi Uezato
Title: Learning Mutual Modulation for Self-Supervised Cross-Modal Super-Resolution
Abstract:
Self-supervised cross-modal super-resolution (SR) can overcome the difficulty of acquiring paired training data, but is challenging because only low-resolution (LR) source and high-resolution (HR) guide images from different modalities are available. Existing methods utilize pseudo or weak supervision in LR space and thus deliver results that are blurry or not faithful to the source modality. To address this issue, we present a mutual modulation SR (MMSR) model, which tackles the task by a mutual modulation strategy, including a source-to-guide modulation and a guide-to-source modulation. In these modulations, we develop cross-domain adaptive filters to fully exploit cross-modal spatial dependency and help induce the source to emulate the resolution of the guide and induce the guide to mimic the modality characteristics of the source. Moreover, we adopt a cycle consistency constraint to train MMSR in a fully self-supervised manner. Experiments on various tasks demonstrate the state-of-the-art performance of our MMSR.



Paperid:762
Authors:Wei He; Quanming Yao; Naoto Yokoya; Tatsumi Uezato; Hongyan Zhang; Liangpei Zhang
Title: Spectrum-Aware and Transferable Architecture Search for Hyperspectral Image Restoration
Abstract:
Convolutional neural networks have been widely developed for hyperspectral image (HSI) restoration. However, making full use of the spatial-spectral information of HSIs still remains a challenge. In this work, we disentangle the 3D convolution into lightweight 2D spatial and spectral convolutions, and build a spectrum-aware search space for HSI restoration. Subsequently, we utilize neural architecture search strategy to automatically learn the most efficient architecture with proper convolutions and connections in order to fully exploit the spatial-spectral information. We also determine that the super-net with global and local skip connections can further boost HSI restoration performance. The searched architecture on the CAVE dataset has been adopted for various dataset denoising and imaging reconstruction tasks, and achieves remarkable performance. On the basis of fruitful experiments, we conclude that the transferability of searched architecture is dependent on the spectral information and independent of the noise levels.



Paperid:763
Authors:Yili Wang; Xin Li; Kun Xu; Dongliang He; Qi Zhang; Fu Li; Errui Ding
Title: Neural Color Operators for Sequential Image Retouching
Abstract:
We propose a novel image retouching method by modeling the retouching process as performing a sequence of newly introduced trainable neural color operators. The neural color operator mimics the behavior of traditional color operators and learns pixelwise color transformation while its strength is controlled by a scalar. To reflect the homomorphism property of color operators, we employ equivariant mapping and adopt an encoder-decoder structure which maps the non-linear color transformation to a much simpler transformation (i.e., translation) in a high dimensional space. The scalar strength of each neural color operator is predicted using CNN based strength predictors by analyzing global image statistics. Overall, our method is rather lightweight and offers flexible controls. Experiments and user studies on public datasets show that our method consistently achieves the best results compared with SOTA methods in both quantitative measures and visual qualities. The code and data will be made publicly available.



Paperid:764
Authors:Ka Leong Cheng; Yueqi Xie; Qifeng Chen
Title: Optimizing Image Compression via Joint Learning with Denoising
Abstract:
High levels of noise usually exist in today’s captured images due to the relatively small sensors equipped in the smartphone cameras, where the noise brings extra challenges to lossy image compression algorithms. Without the capacity to tell the difference between image details and noise, general image compression methods allocate additional bits to explicitly store the undesired image noise during compression and restore the unpleasant noisy image during decompression. Based on the observations, we optimize the image compression algorithm to be noise-aware as joint denoising and compression to resolve the bits misallocation problem. The key is to transform the original noisy images to noise-free bits by eliminating the undesired noise during compression, where the bits are later decompressed as clean images. Specifically, we propose a novel two-branch, weight-sharing architecture with plug-in feature denoisers to allow a simple and effective realization of the goal with little computational cost. Experimental results show that our method gains a significant improvement over the existing baseline methods on both the synthetic and real-world datasets. Our source code is available at: https://github.com/felixcheng97/DenoiseCompression.



Paperid:765
Authors:Xiaotao Hu; Jun Xu; Shuhang Gu; Ming-Ming Cheng; Li Liu
Title: "Restore Globally, Refine Locally: A Mask-Guided Scheme to Accelerate Super-Resolution Networks"
Abstract:
Single image super-resolution (SR) has been boosted by deep convolutional neural networks with growing model complexity and computational costs. To deploy existing SR networks onto edge devices, it is necessary to accelerate them for large image (4K) processing. The different areas in an image often require different SR intensities by networks with different complexity. Motivated by this, in this paper, we propose a Mask Guided Acceleration (MGA) scheme to reduce the computational costs of existing SR networks while maintaining their SR capability. In our MGA scheme, we first decompose a given SR network into a Base-Net and a Refine-Net. The Base-Net is to extract a coarse feature and obtain a coarse SR image. To locate the under-SR areas in the coarse SR image, we then propose a Mask Prediction (MP) module to generate an error mask from the coarse feature. According to the error mask, we select K feature patches from the coarse feature and refine them (instead of the whole feature) by Refine-Net to output the final SR image. Experiments on seven benchmarks demonstrate that our MGA scheme reduces the FLOPs of five popular SR networks by 10% ~ 48% with comparable or even better SR performance. The code will be publicly released.



Paperid:766
Authors:Yushu Wu; Yifan Gong; Pu Zhao; Yanyu Li; Zheng Zhan; Wei Niu; Hao Tang; Minghai Qin; Bin Ren; Yanzhi Wang
Title: Compiler-Aware Neural Architecture Search for On-Mobile Real-Time Super-Resolution
Abstract:
Deep learning-based super-resolution (SR) has gained tremendous popularity in recent years because of its high image quality performance and wide application scenarios. However, prior methods typically suffer from large amounts of computations and huge power consumption, causing difficulties for real-time inference, especially on resourcelimited platforms such as mobile devices. To mitigate this, we propose a compiler-aware SR neural architecture search (NAS) framework that conducts depth search and per-layer width search with adaptive SR blocks. The inference speed is directly taken into the optimization along with the SR loss to derive SR models with high image quality while satisfying the real-time inference requirement. Instead of measuring the speed on mobile devices at each iteration during the search process, a speed model incorporated with compiler optimizations is leveraged to predict the inference latency of the SR block with various width configurations for faster convergence. With the proposed framework, we achieve realtime SR inference for implementing 720p resolution with competitive SR performance (in terms of PSNR and SSIM) on GPU/DSP of mobile platforms (Samsung Galaxy S21).



Paperid:767
Authors:Jiamian Wang; Yulun Zhang; Xin Yuan; Ziyi Meng; Zhiqiang Tao
Title: Modeling Mask Uncertainty in Hyperspectral Image Reconstruction
Abstract:
Recently, hyperspectral imaging (HSI) has attracted increasing research attention, especially for the ones based on a coded aperture snapshot spectral imaging (CASSI) system. Existing deep HSI reconstruction models are generally trained on paired data to retrieve original signals upon 2D compressed measurements given by a particular optical hardware mask in CASSI, during which the mask largely impacts the reconstruction performance and could work as a ""model hyperparameter"" governing on data augmentations. This mask-specific training style will lead to a hardware miscalibration issue, which sets up barriers to deploying deep HSI models among different hardware and noisy environments. To address this challenge, we introduce mask uncertainty for HSI with a complete variational Bayesian learning treatment and explicitly model it through a mask decomposition inspired by real hardware. Specifically, we propose a novel Graph-based Self-Tuning (GST) network to reason uncertainties adapting to varying spatial structures of masks among different hardware. Moreover, we develop a bilevel optimization framework to balance HSI reconstruction and uncertainty estimation, accounting for the hyperparameter property of masks. Extensive experimental results validate the effectiveness (over 33/30 dB) of the proposed method under two miscalibration scenarios and demonstrate a highly competitive performance compared with the state-of-the-art well-calibrated methods. Our source code and pre-trained models are available at https://github.com/Jiamian-Wang/mask_uncertainty_spectral_SCI"



Paperid:768
Authors:Tian Ye; Yunchen Zhang; Mingchao Jiang; Liang Chen; Yun Liu; Sixiang Chen; Erkang Chen
Title: Perceiving and Modeling Density for Image Dehazing
Abstract:
In the real world, the degradation of images taken under haze can be quite complex, where the spatial distribution of haze varies from image to image. Recent methods adopt deep neural networks to recover clean scenes from hazy images directly. However, due to the generic design of network architectures and the failure in estimating an accurate haze degradation model, the generalization ability of recent dehazing methods on real-world hazy images is not ideal. To address the problem of modeling real-world haze degradation, we propose a novel Separable Hybrid Attention (SHA) module to perceive haze density by capturing positional-sensitive features in the orthogonal directions to achieve this goal. Moreover, a density encoding matrix is proposed to model the uneven distribution of the haze explicitly. The density encoding matrix generates positional encoding in a semi-supervised way -- such a haze density perceiving and modeling strategy captures the unevenly distributed degeneration at the feature-level effectively. Through a suitable combination of SHA and density encoding matrix, we design a novel dehazing network architecture, which achieves a good complexity-performance trade-off. Comprehensive evaluation on both synthetic datasets and real-world datasets demonstrates that the proposed method surpasses all the state-of-the-art approaches with a large margin both quantitatively and qualitatively. The code is released in https://github.com/Owen718/ECCV22-Perceiving-and-Modeling-Density-for-Image-Dehazing.



Paperid:769
Authors:Fu-Jen Tsai; Yan-Tsung Peng; Yen-Yu Lin; Chung-Chi Tsai; Chia-Wen Lin
Title: Stripformer: Strip Transformer for Fast Image Deblurring
Abstract:
Images taken in dynamic scenes may contain unwanted motion blur, which significantly degrades visual quality. Such blur causes short- and long-range region-specific smoothing artifacts that are often directional and non-uniform, which is difficult to be removed. Inspired by the current success of transformers on computer vision and image processing tasks, we develop, Stripformer, a transformer-based architecture that constructs intra- and inter-strip tokens to reweight image features in the horizontal and vertical directions to catch blurred patterns with different orientations. It stacks interlaced intra-strip and inter-strip attention layers to reveal blur magnitudes. In addition to detecting region-specific blurred patterns of various orientations and magnitudes, Stripformer is also a token-efficient and parameter-efficient transformer model, demanding much less memory usage and computation cost than the vanilla transformer but works better without relying on tremendous training data. Experimental results show that Stripformer performs favorably against state-of-the-art models in dynamic scene deblurring.



Paperid:770
Authors:Jie Huang; Yajing Liu; Feng Zhao; Keyu Yan; Jinghao Zhang; Yukun Huang; Man Zhou; Zhiwei Xiong
Title: Deep Fourier-Based Exposure Correction Network with Spatial-Frequency Interaction
Abstract:
Images captured under incorrect exposures unavoidably suffer from mixed degradations of lightness and structures. Most existing deep learning-based exposure correction methods separately restore such degradations in the spatial domain. In this paper, we present a new perspective for exposure correction with spatial-frequency interaction. Specifically, we first revisit the frequency properties of different exposure images via Fourier transform where the amplitude component contains most lightness information and the phase component is relevant to structure information. To this end, we propose a deep Fourier-based Exposure Correction Network (FECNet) consisting of an amplitude and a phase sub-networks to progressively reconstruct the representation of lightness and structure components. To facilitate learning these two representations, we introduce a Spatial-Frequency Interaction (SFI) block in two formats tailored to these two sub-networks, which interactively process the local spatial features and the global frequency information to encourage the complementary learning. Extensive experiments demonstrate that our FECNet achieves superior results than other methods with fewer parameters and can be extended to other image enhancement tasks, validating its potential for wide-range applications.



Paperid:771
Authors:Hu Yu; Naishan Zheng; Man Zhou; Jie Huang; Zeyu Xiao; Feng Zhao
Title: Frequency and Spatial Dual Guidance for Image Dehazing
Abstract:
In this paper, we propose a novel image dehazing framework with frequency and spatial dual guidance. In contrast to most existing deep learning-based image dehazing methods that primarily exploit spatial information and neglect the distinguished frequency information, we introduce a new perspective to address image dehazing by jointly exploring the information in the frequency and spatial domain. To implement frequency and spatial dual guidance, we delicately develop two core designs: Amplitude guided phase module in the frequency domain and Global guided local module in the spatial domain. Specifically, the former processes the global frequency information via deep Fourier transform and reconstructs the phase spectrum under the guidance of the amplitude spectrum while the latter integrates the above global frequency information to facilitate the local features learning in the spatial domain. Extensive experiments on synthetic and real-world datasets demonstrate that our method outperforms the state-of-the-art approaches visually and quantitatively.



Paperid:772
Authors:Zhen Cheng; Tao Wang; Yong Li; Fenglong Song; Chang Chen; Zhiwei Xiong
Title: Towards Real-World HDRTV Reconstruction: A Data Synthesis-Based Approach
Abstract:
Existing deep learning based HDRTV reconstruction methods assume one kind of tone mapping operators (TMOs) as the degradation procedure to synthesize SDRTV-HDRTV pairs for supervised training. In this paper, we argue that, although traditional TMOs exploit efficient dynamic range compression priors, they have several drawbacks on modeling the realistic degradation: information over-preservation, color bias and possible artifacts, making the trained reconstruction networks hard to generalize well to real-world cases. To solve this problem, we propose a learning-based data synthesis approach to learn the properties of real-world SDRTVs by integrating several tone mapping priors into both network structures and loss functions. In specific, we design a conditioned two-stream network with prior tone mapping results as a guidance to synthesize SDRTVs by both global and local transformations. To train the data synthesis network, we form a novel self-supervised content loss to constraint different aspects of the synthesized SDRTVs at regions with different brightness distributions and an adversarial loss to emphasize the details to be more realistic. To validate the effectiveness of our approach, we synthesize SDRTV-HDRTV pairs with our method and use them to train several HDRTV reconstruction networks. Then we collect two inference datasets containing both labeled and unlabeled real-world SDRTVs, respectively. Experimental results demonstrate that, the networks trained with our synthesized data generalize significantly better to these two real-world datasets than existing solutions.



Paperid:773
Authors:Pin-Hung Kuo; Jinshan Pan; Shao-Yi Chien; Ming-Hsuan Yang
Title: Learning Discriminative Shrinkage Deep Networks for Image Deconvolution
Abstract:
Most existing methods usually formulate the non-blind deconvolution problem into a maximum-a-posteriori framework and address it by manually designing a variety of regularization terms and data terms of the latent clear images. However, explicitly designing these two terms is quite challenging and usually leads to complex optimization problems which are difficult to solve. This paper proposes an effective non-blind deconvolution approach by learning discriminative shrinkage functions to model these terms implicitly. Most existing methods use deep convolutional neural networks (CNNs) or radial basis functions to learn the regularization term simply. In contrast, we formulate both the data term and regularization term and split the deconvolution model into data-related and regularization-related sub-problems according to the alternating direction method of multipliers. We explore the properties of the Maxout function and develop a deep CNN model with Maxout layers to learn discriminative shrinkage functions, which directly approximates the solutions of these two sub-problems. Moreover, the fast-Fourier-transform-based image restoration usually leads to ringing artifacts. At the same time, the conjugate-gradient-based approach is time-consuming; we develop the Conjugate Gradient Network to restore the latent clear images effectively and efficiently. Experimental results show that the proposed method performs favorably against the state-of-the-art methods in terms of efficiency and accuracy.



Paperid:774
Authors:Jiahong Fu; Hong Wang; Qi Xie; Qian Zhao; Deyu Meng; Zongben Xu
Title: KXNet: A Model-Driven Deep Neural Network for Blind Super-Resolution
Abstract:
Although current deep learning-based methods have gained promising performance in the blind single image super-resolution (SISR) task, most of them mainly focus on heuristically constructing diverse network architectures and put less emphasis on the explicit embedding of the physical generation mechanism between blur kernels and high-resolution (HR) images. To alleviate this issue, we propose a model-driven deep neural network, called KXNet, for blind SISR. Specifically, to solve the classical SISR model, we propose a simple-yet-effective iterative algorithm. Then by unfolding the involved iterative steps into the corresponding network module, we naturally construct the KXNet. The main specificity of the proposed KXNet is that the entire learning process is fully and explicitly integrated with the inherent physical mechanism underlying this SISR task. Thus, the learned blur kernel has clear physical patterns and the mutually iterative process between blur kernel and HR image can soundly guide the KXNet to be evolved in the right direction. Extensive experiments on synthetic and real data finely demonstrate the superior accuracy and generality of our method beyond the current representative state-of-the-art blind SISR methods. Code is available at: https://github.com/jiahong-fu/KXNet.



Paperid:775
Authors:Bohong Chen; Mingbao Lin; Kekai Sheng; Mengdan Zhang; Peixian Chen; Ke Li; Liujuan Cao; Rongrong Ji
Title: ARM: Any-Time Super-Resolution Method
Abstract:
This paper proposes an Any-time super-Resolution Method (ARM) to tackle the over-parameterized single image super-resolution (SISR) models. Our ARM is motivated by three observations: (1) The performance of different image patches varies with SISR networks of different sizes. (2) There is a tradeoff between computation overhead and performance of the reconstructed image. (3) Given an input image, its edge information can be an effective option to estimate its PSNR. Subsequently, we train an ARM supernet containing SISR subnets of different sizes to deal with image patches of various complexity. To that effect, we construct an Edge-to-PSNR lookup table that maps the edge score of an image patch to the PSNR performance for each subnet, together with a set of computation costs for the subnets. In the inference, the image patches are individually distributed to different subnets for a better computation-performance tradeoff. Moreover, each SISR subnet shares weights of the ARM supernet, thus no extra parameters are introduced. The setting of multiple subnets can well adapt the computational cost of SISR model to the dynamically available hardware resources, allowing the SISR task to be in service at any time. Extensive experiments on resolution datasets of different sizes with popular SISR networks as backbones verify the effectiveness and the versatility of our ARM.



Paperid:776
Authors:Haina Qin; Longfei Han; Juan Wang; Congxuan Zhang; Yanwei Li; Bing Li; Weiming Hu
Title: Attention-Aware Learning for Hyperparameter Prediction in Image Processing Pipelines
Abstract:
Between the imaging sensor and the image applications, the hardware image signal processing (ISP) pipelines reconstruct an RGB image from the sensor signal and feed it into downstream tasks. The processing blocks in ISPs depend on a set of tunable hyperparameters that have a complex interaction with the output. Manual setting by image experts is the traditional way of hyperparameter tuning, which is time-consuming and biased towards human perception. Recently, ISP has been optimized by the feedback of the downstream tasks based on different optimization algorithms. Unfortunately, these methods should keep parameters fixed during the inference stage for arbitrary input without considering that each image should have specific parameters based on its feature. To this end, we propose an attention-aware learning method that integrates the parameter prediction network into ISP tuning and utilizes the multi-attention mechanism to generate the attentive mapping between the input RAW image and the parameter space. The proposed method integrates downstream tasks end-to-end, predicting specific parameters for each image while considering feedback from downstream tasks. We validate the proposed method on object detection, image segmentation, and human viewing tasks. In these applications, our method outperforms the existing optimization methods and expert tuning.



Paperid:777
Authors:Yunhui Han; Kunming Luo; Ao Luo; Jiangyu Liu; Haoqiang Fan; Guiming Luo; Shuaicheng Liu
Title: RealFlow: EM-Based Realistic Optical Flow Dataset Generation from Videos
Abstract:
Obtaining the ground truth labels from a video is challenging since the manual annotation of pixel-wise flow labels is prohibitively expensive and laborious. Besides, existing approaches try to adapt the trained model on synthetic datasets to authentic videos, which inevitably suffers from domain discrepancy and hinders the performance for real-world applications. To solve these problems, we propose RealFlow, an Expectation-Maximization based framework that can create large-scale optical flow datasets directly from any unlabeled realistic videos. Specifically, we first estimate optical flow between a pair of video frames, and then synthesize a new image from this pair based on the predicted flow. Thus the new image pairs and their corresponding flows can be regarded as a new training set. Besides, we design a Realistic Image Pair Rendering (RIPR) module that adopts softmax splatting and bi-directional hole filling techniques to alleviate the artifacts of the image synthesis. In the E-step, RIPR renders new images to create a large quantity of training data. In the M-step, we utilize the generated training data to train an optical flow network, which can be used to estimate optical flows in the next E-step. During the iterative learning steps, the capability of the flow network is gradually improved, so is the accuracy of the flow, as well as the quality of the synthesized dataset. Experimental results show that RealFlow outperforms previous dataset generation methods by a considerably large margin. Moreover, based on the generated dataset, our approach achieves state-of-the-art performance on two standard benchmarks compared with both supervised and unsupervised optical flow methods. Our code and dataset are available at https://github.com/megvii-research/RealFlow"



Paperid:778
Authors:Keyu Yan; Man Zhou; Li Zhang; Chengjun Xie
Title: Memory-Augmented Model-Driven Network for Pansharpening
Abstract:
In this paper, we propose a novel memory-augmented model-driven deep unfolding network for pan-sharpening. First, we devise the maximal a posterior estimation (MAP) model with two well-designed priors on the latent multi-spectral (MS) image, i.e., global and local implicit priors to explore the intrinsic knowledge across the modalities of MS and panchromatic (PAN) images. Second, we design an effective alternating minimization algorithm to solve this MAP model, and then unfold the proposed algorithm into a deep network, where each stage corresponds to one iteration. Third, to facilitate the signal flow across adjacent iterations, the persistent memory mechanism is introduced to augment the information representation by exploiting the Long short-term memory unit in the image and feature spaces. With this method, both the interpretability and representation ability of the deep network are improved. Extensive experiments demonstrate the superiority of our method to the existing state-of-the-art approaches. The source code is released at https://github.com/Keyu-Yan/MMNet.



Paperid:779
Authors:Yuxuan Zhang; Bo Dong; Felix Heide
Title: All You Need Is RAW: Defending against Adversarial Attacks with Camera Image Pipelines
Abstract:
Existing neural networks for computer vision tasks are vulnerable to adversarial attacks: adding imperceptible perturbations to the input images can fool these models to make a false prediction on an image that was correctly predicted without the perturbation. Various defense methods have proposed image-to-image mapping methods, either including these perturbations in the training process or removing them in a preprocessing step. In doing so, existing methods often ignore that the natural RGB images in today’s datasets are not captured but, in fact, recovered from RAW color filter array captures that are subject to various degradations in the capture. In this work, we exploit this RAW data distribution as an empirical prior for adversarial defense. Specifically, we proposed a model-agnostic adversarial defensive method, which maps the input RGB images to Bayer RAW space and back to output RGB using a learned camera image signal processing (ISP) pipeline to eliminate potential adversarial patterns. The proposed method acts as an off-the-shelf preprocessing module and, unlike model-specific adversarial training methods, does not require adversarial images to train. As a result, the method generalizes to unseen tasks without additional retraining. Experiments on large-scale datasets (e.g., ImageNet, COCO) for different vision tasks (e.g., classification, semantic segmentation, object detection) validate that the method significantly outperforms existing methods across task domains.



Paperid:780
Authors:Zhen Liu; Yinglong Wang; Bing Zeng; Shuaicheng Liu
Title: Ghost-Free High Dynamic Range Imaging with Context-Aware Transformer
Abstract:
High dynamic range (HDR) deghosting algorithms aim to generate ghost-free HDR images with realistic details. Restricted by the locality of the receptive field, existing CNN-based methods are typically prone to producing ghosting artifacts and intensity distortions in the presence of large motion and severe saturation. In this paper, we propose a novel Context-aware Vision Transformer (CA-ViT) for ghost-free high dynamic range imaging. The CA-ViT is designed as a dual-branch architecture, which can jointly capture both global and local dependencies. Specifically, the global branch employs a window-based Transformer encoder to model long-range object movements and intensity variations to solve ghosting. For the local branch, we design a local context extractor (LCE) to capture short-range image features and use the channel attention mechanism to select informative local details across the extracted features to complement the global branch. By incorporating the CA-ViT as basic components, we further build the HDR-Transformer, a hierarchical network to reconstruct high-quality ghost-free HDR images. Extensive experiments on three benchmark datasets show that our approach outperforms state-of-the-art methods qualitatively and quantitatively with considerably reduced computational budgets. Codes are available at https://github.com/megvii-research/HDR-Transformer"



Paperid:781
Authors:Jin Wan; Hui Yin; Zhenyao Wu; Xinyi Wu; Yanting Liu; Song Wang
Title: Style-Guided Shadow Removal
Abstract:
Shadow removal is an important topic in image restoration, and it can benefit many computer vision tasks. State-of-the-art shadow-removal methods typically employ deep learning by minimizing a pixel-level difference between the de-shadowed region and their corresponding (pseudo) shadow-free version. After shadow removal, the shadow and non-shadow regions may exhibit inconsistent appearance, leading to a visually disharmonious image. To address this problem, we propose a style-guided shadow removal network (SG-ShadowNet) for better image style consistency after shadow removal. In SG-ShadowNet, we first learn the style representation of the non-shadow region via a simple region style estimator. Then we propose a novel effective normalization strategy with the region-level style to adjust the coarsely re-covered shadow region to be more harmonized with the rest of the image. Extensive experiments show that our proposed SG-ShadowNet outperforms all the existing competitive models and achieves a new state-of-the-art performance on ISTD+, SRD, and Video Shadow Removal benchmark datasets. Code is available at: https://github.com/jinwan1994/SG-ShadowNet.



Paperid:782
Authors:Youwei Li; Haibin Huang; Lanpeng Jia; Haoqiang Fan; Shuaicheng Liu
Title: D2C-SR: A Divergence to Convergence Approach for Real-World Image Super-Resolution
Abstract:
In this paper, we present D2C-SR, a novel framework for the task of real-world image super-resolution. As an ill-posed problem, the key challenge in super-resolution related tasks is there can be multiple predictions for a given low-resolution input. Most classical deep learning based approaches ignored the fundamental fact and lack explicit modeling of the underlying high-frequency distribution which leads to blurred results. Recently, some methods of GAN-based or learning super-resolution space can generate simulated textures but do not promise the accuracy of the textures which have low quantitative performance. Rethinking both, we learn the distribution of underlying high-frequency details in a discrete form and propose a two-stage pipeline: divergence stage to convergence stage. At divergence stage, we propose a tree-based structure deep network as our divergence backbone. Divergence loss is proposed to encourage the generated results from the tree-based network to diverge into possible high-frequency representations, which is our way of discretely modeling the underlying high-frequency distribution. At convergence stage, we assign spatial weights to fuse these divergent predictions to obtain the final output with more accurate details. Our approach provides a convenient end-to-end manner to inference. We conduct evaluations on several real-world benchmarks, including a new proposed D2CRealSR dataset with x8 scaling factor. Our experiments demonstrate that D2C-SR achieves better accuracy and visual improvements against state-of-the-art methods, with a significantly less parameters number and our D2C structure can also be applied as a generalized structure to some other methods to obtain improvement. Our codes and dataset are available at https://github.com/megvii-research/D2C-SR"



Paperid:783
Authors:Jaeseok Byun; Taebaek Hwang; Jianlong Fu; Taesup Moon
Title: GRIT-VLP: Grouped Mini-Batch Sampling for Efficient Vision and Language Pre-training
Abstract:
Most of the currently existing vision and language pre-training (VLP) methods have mainly focused on how to extract and align vision and text features. In contrast to the mainstream VLP methods, we highlight that two routinely applied steps during pre-training have crucial impact on the performance of the pre-trained model: in-batch hard negative sampling for image-text matching (ITM) and assigning the large masking probability for the masked language modeling (MLM). After empirically showing the unexpected effectiveness of above two steps, we systematically devise our GRIT-VLP, which adaptively samples mini-batches for more effective mining of hard negative samples for ITM while maintaining the computational cost for pre-training. Our method consists of three components: 1) GRouped mIni-baTch sampling (GRIT) strategy that collects similar examples in a mini-batch, 2) ITC consistency loss for improving the mining ability, and 3) enlarged masking probability for MLM. Consequently, we show our GRIT-VLP achieves a new state-of-the-art performance on various downstream tasks with much less computational cost. Furthermore, we demonstrate that our model is essentially in par with ALBEF, the previous state-of-the-art, only with one-third of training epochs on the same training data. Code is available at https://github.com/jaeseokbyun/GRIT-VLP.



Paperid:784
Authors:Yusheng Wang; Yunfan Lu; Ye Gao; Lin Wang; Zhihang Zhong; Yinqiang Zheng; Atsushi Yamashita
Title: Efficient Video Deblurring Guided by Motion Magnitude
Abstract:
Video deblurring is a highly under-constrained problem due to the spatially and temporally varying blur. An intuitive approach for video deblurring includes two steps: a) detecting the blurry region in the current frame; b) utilizing the information from clear regions in adjacent frames for current frame deblurring. To realize this process, our idea is to detect the pixel-wise blur level of each frame and combine it with video deblurring. To this end, we propose a novel framework that utilizes the motion magnitude prior (MMP) as guidance for efficient deep video deblurring. Specifically, as the pixel movement along its trajectory during the exposure time is positively correlated to the level of motion blur, we first use the average magnitude of optical flow from the high-frequency sharp frames to generate the synthetic blurry frames and their corresponding pixel-wise motion magnitude maps. We then build a dataset including the blurry frame and MMP pairs. The MMP is then learned by a compact CNN by regression. The MMP consists of both spatial and temporal blur level information, which can be further integrated into an efficient recurrent neural network (RNN) for video deblurring. We conduct intensive experiments to validate the effectiveness of the proposed methods on the public datasets.



Paperid:785
Authors:Zhiyuan Mao; Ajay Jaiswal; Zhangyang Wang; Stanley H. Chan
Title: Single Frame Atmospheric Turbulence Mitigation: A Benchmark Study and a New Physics-Inspired Transformer Model
Abstract:
Image restoration algorithms for atmospheric turbulence are known to be much more challenging to design than traditional ones such as blur or noise because the distortion caused by the turbulence is an entanglement of spatially varying blur, geometric distortion, and sensor noise. Existing CNN-based restoration methods built upon convolutional kernels with static weights are insufficient to handle the spatially dynamical atmospheric turbulence effect. To address this problem, in this paper, we propose a physics-inspired transformer model for imaging through atmospheric turbulence. The proposed network utilizes the power of transformer blocks to jointly extract a dynamical turbulence distortion map and restore a turbulence-free image. In addition, recognizing the lack of a comprehensive dataset, we collect and present two new real-world turbulence datasets that allow for evaluation with both classical objective metrics (e.g., PSNR and SSIM) and a new task-driven metric using text recognition accuracy. Both real testing sets and all related code will be made publicly available.



Paperid:786
Authors:A. Burakhan Koyuncu; Han Gao; Atanas Boev; Georgii Gaikov; Elena Alshina; Eckehard Steinbach
Title: Contextformer: A Transformer with Spatio-Channel Attention for Context Modeling in Learned Image Compression
Abstract:
Entropy modeling is a key component for high-performance image compression algorithms. Recent developments in autoregressive context modeling helped learning-based methods to surpass their classical counterparts. However, the performance of those models can be further improved due to the underexploited spatio-channel dependencies in latent space, and the suboptimal implementation of context adaptivity. Inspired by the adaptive characteristics of the transformers, we propose a transformer-based context model, named Contextformer, which generalizes the de facto standard attention mechanism to spatio-channel attention. We replace the context model of a modern compression framework with the Contextformer and test it on the widely used Kodak, CLIC2020, and Tecnick image datasets. Our experimental results show that the proposed model provides up to 11% rate savings compared to the standard Versatile Video Coding (VVC) Test Model (VTM) 16.2, and outperforms various learning-based models in terms of PSNR and MS-SSIM.



Paperid:787
Authors:Shunta Maeda
Title: Image Super-Resolution with Deep Dictionary
Abstract:
Since the first success of Dong et al., the deep-learning-based approach has become dominant in the field of single-image super-resolution. This replaces all the handcrafted image processing steps of traditional sparse-coding-based methods with a deep neural network. In contrast to sparse-coding-based methods, which explicitly create high/low-resolution dictionaries, the dictionaries in deep-learning-based methods are implicitly acquired as a nonlinear combination of multiple convolutions. One disadvantage of deep-learning-based methods is that their performance is degraded for images created differently from the training dataset (out-of-domain images). We propose an end-to-end super-resolution network with a deep dictionary (SRDD), where a high-resolution dictionary is explicitly learned without sacrificing the advantages of deep learning. Extensive experiments show that explicit learning of high-resolution dictionary makes the network more robust for out-of-domain test images while maintaining the performance of the in-domain test images.



Paperid:788
Authors:Mingyang Song; Yang Zhang; Tunç O. Aydın
Title: TempFormer: Temporally Consistent Transformer for Video Denoising
Abstract:
Video denoising is a low-level vision task that aims to restore high quality videos from noisy content. Vision Transformer (ViT) is a new machine learning architecture that has shown promising performance on both high-level and low-level image tasks. In this paper, we propose a modified ViT architecture for video processing tasks, introducing a new training strategy and loss function to enhance temporal consistency without compromising spatial quality. Specifically, we propose an efficient hybrid Transformer-based model which composes Spatio-Temporal Transformer Blocks (STTB) and 3D convolutional layers. The proposed STTB learns the temporal information between neighboring frames implicitly by utilizing the proposed Joint Spatio-Temporal Mixer module for attention calculation and feature aggregation in each ViT block. Moreover, existing methods suffer from temporal inconsistency artifacts that are problematic in practical cases and distracting to the viewers. We propose a sliding block strategy with recurrent architecture, and use a new loss term, Overlap Loss, to alleviate the flickering between adjacent frames. Our method produces state-of-the-art spatio-temporal denoising quality with significantly improved temporal coherency, and requires less computational resources to achieve comparable denoising quality with competing methods.



Paperid:789
Authors:Wooseok Jeong; Seung-Won Jung
Title: RAWtoBit: A Fully End-to-End Camera ISP Network
Abstract:
Image compression is an essential and last processing unit in the camera image signal processing (ISP) pipeline. While many studies have been made to replace the conventional ISP pipeline with a single end-to-end optimized deep learning model, image compression is barely considered as a part of the model. In this paper, we investigate the designing of a fully end-to-end optimized camera ISP incorporating image compression. To this end, we propose RAWtoBit network (RBN) that can effectively perform both tasks simultaneously. RBN is further improved with a novel knowledge distillation scheme by introducing two teacher networks specialized in each task. Extensive experiments demonstrate that our proposed method significantly outperforms alternative approaches in terms of rate-distortion trade-off.



Paperid:790
Authors:Fei Li; Lingfeng Shen; Yang Mi; Zhenbo Li
Title: DRCNet: Dynamic Image Restoration Contrastive Network
Abstract:
Image restoration aims to recover images from spatially-varying degradation. Most existing image-restoration models employed static CNN-based models, where the fixed learned filters cannot fit the diverse degradation well. To address this, in this paper, we propose a novel Dynamic Image Restoration Contrastive Network (DRCNet). The principal block in DRCNet is theDynamic Filter Restoration module (DFR), which mainly consists of the spatial filter branch and the energy-based attention branch. Specifically, the spatial filter branch suppresses spatial noise for varying spatial degradation; the energy-based attention branch guides the feature integration for better spatial detail recovery. To make degraded images and clean images more distinctive in the representation space, we develop a novel Intra-class Contrastive Regularization (Intra-CR) to serve as a constraint in the solution space for DRCNet. Meanwhile, our theoretical derivation proved Intra-CR owns less sensitivity towards hyper-parameter selection than previous contrastive regularization. DRCNet achieves state-of-the-art results on the ten widely-used benchmarks in image restoration. Besides, we conduct ablation studies to show the effectiveness of the DFR module and Intra-CR, respectively.



Paperid:791
Authors:Byeong-Ju Han; Jae-Young Sim
Title: Zero-Shot Learning for Reflection Removal of Single 360-Degree Image
Abstract:
The existing methods for reflection removal mainly focus on removing blurry and weak reflection artifacts and thus often fail to work with severe and strong reflection artifacts. However, in many cases, real reflection artifacts are sharp and intensive enough such that even humans cannot completely distinguish between the transmitted and reflected scenes. In this paper, we attempt to remove such challenging reflection artifacts using 360-degree images. We adopt the zero-shot learning scheme to avoid the burden of collecting paired data for supervised learning and the domain gap between different datasets. We first search for the reference image of the reflected scene in a 360-degree image based on the reflection geometry, which is then used to guide the network to restore the faithful colors of the reflection image. We collect 30 test 360-degree images exhibiting challenging reflection artifacts and demonstrate that the proposed method outperforms the existing state-of-the-art methods on 360-degree images.



Paperid:792
Authors:Yidi Shao; Chen Change Loy; Bo Dai
Title: Transformer with Implicit Edges for Particle-Based Physics Simulation
Abstract:
Particle-based systems provide a flexible and unified way to simulate physics systems with complex dynamics. Most existing data-driven simulators for particle-based systems adopt graph neural networks (GNNs) as their network backbones, as particles and their interactions can be naturally represented by graph nodes and graph edges. However, while particle-based systems usually contain hundreds even thousands of particles, the explicit modeling of particle interactions as graph edges inevitably leads to a significant computational overhead, due to the increased number of particle interactions. Consequently, in this paper we propose a novel Transformer-based method, dubbed as Transformer with Implicit Edges (TIE), to capture the rich semantics of particle interactions in an edge-free manner. The core idea of TIE is to decentralize the computation involving pair-wise particle interactions into per-particle updates. This is achieved by adjusting the self-attention module to resemble the update formula of graph edges in GNN. To improve the generalization ability of TIE, we further amend TIE with learnable material-specific abstract particles to disentangle global material-wise semantics from local particle-wise semantics. We evaluate our model on diverse domains of varying complexity and materials. Compared with existing GNN-based methods, without bells and whistles, TIE achieves superior performance and generalization across all these domains. Codes and models are available at https://github.com/ftbabi/TIE_ECCV2022.git.



Paperid:793
Authors:Shuai Wang; Lei Zhu; Huazhu Fu; Jing Qin; Carola-Bibiane Schönlieb; Wei Feng; Song Wang
Title: Rethinking Video Rain Streak Removal: A New Synthesis Model and a Deraining Network with Video Rain Prior
Abstract:
Existing video synthetic models and deraining methods are mostly built on a simplified video rain model assuming that rain streak layers of different video frames are uncorrelated, thereby producing degraded performance on real-world rainy videos. To address this problem, we devise a new video rain synthesis model with the concept of rain streak motions to enforce a consistency of rain layers between video frames, thereby generating more realistic rainy video data for network training, and then develop a recurrent disentangled deraining network (RDD-Net) based on our video rain model for boosting video deraining. More specifically, taking adjacent frames of a key frame as the input, our RDD-Net recurrently aggregates each adjacent frame and the key frame by a fusion module, and then devise a disentangle model to decouple the fused features by predicting not only a clean background layer and a rain layer, but also a rain streak motion layer. After that, we develop three attentive recovery modules to combine the decoupled features from different adjacent frames for predicting the final derained result of the key frame. Experiments on three widely-used benchmark datasets and a collected dataset, as well as real-world rainy videos show that our RDD-Net quantitatively and qualitatively outperforms state-of-the-art deraining methods. Our code, our dataset, and our results on four datasets are released at https://github.com/wangshauitj/RDD-Net.



Paperid:794
Authors:Jinjin Gu; Haoming Cai; Chenyu Dong; Ruofan Zhang; Yulun Zhang; Wenming Yang; Chun Yuan
Title: Super-Resolution by Predicting Offsets: An Ultra-Efficient Super-Resolution Network for Rasterized Images
Abstract:
Rendering high-resolution (HR) graphics brings substantial computational costs. Efficient graphics super-resolution (SR) methods may achieve HR rendering with small computing resources and have attracted extensive research interests in industry and research communities. We present a new method for real-time SR for computer graphics, namely Super-Resolution by Predicting Offsets (SRPO). Our algorithm divides the image into two parts for processing, i.e., sharp edges and flatter areas. For edges, different from the previous SR methods that take the anti-aliased images as inputs, our proposed SRPO takes advantage of the characteristics of rasterized images to conduct SR on the rasterized images. To complement the residual between HR and low-resolution (LR) rasterized images, we train an ultra-efficient network to predict the offset maps to move the appropriate surrounding pixels to the new positions. For flat areas, we found simple interpolation methods can already generate reasonable output. We finally use a guided fusion operation to integrate the sharp edges generated by the network and flat areas by the interpolation method to get the final SR image. The proposed network only contains 8,434 parameters and can be accelerated by network quantization. Extensive experiments show that the proposed SRPO can achieve superior visual effects at a smaller computational cost than the existing state-of-the-art methods.



Paperid:795
Authors:Zhihang Zhong; Xiao Sun; Zhirong Wu; Yinqiang Zheng; Stephen Lin; Imari Sato
Title: Animation from Blur: Multi-modal Blur Decomposition with Motion Guidance
Abstract:
We study the challenging problem of recovering detailed motion from a single motion-blurred image. Existing solutions to this problem estimate a single image sequence without considering the motion ambiguity for each region. Therefore, the results tend to converge to the mean of the multi-modal possibilities. In this paper, we explicitly account for such motion ambiguity, allowing us to generate multiple plausible solutions all in sharp detail. The key idea is to introduce a motion guidance representation, which is a compact quantization of 2D optical flow with only four discrete motion directions. Conditioned on the motion guidance, the blur decomposition is led to a specific, unambiguous solution by using a novel two-stage decomposition network. We propose a unified framework for blur decomposition, which supports various interfaces for generating our motion guidance, including human input, motion information from adjacent video frames, and learning from a video dataset. Extensive experiments on synthesized datasets and real-world data show that the proposed framework is qualitatively and quantitatively superior to previous methods, and also offers the merit of producing physically plausible and diverse solutions. Code is available at https://github.com/zzh-tech/Animation-from-Blur.



Paperid:796
Authors:Yibo Shi; Yunying Ge; Jing Wang; Jue Mao
Title: AlphaVC: High-Performance and Efficient Learned Video Compression
Abstract:
Recently, learned video compression has drawn lots of attention and show a rapid development trend with promising results. However, the previous works still suffer from some criticial issues and have a performance gap with traditional compression standards in terms of widely used PSNR metric. In this paper, we propose several techniques to effectively improve the performance. First, to address the problem of accumulative error, we introduce a conditional-I-frame as the first frame in the GoP, which stabilizes the reconstructed quality and saves the bit-rate. Second, to efficiently improve the accuracy of inter prediction without increasing the complexity of decoder, we propose a pixel-to-feature motion prediction method at encoder side that helps us to obtain high-quality motion information. Third, we propose a probability-based entropy skipping method, which not only brings performance gain, but also greatly reduces the runtime of entropy coding. With these powerful techniques, this paper proposes AlphaVC, a high-performance and efficient learned video compression scheme. To the best of our knowledge, AlphaVC is the first E2E AI codec that exceeds the latest compression standard VVC on all common test datasets for both PSNR (-28.2% BD-rate saving) and MSSSIM (-52.2% BD-rate saving), and has very fast encoding (0.001x VVC) and decoding (1.69x VVC) speeds.



Paperid:797
Authors:Meng Li; Shangyin Gao; Yihui Feng; Yibo Shi; Jing Wang
Title: Content-Oriented Learned Image Compression
Abstract:
In recent years, with the development of deep neural networks, end-to-end optimized image compression has made significant progress and exceeded the classic methods in terms of rate-distortion performance. However, most learning-based image compression methods are unlabeled and do not consider image semantics or content when optimizing the model. In fact, human eyes have different sensitivities to different content, so the image content also needs to be considered when optimizing the model. In this paper, we propose a content-oriented image compression method, which handles different kinds of image contents with different strategies. Extensive experiments show that the proposed method achieves competitive results compared with state-of-the-art end-to-end learned image compression methods or classic methods.



Paperid:798
Authors:Lin Zhang; Xin Li; Dongliang He; Fu Li; Yili Wang; Zhaoxiang Zhang
Title: RRSR:Reciprocal Reference-Based Image Super-Resolution with Progressive Feature Alignment and Selection
Abstract:
Reference-based image super-resolution (RefSR) is a promising SR branch and has shown great potential in overcoming the limitations of single image super-resolution. While previous state-of-the-art RefSR methods mainly focus on improving the efficacy and robustness of reference feature transfer, it is generally overlooked that a well reconstructed SR image should enable better SR reconstruction for its similar LR images when it is referred to as. Therefore, in this work, we propose a reciprocal learning framework that can appropriately leverage such a fact to reinforce the learning of a RefSR network. Besides, we deliberately design a progressive feature alignment and selection module for further improving the RefSR task. The newly proposed module aligns reference-input images at multi-scale feature spaces and performs reference-aware feature selection in a progressive manner, thus more precise reference features can be transferred into the input features and the network capability is enhanced. Our reciprocal learning paradigm is model-agnostic and it can be applied to arbitrary RefSR models. We empirically show that multiple recent state-of-the-art RefSR models can be consistently improved with our reciprocal learning paradigm. Furthermore, our proposed model together with the reciprocal learning strategy sets new state-of-the-art performances on multiple benchmarks.



Paperid:799
Authors:Haoqing Wang; Zhi-Hong Deng
Title: Contrastive Prototypical Network with Wasserstein Confidence Penalty
Abstract:
Unsupervised few-shot learning aims to learn the inductive bias from unlabeled dataset for solving the novel few-shot tasks. The existing unsupervised few-shot learning models and the contrastive learning models follow a unified paradigm. Therefore, we conduct empirical study under this paradigm and find that pairwise contrast, meta losses and large batch size are the important design factors. This results in our CPN (Contrastive Prototypical Network) model, which combines the prototypical loss with pairwise contrast and outperforms the existing models from this paradigm with modestly large batch size. Furthermore, the one-hot prediction target in CPN could lead to learning the sample-specific information. To this end, we propose Wasserstein Confidence Penalty which can impose appropriate penalty on overconfident predictions based on the semantic relationships among pseudo classes. Our full model, CPNWCP (Contrastive Prototypical Network with Wasserstein Confidence Penalty), achieves state-of-the-art performance on miniImageNet and tieredImageNet under unsupervised setting. Our code is available at https://github.com/Haoqing-Wang/CPNWCP.



Paperid:800
Authors:Xinyi Zou; Yan Yan; Jing-Hao Xue; Si Chen; Hanzi Wang
Title: Learn-to-Decompose: Cascaded Decomposition Network for Cross-Domain Few-Shot Facial Expression Recognition
Abstract:
Most existing compound facial expression recognition (FER) methods rely on large-scale labeled compound expression data for training. However, collecting such data is labor-intensive and time-consuming. In this paper, we address the compound FER task in the cross-domain few-shot learning (FSL) setting, which requires only a few samples of compound expressions in the target domain. Specifically, we propose a novel cascaded decomposition network (CDNet), which cascades several learn-to-decompose modules with shared parameters based on a sequential decomposition mechanism, to obtain a transferable feature space. To alleviate the overfitting problem caused by limited base classes in our task, a partial regularization strategy is designed to effectively exploit the best of both episodic training and batch training. By training across similar tasks on multiple basic expression datasets, CDNet learns the ability of learn-to-decompose that can be easily adapted to identify unseen compound expressions. Extensive experiments on both in-the-lab and in-the-wild compound expression datasets demonstrate the superiority of our proposed CDNet against several state-of-the-art FSL methods. Code is available at: https://github.com/zouxinyi0625/CDNet.



Paperid:801
Authors:Qi Fan; Wenjie Pei; Yu-Wing Tai; Chi-Keung Tang
Title: Self-Support Few-Shot Semantic Segmentation
Abstract:
Existing few-shot segmentation methods have achieved great progress based on the support-query matching framework. But they still heavily suffer from the limited coverage of intra-class variations from the few-shot supports. Motivated by the simple Gestalt principle that pixels belonging to the same object are more similar than those to different objects of same class, we propose a novel self-support matching idea to alleviate this problem. It uses query prototypes to match query features, where the query prototypes are collected from high-confidence prediction regions. This strategy can effectively capture the consistent underlying characteristics of the query objects, and thus fittingly match query features. We also propose an adaptive self-support background prototype generation module and self-support loss to further facilitate the self-support matching procedure. Our self-support network substantially improves the prototype quality, benefits more improvement from stronger backbones and more supports, and achieves SOTA on multiple datasets. Codes are at https://github.com/fanq15/SSP.



Paperid:802
Authors:Qi Fan; Chi-Keung Tang; Yu-Wing Tai
Title: Few-Shot Object Detection with Model Calibration
Abstract:
Few-shot object detection (FSOD) targets at transferring knowledge from known to unknown classes to detect objects of novel classes. However, previous works ignore the model bias problem inherent in the transfer learning paradigm. Such model bias causes overfitting toward the training classes and destructs the well-learned transferable knowledge. In this paper, we pinpoint and comprehensively investigate the model bias problem in FSOD models and propose a simple yet effective method to address the model bias problem with the facilitation of model calibrations in three levels: 1) Backbone calibration to preserve the well-learned prior knowledge and relieve the model bias toward base classes, 2) RPN calibration to rescue unlabeled objects of novel classes and, 3) Detector calibration to prevent the model bias toward a few training samples for novel classes. Specifically, we leverage the overlooked classification dataset to facilitate our model calibration procedure, which has only been used for pre-training in other related works. We validate the effectiveness of our model calibration method on the popular Pascal VOC and MS COCO datasets, where our method achieves very promising performance. Codes are released at https://github.com/fanq15/FewX.



Paperid:803
Authors:Yuning Lu; Liangjian Wen; Jianzhuang Liu; Yajing Liu; Xinmei Tian
Title: Self-Supervision Can Be a Good Few-Shot Learner
Abstract:
Existing few-shot learning (FSL) methods rely on training with a large labeled dataset, which prevents them from leveraging abundant unlabeled data. From an information-theoretic perspective, we propose an effective unsupervised FSL method, learning representations with self-supervision. Following the InfoMax principle, our method learns comprehensive representations by capturing the intrinsic structure of the data. Specifically, we maximize the mutual information (MI) of instances and their representations with a low-bias MI estimator to perform self-supervised pre-training. Rather than supervised pre-training focusing on the discriminable features of the seen classes, our self-supervised model has less bias toward the seen classes, resulting in better generalization for unseen classes. We explain that supervised pre-training and self-supervised pre-training are actually maximizing different MI objectives. Extensive experiments are further conducted to analyze their FSL performance with various training settings. Surprisingly, the results show that self-supervised pre-training can outperform supervised pre-training under the appropriate conditions. Compared with state-of-the-art FSL methods, our approach achieves comparable performance on widely used FSL benchmarks without any labels of the base classes.



Paperid:804
Authors:Jinxiang Lai; Siqian Yang; Wenlong Liu; Yi Zeng; Zhongyi Huang; Wenlong Wu; Jun Liu; Bin-Bin Gao; Chengjie Wang
Title: tSF: Transformer-Based Semantic Filter for Few-Shot Learning
Abstract:
Few-Shot Learning (FSL) alleviates the data shortage challenge via embedding discriminative target-aware features among plenty seen (base) and few unseen (novel) labeled samples. Most feature embedding modules in recent FSL methods are specially designed for corresponding learning tasks (e.g., classification, segmentation, and object detection), which limits the utility of embedding features. To this end, we propose a light and universal module named transformer-based Semantic Filter (tSF), which can be applied for different FSL tasks. The proposed tSF redesigns the inputs of a transformer-based structure by a semantic filter, which not only embeds the knowledge from whole base set to novel set but also filters semantic features for target category. Furthermore, the parameters of tSF is equal to half of a standard transformer block (less than 1M). In the experiments, our tSF is able to boost the performances in different classic few-shot learning tasks (about 2% improvement), especially outperforms the state-of-the-arts on multiple benchmark datasets in few-shot classification task.



Paperid:805
Authors:Yanxu Hu; Andy J. Ma
Title: Adversarial Feature Augmentation for Cross-Domain Few-Shot Classification
Abstract:
Few-shot classification is a promising approach to solving the problem of classifying novel classes with only limited annotated data for training. Existing methods based on meta-learning predict novel-class labels for (target domain) testing tasks via meta knowledge learned from (source domain) training tasks of base classes. However, most existing works may fail to generalize to novel classes due to the probably large domain discrepancy across domains. To address this issue, we propose a novel adversarial feature augmentation (AFA) method to bridge the domain gap in few-shot learning. The feature augmentation is designed to simulate distribution variations by maximizing the domain discrepancy. During adversarial training, the domain discriminator is learned by distinguishing the augmented features (unseen domain) from the original ones (seen domain), while the domain discrepancy is minimized to obtain the optimal feature encoder. The proposed method is a plug-and-play module that can be easily integrated into existing few-shot learning methods based on meta-learning. Extensive experiments on nine datasets demonstrate the superiority of our method for cross-domain few-shot classification compared with the state of the art. Code is available at https://github.com/youthhoo/AFA_For_Few_shot_learning"



Paperid:806
Authors:Yue Xu; Yong-Lu Li; Jiefeng Li; Cewu Lu
Title: Constructing Balance from Imbalance for Long-Tailed Image Recognition
Abstract:
Long-tailed image recognition presents massive challenges to deep learning systems since the imbalance between majority (head) classes and minority (tail) classes severely skews the data-driven deep neural networks. Previous methods tackle with data imbalance from the viewpoints of data distribution, feature space, and model design, etc.In this work, instead of directly learning a recognition model, we suggest confronting the bottleneck of head-to-tail bias before classifier learning, from the previously omitted perspective of balancing label space. To alleviate the head-to-tail bias, we propose a concise paradigm by progressively adjusting label space and dividing the head classes and tail classes, dynamically constructing balance from imbalance to facilitate the classification. With flexible data filtering and label space mapping, we can easily embed our approach to most classification models, especially the decoupled training methods. Besides, we find the separability of head-tail classes varies among different features with different inductive biases. Hence, our proposed model also provides a feature evaluation method and paves the way for long-tailed feature learning. Extensive experiments show that our method can boost the performance of state-of-the-arts of different types on widely-used benchmarks. Code is available at https://github.com/silicx/DLSA.



Paperid:807
Authors:Yuzhe Yang; Hao Wang; Dina Katabi
Title: "On Multi-Domain Long-Tailed Recognition, Imbalanced Domain Generalization and Beyond"
Abstract:
Real-world data often exhibit imbalanced label distributions. Existing studies on data imbalance focus on single-domain settings, i.e., samples are from the same data distribution. However, natural data can originate from distinct domains, where a minority class in one domain could have abundant instances from other domains. We formalize the task of Multi-Domain Long-Tailed Recognition (MDLT), which learns from multi-domain imbalanced data, addresses label imbalance, domain shift, and divergent label distributions across domains, and generalizes to all domain-class pairs. We first develop the domain-class transferability graph, and show that such transferability governs the success of learning in MDLT. We then propose BoDA, a theoretically grounded learning strategy that tracks the upper bound of transferability statistics, and ensures balanced alignment and calibration across imbalanced domain-class distributions. We curate five MDLT benchmarks based on widely-used multi-domain datasets, and compare BoDA to twenty algorithms that span different learning strategies. Extensive and rigorous experiments verify the superior performance of BoDA. Further, as a byproduct, BoDA establishes new state-of-the-art on Domain Generalization benchmarks, highlighting the importance of addressing data imbalance across domains, which can be crucial for improving generalization to unseen domains. Code and data are available at: https://github.com/YyzHarry/multi-domain-imbalance.



Paperid:808
Authors:Qi Fan; Chi-Keung Tang; Yu-Wing Tai
Title: Few-Shot Video Object Detection
Abstract:
We introduce Few-Shot Video Object Detection (FSVOD) with three contributions to visual learning in our highly diverse and dynamic world: 1) a large-scale video dataset FSVOD-500 comprising of 500 classes with class-balanced videos in each category for few-shot learning; 2) a novel Tube Proposal Network (TPN) to generate high-quality video tube proposals for aggregating feature representation for the target video object which can be highly dynamic; 3) a strategically improved Temporal Matching Network (TMN+) for matching representative query tube features with better discriminative ability thus achieving higher diversity. Our TPN and TMN+ are jointly and end-to-end trained. Extensive experiments demonstrate that our method produces significantly better detection results on two few-shot video object detection datasets compared to image-based methods and other naive video-based extensions. Codes and datasets are released at https://github.com/fanq15/FewX.



Paperid:809
Authors:Minghao Fu; Yun-Hao Cao; Jianxin Wu
Title: Worst Case Matters for Few-Shot Recognition
Abstract:
Few-shot recognition learns a recognition model with very few (e.g., 1 or 5) images per category, and current few-shot learning methods focus on improving the average accuracy over many episodes. We argue that in real-world applications we may often only try one episode instead of many, and hence maximizing the worst-case accuracy is more important than maximizing the average accuracy. We empirically show that a high average accuracy not necessarily means a high worst-case accuracy. Since this objective is not accessible, we propose to reduce the standard deviation and increase the average accuracy simultaneously. In turn, we devise two strategies from the bias-variance tradeoff perspective to implicitly reach this goal: a simple yet effective stability regularization (SR) loss together with model ensemble to reduce variance during fine-tuning, and an adaptability calibration mechanism to reduce the bias. Extensive experiments on benchmark datasets demonstrate the effectiveness of the proposed strategies, which outperforms current state-of-the-art methods with a significant margin in terms of not only average, but also worst-case accuracy.



Paperid:810
Authors:Kai Yi; Xiaoqian Shen; Yunhao Gou; Mohamed Elhoseiny
Title: Exploring Hierarchical Graph Representation for Large-Scale Zero-Shot Image Classification
Abstract:
The main question we address in this paper is how to scale up visual recognition of unseen classes, also known as zero-shot learning, to tens of thousands of categories as in the ImageNet-21K benchmark. At this scale, especially with many fine-grained categories included in ImageNet-21K, it is critical to learn quality visual semantic representations that are discriminative enough to recognize unseen classes and distinguish them from seen ones. We propose a Hierarchical Graphical knowledge Representation framework for the confidence-based classification method, dubbed as HGR-Net. Our experimental results demonstrate that HGR-Net can grasp class inheritance relations by utilizing hierarchical conceptual knowledge. Our method significantly outperformed all existing techniques, boosting the performance by 7% compared to the runner-up approach on the ImageNet-21K benchmark. We show that HGR-Net is learning-efficient in few-shot scenarios. We also analyzed our method on smaller datasets like ImageNet-21K-P, 2-hops, and 3-hops, demonstrating its generalization ability. Our benchmark and code are available at https://kaiyi.me/p/hgrnet.html.



Paperid:811
Authors:Zhitong Xiong; Haopeng Li; Xiao Xiang Zhu
Title: Doubly Deformable Aggregation of Covariance Matrices for Few-Shot Segmentation
Abstract:
Training semantic segmentation models with few annotated samples has great potential in various real-world applications. For the few-shot segmentation task, the main challenge is how to accurately measure the semantic correspondence between the support and query samples with limited training data. To address this problem, we propose to aggregate the learnable covariance matrices with a deformable 4D Transformer to effectively predict the segmentation map. Specifically, in this work, we first devise a novel hard example mining mechanism to learn covariance kernels for the Gaussian process. The learned covariance kernel functions have great advantages over existing cosine similarity-based methods in correspondence measurement. Based on the learned covariance kernels, an efficient doubly deformable 4D Transformer module is designed to adaptively aggregate feature similarity maps into segmentation results. By combining these two designs, the proposed method can not only set new state-of-the-art performance on public benchmarks, but also converge extremely faster than existing methods. Experiments on three public datasets have demonstrated the effectiveness of our method.



Paperid:812
Authors:Xinyu Shi; Dong Wei; Yu Zhang; Donghuan Lu; Munan Ning; Jiashun Chen; Kai Ma; Yefeng Zheng
Title: Dense Cross-Query-and-Support Attention Weighted Mask Aggregation for Few-Shot Segmentation
Abstract:
Research into Few-shot Semantic Segmentation (FSS) has attracted great attention, with the goal to segment target objects in a query image given only a few annotated support images of the target class. A key to this challenging task is to fully utilize the information in the support images by exploiting fine-grained correlations between the query and support images. However, most existing approaches either compressed the support information into a few class-wise prototypes, or used partial support information (e.g., only foreground) at the pixel level, causing non-negligible information loss. In this paper, we propose Dense pixel-wise Cross-query-and-support Attention weighted Mask Aggregation (DCAMA), where both foreground and background support information are fully exploited via multi-level pixel-wise correlations between paired query and support features. Implemented with the scaled dot-product attention in the Transformer architecture, DCAMA treats every query pixel as a token, computes its similarities with all support pixels, and predicts its segmentation label as an additive aggregation of all the support pixels’ labels--weighted by the similarities. Based on the unique formulation of DCAMA, we further propose efficient and effective one-pass inference for n-shot segmentation, where pixels of all support images are collected for the mask aggregation at once. Experiments show that our DCAMA significantly advances the state of the art on standard FSS benchmarks of PASCAL-5i, COCO-20i, and FSS-1000, e.g., with 3.1%, 9.7%, and 3.6% absolute improvements in 1-shot mIoU over previous best records. Ablative studies also verify the design DCAMA.



Paperid:813
Authors:Xingping Dong; Jianbing Shen; Ling Shao
Title: Rethinking Clustering-Based Pseudo-Labeling for Unsupervised Meta-Learning
Abstract:
The pioneering unsupervised meta-learning work is a clustering-based pseudo-labeling method, which is model-agnostic and can utilize supervised algorithms for learning from unlabeled data. However, it often suffers from label inconsistency and limited diversity, which leads to poor performance. In this work, we prove that the core reason for this comes from the lack of a clustering-friendly property in the embedding space. Through comprehensive experimental validations, we break this restriction by minimizing the inter-class to intra-class similarity ratio to provide clustering-friendly embedding features. Surprisingly, we only utilize a simple clustering algorithm (k-means) on our embedding space to obtain pseudo-labels and achieve significant improvement. Moreover, we adopt a progressive evaluation mechanism to seek more diverse samples in order to further alleviate the limited diversity problem. Besides, our approach is model-agnostic and can easily be integrated into existing supervised methods. To demonstrate its generalization ability, we integrate it into two representative algorithms: MAML and EP. The results on three major few-shot benchmarks clearly show that the proposed method achieves significant improvement compared to the state-of-the-art models. Notably, our approach outperforms the corresponding supervised method in two tasks.



Paperid:814
Authors:Shreyank N Gowda; Laura Sevilla-Lara; Frank Keller; Marcus Rohrbach
Title: CLASTER: Clustering with Reinforcement Learning for Zero-Shot Action Recognition
Abstract:
Zero-Shot action recognition is the task of recognizing action classes without visual examples. The problem can be seen as learning a representation on seen classes which generalizes well to instances of unseen classes, without losing discriminability between classes. Neural networks are able to model highly complex boundaries between visual classes, which explains their success as supervised models. However, in Zero-Shot learning, these highly specialized class boundaries may overfit to the seen classes and not transfer well from seen to unseen classes. We propose a novel cluster-based representation, which regularizes the learning process, yielding a representation that generalizes well to instances from unseen classes. We optimize the clustering using reinforcement learning, which we observe is critical. We call the proposed method CLASTER and observe that it consistently outperforms the state-of-the-art in all standard Zero-Shot video datasets, including UCF101, HMDB51 and Olympic Sports; both in the standard Zero-Shot evaluation and the generalized Zero-Shot learning. We see improvements of up to 11.9% over SOTA.



Paperid:815
Authors:Townim Chowdhury; Ali Cheraghian; Sameera Ramasinghe; Sahar Ahmadi; Morteza Saberi; Shafin Rahman
Title: Few-Shot Class-Incremental Learning for 3D Point Cloud Objects
Abstract:
Few-shot class-incremental learning (FSCIL) aims to incrementally fine-tune a model trained on base classes for a novel set of classes using a few examples without forgetting the previous training. Recent efforts of FSCIL addresses this problem primarily on 2D image data. However, due to the advancement of camera technology, 3D point cloud data has become more available than ever, which warrants considering FSCIL on 3D data. In this paper, we address FSCIL in the 3D domain. In addition to well-known problems of catastrophic forgetting of past knowledge and overfitting of few-shot data, 3D FSCIL can bring newer challenges. For example, base classes may contain many synthetic instances in a realistic scenario. In contrast, only a few real-scanned samples (from RGBD sensors) of novel classes are available in incremental steps. Due to the data variation from synthetic to real, FSCIL endures additional challenges, degrading performance in later incremental steps. We attempt to solve this problem by using Microshapes (orthogonal basis vectors) describing any 3D objects using a pre-defined set of rules. It supports incremental training with few-shot examples minimizing synthetic to real data variation. We propose new test protocols for 3D FSCIL using popular synthetic datasets, ModelNet and ShapeNet and 3D real-scanned datasets, ScanObjectNN and Common Objects in 3D (CO3D). By comparing state-of-the-art methods, we establish the effectiveness of our approach on the 3D domain.



Paperid:816
Authors:Zhenyi Wang; Li Shen; Le Fang; Qiuling Suo; Donglin Zhan; Tiehang Duan; Mingchen Gao
Title: Meta-Learning with Less Forgetting on Large-Scale Non-stationary Task Distributions
Abstract:
The paradigm of machine intelligence moves from purely supervised learning to a more practical scenario when many loosely related unlabeled data are available and labeled data is scarce. Most existing algorithms assume that the underlying task distribution is stationary. Here we consider a more realistic and challenging setting in that task distributions evolve over time. We name this problem as Semi-supervised meta-learning with Evolving Task diStributions, abbreviated as SETS. Two key challenges arise in this more realistic setting: (i) how to use unlabeled data in the presence of a large amount of unlabeled out-of-distribution (OOD) data; and (ii) how to prevent catastrophic forgetting on previously learned task distributions due to the task distribution shift. We propose an OOD Robust and knowleDge presErved semi-supeRvised meta-learning approach (ORDER) we use ORDER to denote the task distributions sequentially arrive with some ORDER, to tackle these two major challenges. Specifically, our ORDER introduces a novel mutual information regularization to robustify the model with unlabeled OOD data and adopts an optimal transport regularization to remember previously learned knowledge in feature space. In addition, we test our method on a very challenging dataset: SETS on large-scale non-stationary semi-supervised task distributions consisting of (at least) 72K tasks. With extensive experiments, we demonstrate the proposed ORDER alleviates forgetting on evolving task distributions and is more robust to OOD data than related strong baselines.



Paperid:817
Authors:Ziyu Jiang; Tianlong Chen; Xuxi Chen; Yu Cheng; Luowei Zhou; Lu Yuan; Ahmed Awadallah; Zhangyang Wang
Title: DNA: Improving Few-Shot Transfer Learning with Low-Rank Decomposition and Alignment
Abstract:
Self-supervised (SS) learning has achieved remarkable success in learning strong representation for in-domain few-shot and semi-supervised tasks. However, when transferring such representations to downstream tasks with domain shifts, the performance degrades compared to its supervised counterpart, especially at the few-shot regime. In this paper, we proposed to boost the transferability of the self-supervised pre-trained models on cross-domain tasks via a novel self-supervised alignment step on the target domain using only unlabeled data before conducting the downstream supervised fine-tuning. A new reparameterization of the pre-trained weights is also presented to mitigate the potential catastrophic forgetting during the alignment step. It involves low-rank and sparse decomposition, that can elegantly balance between preserving the source domain knowledge without forgetting (via fixing the low-rank subspace), and the extra flexibility to absorb the new out-of-the-domain knowledge (via freeing the sparse residual). Our resultant framework, termed Decomposition-and-Alignment (DnA), significantly improves the few-shot transfer performance of the SS pre-trained model to downstream tasks with domain gaps.



Paperid:818
Authors:Rongkai Ma; Pengfei Fang; Gil Avraham; Yan Zuo; Tianyu Zhu; Tom Drummond; Mehrtash Harandi
Title: Learning Instance and Task-Aware Dynamic Kernels for Few-Shot Learning
Abstract:
Learning and generalizing to novel concepts with few samples (Few-Shot Learning) is still an essential challenge to real-world applications. A principle way of achieving few-shot learning is to realize a model that can rapidly adapt to the context of a given task. Dynamic networks have been shown capable of learning content-adaptive parameters efficiently, making them suitable for few-shot learning. In this paper, we propose to learn the dynamic kernels of a convolution network as a function of the task at hand, enabling faster generalization. To this end, we obtain our dynamic kernels based on the entire task and each sample and develop a mechanism further conditioning on each individual channel and position independently. This results in dynamic kernels that simultaneously attend to the global information whilst also considering minuscule details available. We empirically show that our model improves performance on few-shot classification and detection tasks, achieving a tangible improvement over several baseline models. This includes state-of-the-art results on 4 few-shot classification benchmarks: mini-ImageNet, tiered-ImageNet, CUB and FC100 and competitive results on a few-shot detection dataset: MS COCO-PASCAL-VOC.



Paperid:819
Authors:Quande Liu; Youpeng Wen; Jianhua Han; Chunjing Xu; Hang Xu; Xiaodan Liang
Title: Open-World Semantic Segmentation via Contrasting and Clustering Vision-Language Embedding
Abstract:
To bridge the gap between supervised semantic segmentation and real-world applications that acquire one model to recognize arbitrary new concepts, recent zero-shot segmentation attracts a lot of attention by exploring the relationships between unseen and seen object categories, yet requiring large amounts of densely-annotated data with diverse base classes. In this paper, we propose a new open-world semantic segmentation pipeline that makes the first attempt to learn to segment semantic objects of various open-world categories without any efforts on dense annotations, by purely exploiting the image-caption data that naturally exist on the Internet. Our method, Vision-language-driven Semantic Segmentation (ViL-Seg), employs an image and a text encoder to generate visual and text embeddings for the image-caption data, with two core components that endow its segmentation ability: First, the image encoder is jointly trained with a vision-based contrasting and a cross-modal contrasting, which encourage the visual embeddings to preserve both fine-grained semantics and high-level category information that are crucial for the segmentation task. Furthermore, an online clustering head is devised over the image encoder, which allows us to dynamically segment the visual embeddings into distinct semantic groups such that they can be classified by comparing with various text embeddings to complete our segmentation pipeline. Experiments show that without using any data with dense annotations, our method can directly segment objects of arbitrary categories, outperforming zero-shot segmentation methods that require data labeling on three benchmark datasets.



Paperid:820
Authors:Zhanyuan Yang; Jinghua Wang; Yingying Zhu
Title: Few-Shot Classification with Contrastive Learning
Abstract:
A two-stage training paradigm consisting of sequential pre-training and meta-training stages has been widely used in current few-shot learning (FSL) research. Many of these methods use self-supervised learning and contrastive learning to achieve new state-of-the-art results. However, the potential of contrastive learning in both stages of FSL training paradigm is still not fully exploited. In this paper, we propose a novel contrastive learning-based framework that seamlessly integrates contrastive learning into both stages to improve the performance of few-shot classification. In the pre-training stage, we propose a self-supervised contrastive loss in the forms of feature vector vs. feature map and feature map vs. feature map, which uses global and local information to learn good initial representations. In the meta-training stage, we propose a cross-view episodic training mechanism to perform the nearest centroid classification on two different views of the same episode and adopt a distance-scaled contrastive loss based on them. These two strategies force the model to overcome the bias between views and promote the transferability of representations. Extensive experiments on three benchmark datasets demonstrate that our method achieves competitive results.



Paperid:821
Authors:Shan Zhang; Naila Murray; Lei Wang; Piotr Koniusz
Title: Time-rEversed diffusioN tEnsor Transformer: A New TENET of Few-Shot Object Detection
Abstract:
In this paper, we tackle the challenging problem of Few-shot Object Detection. Existing FSOD pipelines (i) use average-pooled representations that result in information loss; and/or (ii) discard position information that can help detect object instances. Consequently, such pipelines are sensitive to large intra-class appearance and geometric variations between support and query images. To address these drawbacks, we propose a Time-rEversed diffusioN tEnsor Transformer (TENET), which i) forms high-order tensor representations that capture multi-way feature occurrences that are highly discriminative, and ii) uses a transformer that dynamically extracts correlations between the query image and the entire support set, instead of a single average-pooled support embedding. We also propose a Transformer Relation Head (TRH), equipped with higher-order representations, which encodes correlations between query regions and the entire support set, while being sensitive to the positional variability of object instances. Our model achieves state-of-the-art results on PASCAL VOC, FSOD, and COCO.



Paperid:822
Authors:Bowen Dong; Pan Zhou; Shuicheng Yan; Wangmeng Zuo
Title: Self-Promoted Supervision for Few-Shot Transformer
Abstract:
The few-shot learning ability of vision transformers (ViTs) is rarely investigated though heavily desired. In this work, we empirically find that with the same few-shot learning frameworks, replacing the widely used CNN feature extractor with a ViT model often severely impairs few-shot classification performance. Moreover, our empirical study shows that in the absence of inductive bias, ViTs often learn the low-qualified token dependencies under few-shot learning regime where only a few labeled training data are available, which largely contributes to the above performance degradation. To alleviate this issue, we propose a simple yet effective few-shot training framework for ViTs, namely Self-promoted sUpervisioN (SUN). Specifically, besides the conventional global supervision for global semantic learning, SUN further pretrains the ViT on the few-shot learning dataset and then uses it to generate individual location-specific supervision for guiding each patch token. This location-specific supervision tells the ViT which patch tokens are similar or dissimilar and thus accelerates token dependency learning. Moreover, it models the local semantics in each patch token to improve the object grounding and recognition capability which helps learn generalizable patterns. To improve the quality of location-specific supervision, we further propose: 1) background patch filtration to filtrate background patches out and assign them into an extra background class; and 2) spatial-consistent augmentation to introduce sufficient diversity for data augmentation while keeping the accuracy of the generated local supervisions. Experimental results show that SUN using ViTs significantly surpasses other few-shot learning frameworks with ViTs and is the first one that achieves higher performance than those CNN state-of-the-arts. Our code is publicly available at https://github.com/DongSky/few-shot-vit.



Paperid:823
Authors:Thanh Nguyen; Chau Pham; Khoi Nguyen; Minh Hoai
Title: Few-Shot Object Counting and Detection
Abstract:
We tackle a new task of few-shot object counting and detection. Given a few exemplar bounding boxes of a target object class, we seek to count and detect all the objects of the target class. This task shares the same supervision as the few-shot counting but additionally outputs the object bounding boxes along with the total object count. To address this challenging problem, we introduce a novel two-stage training strategy and a novel uncertainty-aware few-shot object detector Counting-DETR. The former is aimed at generating pseudo ground truth bounding boxes to train the latter. The latter leverages the pseudo ground-truth provided by the former, but taking the necessary steps to account for the imperfection of pseudo ground truth. To validate the performance of our method on the new task, we introduce a two new datasets named FSCD-147 and FSCD-LVIS. Both datasets contain images with complex scene, multiple object classes per image, and huge variation in object shapes, sizes, and appearance. Our propose approach outperforms very strong baselines adapted from few-shot object counting and few-shot object detection with a large margin in both counting and detection metrics.



Paperid:824
Authors:Kibok Lee; Hao Yang; Satyaki Chakraborty; Zhaowei Cai; Gurumurthy Swaminathan; Avinash Ravichandran; Onkar Dabeer
Title: Rethinking Few-Shot Object Detection on a Multi-Domain Benchmark
Abstract:
Most existing works on few-shot object detection (FSOD) focus on a setting where both pre-training and few-shot learning datasets are from a similar domain. However, few-shot algorithms are important in multiple domains; hence evaluation needs to reflect the broad applications. We propose a Multi-dOmain Few-Shot Object Detection (MoFSOD) benchmark consisting of 10 datasets from a wide range of domains to evaluate FSOD algorithms. We comprehensively analyze the impacts of freezing layers, different architectures, and different pre-training datasets on FSOD performance. Our empirical results show several key factors that have not been explored in previous works: 1) contrary to previous belief, on a multi-domain benchmark, fine-tuning (FT) is a strong baseline for FSOD, performing on par or better than the state-of-the-art (SOTA) algorithms; 2) utilizing FT as the baseline allows us to explore multiple architectures, and we found them to have a significant impact on down-stream few-shot tasks, even with similar pre-training performances; 3) by decoupling pre-training and few-shot learning, MoFSOD allows us to explore the impact of different pre-training datasets, and the right choice can boost the performance of the down-stream tasks significantly. Based on these findings, we list possible avenues of investigation for improving FSOD performance and propose two simple modifications to existing algorithms that lead to SOTA performance on the MoFSOD benchmark. The code is available at https://github.com/amazon-research/few-shot-object-detection-benchmark.



Paperid:825
Authors:Wentao Chen; Zhang Zhang; Wei Wang; Liang Wang; Zilei Wang; Tieniu Tan
Title: Cross-Domain Cross-Set Few-Shot Learning via Learning Compact and Aligned Representations
Abstract:
Few-shot learning (FSL) aims to recognize novel queries with only a few support samples through leveraging prior knowledge from a base dataset. In this paper, we consider the domain shift problem in FSL and aim to address the domain gap between the support set and the query set. Different from previous cross-domain FSL work (CD-FSL) that considers the domain shift between base and novel classes, the new problem, termed cross-domain cross-set FSL (CDSC-FSL), requires few-shot learners not only to adapt to the new domain, but also to be consistent between different domains within each novel class. To this end, we propose a novel approach, namely stabPA, to learn prototypical compact and cross-domain aligned representations, so that the domain shift and few-shot learning can be addressed simultaneously. We evaluate our approach on two new CDCS-FSL benchmarks built from the DomainNet and Office-Home datasets respectively. Remarkably, our approach outperforms multiple elaborated baselines by a large margin, e.g., improving 5-shot accuracy by 6.0 points on average on DomainNet. Code is available at https://github.com/WentaoChen0813/CDCS-FSL"



Paperid:826
Authors:Tianxue Ma; Mingwei Bi; Jian Zhang; Wang Yuan; Zhizhong Zhang; Yuan Xie; Shouhong Ding; Lizhuang Ma
Title: Mutually Reinforcing Structure with Proposal Contrastive Consistency for Few-Shot Object Detection
Abstract:
Few-shot object detection is based on the base set with abundant labeled samples to detect novel categories with scarce samples. The majority of former solutions are mainly based on meta-learning or transfer-learning, neglecting the fact that images from the base set might contain unlabeled novel-class objects, which easily leads to performance degradation and poor plasticity since those novel objects are served as the background. Based on the above phenomena, we propose a Mutually Reinforcing Structure Network (MRSN) to make rational use of unlabeled novel class instances in the base set. In particular, MRSN consists of a mining model which unearths unlabeled novel-class instances and an absorbed model which learns variable knowledge. Then, we design a Proposal Contrastive Consistency (PCC) module in the absorbed model to fully exploit class characteristics and avoid bias from unearthed labels. Furthermore,we propose a simple and effective data synthesis method undirectional-CutMix (UD-CutMix) to improve the robustness of model mining novel class instances, urge the model to pay attention to discriminative parts of objects and eliminate the interference of background information. Extensive experiments illustrate that our proposed approach achieves state-of-the-art results on PASCAL VOC and MS-COCO datasets. Our code will be released at https://github.com/MMatx/MRSN.



Paperid:827
Authors:Huisi Wu; Fangyan Xiao; Chongxin Liang
Title: Dual Contrastive Learning with Anatomical Auxiliary Supervision for Few-Shot Medical Image Segmentation
Abstract:
Few-shot semantic segmentation is a promising solution for scarce data scenarios, especially for medical imaging challenges with limited training data. However, most of the existing few-shot segmentation methods tend to over rely on the images containing target classes, which may hinder its utilization of medical imaging data. In this paper, we present a few-shot segmentation model that employs anatomical auxiliary information from medical images without target classes for dual contrastive learning. The dual contrastive learning module performs comparison among vectors from the perspectives of prototypes and contexts, to enhance the discriminability of learned features and the data utilization. Besides, to distinguish foreground features from background features more friendly, a constrained iterative prediction module is designed to optimize the segmentation of the query image. Experiments on two medical image datasets show that the proposed method achieves performance comparable to state-of-the-art methods.



Paperid:828
Authors:Quentin Bouniot; Ievgen Redko; Romaric Audigier; Angélique Loesch; Amaury Habrard
Title: Improving Few-Shot Learning through Multi-task Representation Learning Theory
Abstract:
In this paper, we consider the framework of multi-task representation (MTR) learning where the goal is to use source tasks to learn a representation that reduces the sample complexity of solving a target task. We start by reviewing recent advances in MTR theory and show that they can provide novel insights for popular meta-learning algorithms when analyzed within this framework. In particular, we highlight a fundamental difference between gradient-based and metric-based algorithms in practice and put forward a theoretical analysis to explain it. Finally, we use the derived insights to improve the performance of meta-learning methods via a new spectral-based regularization term and confirm its efficiency through experimental studies on few-shot classification benchmarks. To the best of our knowledge, this is the first contribution that puts the most recent learning bounds of MTR theory into practice for the task of few-shot classification.



Paperid:829
Authors:Min Zhang; Siteng Huang; Wenbin Li; Donglin Wang
Title: Tree Structure-Aware Few-Shot Image Classification via Hierarchical Aggregation
Abstract:
In this paper, we mainly focus on the problem of how to learn additional feature representations for few-shot image classification through pretext tasks (e.g., rotation or color permutation and so on). This additional knowledge generated by pretext tasks can further improve the performance of few-shot learning (FSL) as it differs from human-annotated supervision (i.e., class labels of FSL tasks). To solve this problem, we present a plug-in Hierarchical Tree Structure-aware (HTS) method, which not only learns the relationship of FSL and pretext tasks, but more importantly, can adaptively select and aggregate feature representations generated by pretext tasks to maximize the performance of FSL tasks. A hierarchical tree constructing component and a gated selection aggregating component is introduced to construct the tree structure and find richer transferable knowledge that can rapidly adapt to novel classes with a few labeled images. Extensive experiments show that our HTS can significantly enhance multiple few-shot methods to achieve new state-of-the-art performance on four benchmark datasets. The code is available at: https://github.com/remiMZ/HTS-ECCV22.



Paperid:830
Authors:Khoi D. Nguyen; Quoc-Huy Tran; Khoi Nguyen; Binh-Son Hua; Rang Nguyen
Title: Inductive and Transductive Few-Shot Video Classification via Appearance and Temporal Alignments
Abstract:
We present a novel method for few-shot video classification, which performs appearance and temporal alignments. In particular, given a pair of query and support videos, we conduct appearance alignment via frame-level feature matching to achieve the appearance similarity score between the videos, while utilizing temporal order-preserving priors for obtaining the temporal similarity score between the videos. Moreover, we introduce a few-shot video classification framework that leverages the above appearance and temporal similarity scores across multiple steps, namely prototype-based training and testing as well as inductive and transductive prototype refinement. To the best of our knowledge, our work is the first to explore transductive few-shot video classification. Extensive experiments on both Kinetics and Something-Something V2 datasets show that both appearance and temporal alignments are crucial for datasets with temporal order sensitivity such as Something-Something V2. Our approach achieves similar or better results than previous methods on both datasets.



Paperid:831
Authors:Otniel-Bogdan Mercea; Thomas Hummel; A. Sophia Koepke; Zeynep Akata
Title: Temporal and Cross-Modal Attention for Audio-Visual Zero-Shot Learning
Abstract:
Audio-visual generalised zero-shot learning for video classification requires understanding the relations between the audio and visual information in order to be able to recognise samples from novel, previously unseen classes at test time. The natural semantic and temporal alignment between audio and visual data in video data can be exploited to learn powerful representations that generalise to unseen classes at test time. We propose a multi-modal and Temporal Cross-attention Framework for audio-visual generalised zero-shot learning. Its inputs are temporally aligned audio and visual features that are obtained from pre-trained networks. Encouraging the framework to focus on cross-modal correspondence across time instead of self-attention within the modalities boosts the performance significantly. We show that our proposed framework that ingests temporal features yields state-of-the-art performance on the UCF-GZSL, VGGSound-GZSL, and ActivityNet-GZSL benchmarks for (generalised) zero-shot learning.



Paperid:832
Authors:Seonghyeon Moon; Samuel S. Sohn; Honglu Zhou; Sejong Yoon; Vladimir Pavlovic; Muhammad Haris Khan; Mubbasir Kapadia
Title: HM: Hybrid Masking for Few-Shot Segmentation
Abstract:
We study few-shot semantic segmentation that aims to segment a target object from a query image when provided with a few annotated support images of the target class. Several recent methods resort to a feature masking (FM) technique to discard irrelevant feature activations which eventually facilitates the reliable prediction of segmentation mask. A fundamental limitation of FM is the inability to preserve the fine-grained spatial details that affect the accuracy of segmentation mask, especially for small target objects. In this paper, we develop a simple, effective, and efficient approach to enhance feature masking (FM). We dub the enhanced FM as hybrid masking (HM). Specifically, we compensate for the loss of fine-grained spatial details in FM technique by investigating and leveraging a complementary basic input masking method. Experiments have been conducted on three publicly available benchmarks with strong few-shot segmentation (FSS) baselines. We empirically show improved performance against the current state-of-the-art methods by visible margins across different benchmarks. Our code and trained models are available at: https://github.com/moonsh/HM-Hybrid-Masking"



Paperid:833
Authors:Haoquan Li; Laoming Zhang; Daoan Zhang; Lang Fu; Peng Yang; Jianguo Zhang
Title: TransVLAD: Focusing on Locally Aggregated Descriptors for Few-Shot Learning
Abstract:
This paper presents a transformer framework for few-shot learning, termed TransVLAD, with one focus showing the power of locally aggregated descriptors for few-shot learning. Our TransVLAD model is simple: a standard transformer encoder following a NeXtVLAD aggregation module to output the locally aggregated descriptors. In contrast to the prevailing use of CNN as part of the feature extractor, we are the first to prove self-supervised learning like masked autoencoders (MAE) can deal with the overfitting of transformers in few-shot image classification. Besides, few-shot learning can benefit from this general-purpose pre-training. Then, we propose two methods to mitigate few-shot biases, supervision bias and simple-characteristic bias. The first method is introducing masking operation into fine-tuning, by which we accelerate fine-tuning (by more than 3x) and improve accuracy. The second one is adapting focal loss into soft focal loss to focus on hard characteristics learning. Our TransVLAD finally tops 10 benchmarks on five popular few-shot datasets by an average of more than 2%.



Paperid:834
Authors:Tao Zhang; Wu Huang
Title: Kernel Relative-Prototype Spectral Filtering for Few-Shot Learning
Abstract:
Few-shot learning performs classification tasks and regression tasks on scarce samples. As one of the most representative few-shot learning models, Prototypical Network represents each class as sample average, or a prototype, and measures the similarity of samples and prototypes by Euclidean distance. In this paper, we propose a framework of spectral filtering (shrinkage) for measuring the difference between query samples and prototypes, or namely the relative prototypes, in a reproducing kernel Hilbert space (RKHS). In this framework, we further propose a method utilizing Tikhonov regularization as the filter function for few-shot classification. We conduct several experiments to verify our method utilizing different kernels based on the miniImageNet dataset, tiered-ImageNet dataset and CIFAR-FS dataset. The experimental results show that the proposed model can perform the state-of-the-art. In addition, the experimental results show that the proposed shrinkage method can boost the performance. Source code is available at https://github.com/zhangtao2022/DSFN.



Paperid:835
Authors:Niv Cohen; Rinon Gal; Eli A. Meirom; Gal Chechik; Yuval Atzmon
Title: "“This Is My Unicorn, Fluffy”: Personalizing Frozen Vision-Language Representations"
Abstract:
Large Vision & Language models pretrained on web-scale data provide representations that are invaluable for numerous V&L problems. However, it is unclear how they can be extended to reason about user-specific visual concepts in unstructured language. This problem arises in multiple domains, from personalized image retrieval to personalized interaction with smart devices. We introduce a new learning setup called Personalized Vision & Language (PerVL) with two new benchmark datasets for retrieving and segmenting user-specific (""""personalized"""") concepts “in the wild"""". In PerVL, one should learn personalized concepts (1) independently of the downstream task (2) allowing a pretrained model to reason about them with free language, and (3) without providing personalized negative examples. We propose an architecture for solving PerVL that operates by expanding the input vocabulary of a pretrained model with new word embeddings for the personalized concepts. The model can then simply employ them as part of a sentence. We demonstrate that our approach learns personalized visual concepts from a few examples and effectively applies them in image retrieval and semantic segmentation using rich textual queries. For example the model improves MRR by 51.1% (28.4% vs 18.8%) compared to the strongest baseline.



Paperid:836
Authors:Zixuan Zhou; Xuefei Ning; Yi Cai; Jiashu Han; Yiping Deng; Yuhan Dong; Huazhong Yang; Yu Wang
Title: CLOSE: Curriculum Learning on the Sharing Extent towards Better One-Shot NAS
Abstract:
One-shot Neural Architecture Search (NAS) has been widely used to discover architectures due to its efficiency. However, previous studies reveal that one-shot performance estimations of architectures might not be well correlated with their performances in stand-alone training because of the excessive sharing of operation parameters (i.e., large sharing extent) between architectures. Thus, recent methods construct even more over-parameterized supernets to reduce the sharing extent. But these improved methods introduce a large number of extra parameters and thus cause an undesirable trade-off between the training costs and the ranking quality. To alleviate the above issues, we propose to apply Curriculum Learning On Sharing Extent (CLOSE) to train the supernet both efficiently and effectively. Specifically, we train the supernet with a large sharing extent (an easier curriculum) at the beginning and gradually decrease the sharing extent of the supernet (a harder curriculum). To support this training strategy, we design a novel supernet (CLOSENet) that decouples the parameters from operations to realize a flexible sharing scheme and adjustable sharing extent. Extensive experiments demonstrate that CLOSE can obtain a better ranking quality across different computational budget constraints than other one-shot supernets, and is able to discover superior architectures when combined with various search strategies. Code is available at https://github.com/walkerning/aw_nas.



Paperid:837
Authors:Junwoo Cho; Seungtae Nam; Daniel Rho; Jong Hwan Ko; Eunbyung Park
Title: Streamable Neural Fields
Abstract:
Neural fields have emerged as a new data representation paradigm and have shown remarkable success in various signal representations. Since they preserve signals in their network parameters, the data transfer by sending and receiving the entire model parameters prevents this emerging technology from being used in many practical scenarios. We propose streamable neural fields, a single model that consists of executable sub-networks of various widths. The proposed architectural and training techniques enable a single network to be streamable over time and reconstruct different qualities and parts of signals. For example, a smaller sub-network produces smooth and low-frequency signals, while a larger sub-network can represent fine details. Experimental results have shown the effectiveness of our method in various domains, such as 2D images, videos, and 3D signed distance functions. Finally, we demonstrate that our proposed method improves training stability, by exploiting parameter sharing.



Paperid:838
Authors:Julia Hornauer; Vasileios Belagiannis
Title: Gradient-Based Uncertainty for Monocular Depth Estimation
Abstract:
In monocular depth estimation, disturbances in the image context, like moving objects or reflecting materials, can easily lead to erroneous predictions. For that reason, uncertainty estimates for each pixel are necessary, in particular for safety-critical applications such as automated driving. We propose a post hoc uncertainty estimation approach for an already trained and thus fixed depth estimation model, represented by a deep neural network. The uncertainty is estimated with the gradients which are extracted with an auxiliary loss function. To avoid relying on ground-truth information for the loss definition, we present an auxiliary loss function based on the correspondence of the depth prediction for an image and its horizontally flipped counterpart. Our approach achieves state-of-the-art uncertainty estimation results on the KITTI and NYU Depth V2 benchmarks without the need to retrain the neural network. Models and code are publicly available at https://github.com/jhornauer/GrUMoDepth.



Paperid:839
Authors:Zhen Wang; Liu Liu; Yajing Kong; Jiaxian Guo; Dacheng Tao
Title: Online Continual Learning with Contrastive Vision Transformer
Abstract:
Online continual learning (online CL) studies the problem of learning sequential tasks from an online data stream without task boundaries, aiming to adapt to new data while alleviating catastrophic forgetting on the past tasks. This paper proposes a framework Contrastive Vision Transformer (CVT), which designs a focal contrastive learning strategy based on a transformer architecture, to achieve a better stability-plasticity trade-off for online CL. Specifically, we design a new external attention mechanism for online CL that implicitly captures previous tasks’ information. Besides, CVT contains learnable focuses for each class, which could accumulate the knowledge of previous classes to alleviate forgetting. Based on the learnable focuses, we design a focal contrastive loss to rebalance contrastive learning between new and past classes and consolidate previously learned representations. Moreover, CVT contains a dual-classifier structure for decoupling learning current classes and balancing all observed classes. The extensive experimental results show that our approach achieves state-of-the-art performance with even fewer parameters on online CL benchmarks and effectively alleviates the catastrophic forgetting.



Paperid:840
Authors:Taeho Kim; Yongin Kwon; Jemin Lee; Taeho Kim; Sangtae Ha
Title: CPrune: Compiler-Informed Model Pruning for Efficient Target-Aware DNN Execution
Abstract:
Mobile devices run deep learning models for various purposes, such as image classification and speech recognition. Due to the resource constraints of mobile devices, researchers have focused on either making a lightweight deep neural network (DNN) model using model pruning or generating an efficient code using compiler optimization. We found that the straightforward integration between model compression and compiler auto-tuning often does not produce the most efficient model for a target device. We propose CPrune, a compiler-informed model pruning for efficient target-aware DNN execution to support an application with a required target accuracy. CPrune makes a lightweight DNN model through informed pruning based on the structural information of programs built during the compiler tuning process. Our experimental results show that CPrune increases the DNN execution speed up to 2.73x compared to the state-of-the-art TVM auto-tune while satisfying the accuracy requirement.



Paperid:841
Authors:Xiaoxing Wang; Jiale Lin; Juanping Zhao; Xiaokang Yang; Junchi Yan
Title: EAutoDet: Efficient Architecture Search for Object Detection
Abstract:
Training CNN for detection is time-consuming due to the large dataset and complex network modules, making it hard to search architectures on detection datasets directly, which usually requires vast search costs (usually tens and even hundreds of GPU-days). In contrast, this paper introduces an efficient framework, named EAutoDet, that can discover practical backbone and FPN architectures for object detection in 1.4 GPU-days. Specifically, we construct a supernet for both backbone and FPN modules and adopt the differentiable method. To reduce the GPU memory requirement and computational cost, we propose a kernel reusing technique by sharing the weights of candidate operations on one edge and consolidating them into one convolution. A dynamic channel refinement strategy is also introduced to search channel numbers. Extensive experiments show significant efficacy and efficiency of our method. In particular, the discovered architectures surpass state-of-the-art object detection NAS methods and achieve 40.1 mAP with 120 FPS and 49.2 mAP with 41.3 FPS on COCO test-dev set. We also transfer the discovered architectures to rotation detection task, which achieve 77.05 mAP on DOTA-v1.0 test set with 21.1M parameters. The code is publicly available at https://github.com/vicFigure/EAutoDet.



Paperid:842
Authors:Chao Xue; Xiaoxing Wang; Junchi Yan; Chun-Guang Li
Title: A Max-Flow Based Approach for Neural Architecture Search
Abstract:
Neural Architecture Search (NAS) aims to automatically produce network architectures suitable to specific tasks on given datasets. Unlike previous NAS strategies based on reinforcement learning, genetic algorithm, Bayesian optimization, and differential programming method, we formulate the NAS task as a Max-Flow problem on search space consisting of Directed Acyclic Graph (DAG) and thus propose a novel NAS approach, called MF-NAS, which defines the search space and designs the search strategy in a fully graphic manner. In MF-NAS, parallel edges with capacities are induced by combining different operations, including skip connection, convolutions, and pooling, and the weights and capacities of the parallel edges are updated iteratively during the search process. Moreover, we interpret MF-NAS from the perspective of nonparametric density estimation and show the relationship between the flow of a graph and the corresponding classification accuracy of neural network architecture. We evaluate the competitive efficacy of our proposed MF-NAS across different datasets with different search spaces that are used in DARTS/ENAS and NAS-Bench-201.



Paperid:843
Authors:Robik Shrestha; Kushal Kafle; Christopher Kanan
Title: OccamNets: Mitigating Dataset Bias by Favoring Simpler Hypotheses
Abstract:
Dataset bias and spurious correlations can significantly impair generalization in deep neural networks. Many prior efforts have addressed this problem using either alternative loss functions or sampling strategies that focus on rare patterns. We propose a new direction: modifying the network architecture to impose inductive biases that make the network robust to dataset bias. Specifically, we propose OccamNets, which are biased to favor simpler solutions by design. OccamNets have two inductive biases. First, they are biased to use as little network depth as needed for an individual example. Second, they are biased toward using fewer image locations for prediction. While OccamNets are biased toward simpler hypotheses, they can learn more complex hypotheses if necessary. In experiments, OccamNets outperform or rival state-of-the-art methods using architectures that do not incorporate these inductive biases. Furthermore, we demonstrate that when these methods are combined with OccamNets results further improve.



Paperid:844
Authors:Martin Trimmel; Mihai Zanfir; Richard Hartley; Cristian Sminchisescu
Title: ERA: Enhanced Rational Activations
Abstract:
Activation functions play a central role in deep learning since they form an essential building stone of neural networks. In the last few years, the focus has been shifting towards investigating new types of activations that outperform the classical Rectified Linear Unit (ReLU) in modern neural architectures. Most recently, rational activation functions (RAFs) have awakened interest because they were shown to perform on par with state-of-the-art activations on image classification. Despite their apparent potential, prior formulations are either not safe, not smooth or not ""true"" rational functions, and they only work with careful initialisation. Aiming to mitigate these issues, we propose a novel, enhanced rational function, ERA, and investigate how to better accommodate the specific needs of these activations, to both network components and training regime. In addition to being more stable, the proposed function outperforms other standard ones across a range of lightweight network architectures on two different tasks: image classification and 3d human pose and shape reconstruction.



Paperid:845
Authors:Cong Wang; Hongmin Xu; Xiong Zhang; Li Wang; Zhitong Zheng; Haifeng Liu
Title: Convolutional Embedding Makes Hierarchical Vision Transformer Stronger
Abstract:
Vision Transformers (ViTs) have recently dominated a range of computer vision tasks, yet it suffers from low training data efficiency and inferior local semantic representation capability without appropriate inductive bias. Convolutional neural networks (CNNs) inherently capture regional-aware semantics, inspiring researchers to introduce CNNs back into the architecture of the ViTs to provide desirable inductive bias for ViTs. However, is the locality achieved by the micro-level CNNs embedded in ViTs good enough? In this paper, we investigate the problem by profoundly exploring how the macro architecture of the hybrid CNNs/ViTs enhances the performances of hierarchical ViTs. Particularly, we study the role of token embedding layers, alias convolutional embedding (CE), and systemically reveal how CE injects desirable inductive bias in ViTs. Besides, we apply the optimal CE configuration to 4 recently released state-of-the-art ViTs, effectively boosting the corresponding performances. Finally, a family of efficient hybrid CNNs/ViTs, dubbed CETNets, are released, which may serve as generic vision backbones. Specifically, CETNets achieve 84.9% Top-1 accuracy on ImageNet-1K (training from scratch), 48.6% box mAP on the COCO benchmark, and 51.6% mIoU on the ADE20K, substantially improving the performances of the corresponding state-of-the-art baselines.



Paperid:846
Authors:Kwang In Kim
Title: Active Label Correction Using Robust Parameter Update and Entropy Propagation
Abstract:
Label noise is prevalent in real-world visual learning applications and correcting all label mistakes can be prohibitively costly. Training neural network classifiers on such noisy datasets may lead to significant performance degeneration. Active label correction (ALC) attempts to minimize the re-labeling costs by identifying examples for which providing correct labels will yield maximal performance improvements. Existing ALC approaches typically select the examples that the classifier is least confident about (\eg with the largest entropies). However, such confidence estimates can be unreliable as the classifier itself is initially trained on noisy data. Also, naively selecting a batch of low confidence examples can result in redundant labeling of spatially adjacent examples. We present a new ALC algorithm that addresses these challenges. Our algorithm robustly estimates label confidence values by regulating the contributions of individual examples in the parameter update of the network. Further, our algorithm avoids redundant labeling by promoting diversity in batch selection through propagating the confidence of each newly labeled example to the entire dataset. Experiments involving four benchmark datasets and two types of label noise demonstrate that our algorithm offers a significant improvement in re-labeling efficiency over state-of-the-art ALC approaches.



Paperid:847
Authors:Justin Theiss; Jay Leverett; Daeil Kim; Aayush Prakash
Title: Unpaired Image Translation via Vector Symbolic Architectures
Abstract:
Image-to-image translation has played an important role in enabling synthetic data for computer vision. However, if the source and target domains have a large semantic mismatch, existing techniques often suffer from source content corruption aka semantic flipping. To address this problem, we propose a new paradigm for image-to-image translation using Vector Symbolic Architectures (VSA), a theoretical framework which defines algebraic operations in a high-dimensional vector (hypervector) space. We introduce VSA-based constraints on adversarial learning for source-to-target translations by learning a hypervector mapping that inverts the translation to ensure consistency with source content. We show both qualitatively and quantitatively that our method improves over other state-of-the-art techniques.



Paperid:848
Authors:Jihao Liu; Xin Huang; Guanglu Song; Hongsheng Li; Yu Liu
Title: "UniNet: Unified Architecture Search with Convolution, Transformer, and MLP"
Abstract:
Recently, transformer and multi-layer perceptron (MLP) architectures have achieved impressive results on various vision tasks. However, how to effectively combine those operators to form high-performance hybrid visual architectures still remains a challenge. In this work, we study the learnable combination of convolution, transformer, and MLP by proposing a novel unified architecture search approach. Our approach contains two key designs to achieve the search for high-performance networks. First, we model the very different searchable operators in a unified form and thus enable the operators to be characterized with the same set of configuration parameters. In this way, the overall search space size is significantly reduced, and the total search cost becomes affordable. Second, we propose context-aware downsampling modules (DSMs) to mitigate the gap between the different types of operators. Our proposed DSMs are able to better adapt features from different types of operators, which is important for identifying high-performance hybrid architectures. Finally, we integrate configurable operators and DSMs into a unified search space and search with a Reinforcement Learning-based search algorithm to fully explore the optimal combination of the operators. To this end, we search a baseline network and scale it up to obtain a family of models, named UniNets, which achieve much better accuracy and efficiency than previous ConvNets and Transformers. In particular, our UniNet-B5 achieves 84.9% top-1 accuracy on ImageNet, outperforming EfficientNet-B7 and BoTNet-T7 with 44% and 55% fewer FLOPs respectively. By pretraining on the ImageNet-21K, our UniNet-B6 achieves 87.4%, outperforming Swin-L with 51% fewer FLOPs and 41% fewer parameters.



Paperid:849
Authors:Yongming Rao; Wenliang Zhao; Jie Zhou; Jiwen Lu
Title: AMixer: Adaptive Weight Mixing for Self-Attention Free Vision Transformers
Abstract:
Vision Transformers have shown state-of-the-art results for various visual recognition tasks. The dot-product self-attention mechanism that replaces convolution to mix spatial information is commonly recognized as the indispensable ingredient behind the success of vision Transformers. In this paper, we thoroughly investigate the key differences between vision Transformers and recent all-MLP models. Our empirical results show the superiority of vision Transformers mainly comes from the data-dependent token mixing strategy and the multi-head scheme instead of query-key interactions. Inspired by this observation, we propose a computationally and parametrically efficient operation named adaptive weight mixing to generate attention weights without token-token interactions. Based on this operation, we develop a new architecture named as AMixer to capture both long-term and short-term spatial dependencies without self-attention. Extensive experiments demonstrate that our adaptive weight mixing is more efficient and effective than previous weight generation methods and our AMixer can achieve a better trade-off between accuracy and complexity than vision Transformers and MLP models on both ImageNet and downstream tasks.



Paperid:850
Authors:Kan Wu; Jinnian Zhang; Houwen Peng; Mengchen Liu; Bin Xiao; Jianlong Fu; Lu Yuan
Title: TinyViT: Fast Pretraining Distillation for Small Vision Transformers
Abstract:
Vision transformer (ViT) recently has drawn great attention in computer vision due to its remarkable model capability. However, most prevailing ViT models suffer from huge number of parameters, restricting their applicability on devices with limited resources. To alleviate this issue, we propose TinyViT, a new family of tiny and efficient small vision transformers pretrained on large-scale datasets with our proposed fast distillation framework. The central idea is to transfer knowledge from large pretrained models to small ones, while enabling small models to get the dividends of massive pretraining data. More specifically, we apply distillation during pretraining for knowledge transfer. The logits of large teacher models are sparsified and stored in disk in advance to save the memory cost and computation overheads. The tiny student transformers are automatically scaled down from a large pretrained model with computation and parameter constraints. Comprehensive experiments demonstrate the efficacy of TinyViT. It achieves a top-1 accuracy of 84.8% on ImageNet-1k with only 21M parameters, being comparable to Swin-B pretrained on ImageNet-21k while using 4.2 times fewer parameters. Moreover, increasing image resolutions, TinyViT can reach 86.5% accuracy, being slightly better than Swin-L while using only 11% parameters. Last but not the least, we demonstrate a good transfer ability of TinyViT on various downstream tasks. Code and models are available at https://github.com/microsoft/Cream/tree/main/TinyViT.



Paperid:851
Authors:Jinwoo Kim; Saeyoon Oh; Sungjun Cho; Seunghoon Hong
Title: Equivariant Hypergraph Neural Networks
Abstract:
Many problems in computer vision and machine learning can be cast as learning on hypergraphs that represent higher-order relations. Recent approaches for hypergraph learning extend graph neural networks based on message passing, which is simple yet fundamentally limited in modeling long-range dependencies and expressive power. On the other hand, tensor-based equivariant neural networks enjoy maximal expressiveness, but their application has been limited in hypergraphs due to heavy computation and strict assumptions on fixed-order hyperedges. We resolve these problems and present Equivariant Hypergraph Neural Network (EHNN), the first attempt to realize maximally expressive equivariant layers for general hypergraph learning. We also present two practical realizations of our framework based on hypernetworks (EHNN-MLP) and self-attention (EHNN-Transformer), which are easy to implement and theoretically more expressive than most message passing approaches. We demonstrate their capability in a range of hypergraph learning problems, including synthetic k-edge identification, semi-supervised classification, and visual keypoint matching, and report improved performances over strong message passing baselines. Our implementation is available at https://github.com/jw9730/ehnn.



Paperid:852
Authors:Jiyang Xie; Xiu Su; Shan You; Zhanyu Ma; Fei Wang; Chen Qian
Title: ScaleNet: Searching for the Model to Scale
Abstract:
Recently, community has paid increasing attention on model scaling and contributed to developing a model family with a wide spectrum of scales. Current methods either simply resort to a one-shot NAS manner to construct a non-structural and non-scalable model family or rely on a manual yet fixed scaling strategy to scale an unnecessarily best base model. In this paper, we bridge both two components and propose ScaleNet to jointly search base model and scaling strategy so that the scaled large model can have more promising performance. Concretely, we design a super-supernet to embody models with different spectrum of sizes (e.g., FLOPs). Then, the scaling strategy can be learned interactively with the base model via a Markov chain-based evolution algorithm and generalized to develop even larger models. To obtain a decent super-supernet, we design a hierarchical sampling strategy to enhance its training sufficiency and alleviate the disturbance. Experimental results show our scaled networks enjoy significant performance superiority on various FLOPs, but with at least 2.53x reduction on search cost.



Paperid:853
Authors:Vincent Le Guen; Clément Rambour; Nicolas Thome
Title: Complementing Brightness Constancy with Deep Networks for Optical Flow Prediction
Abstract:
State-of-the-art methods for optical flow estimation rely on deep learning, which require complex sequential training schemes to reach optimal performances on real-world data. In this work, we introduce the COMBO deep network that explicitly exploits the brightness constancy (BC) model used in traditional methods. Since State-of-the-art methods for optical flow estimation rely on deep learning, which require complex sequential training schemes to reach optimal performances on real-world data. In this work, we introduce the COMBO deep network that explicitly exploits the brightness constancy (BC) model used in traditional methods. Since BC is an approximate physical model violated in several situations, we propose to train a physically-constrained network complemented with a data-driven network. We introduce a unique and meaningful flow decomposition between the physical prior and the data-driven complement, including an uncertainty quantification of the BC model. We derive a training scheme for learning the different components of the decomposition, in a supervised but also in a semi-supervised context. Experiments show that COMBO can improve performances over state-of-the-art supervised networks, eg RAFT, reaching state-of-the-art performances on several benchmarks. We highlight how COMBO can leverage the BC model and adapt to its limitations. Finally, we show that our semi-supervised method can significantly simplify the training procedure.



Paperid:854
Authors:Xiu Su; Shan You; Jiyang Xie; Mingkai Zheng; Fei Wang; Chen Qian; Changshui Zhang; Xiaogang Wang; Chang Xu
Title: ViTAS: Vision Transformer Architecture Search
Abstract:
Vision transformers (ViTs) inherited the success of NLP but their structures have not been sufficiently investigated and optimized for visual tasks. One of the simplest solutions is to directly search the optimal one via the widely used neural architecture search (NAS) in CNNs. However, we empirically find this straightforward adaptation would encounter catastrophic failures and be frustratingly unstable for the training of superformer. In this paper, we argue that since ViTs mainly operate on token embeddings with little inductive bias, imbalance of channels for different architectures would worsen the weight-sharing assumption and cause the training instability as a result. Therefore, we develop a new cyclic weight-sharing mechanism for token embeddings of the ViTs, which enables each channel could more evenly contribute to all candidate architectures. Besides, we also propose identity shifting to alleviate the many-to-one issue in superformer and leverage weak augmentation and regularization techniques for more steady training empirically. Based on these, our proposed method, ViTAS, has achieved significant superiority in both DeiT- and Twins-based ViTs. For example, with only $1.4$G FLOPs budget, our searched architecture has $3.3\%$ ImageNet-$1$k accuracy than the baseline DeiT. With $3.0$G FLOPs, our results achieve $82.0\%$ accuracy on ImageNet-$1$k, and $45.9\%$ mAP on COCO$2017$ which is $2.4\%$ superior than other ViTs.



Paperid:855
Authors:Chenxi Liu; Zhaoqi Leng; Pei Sun; Shuyang Cheng; Charles R. Qi; Yin Zhou; Mingxing Tan; Dragomir Anguelov
Title: LidarNAS: Unifying and Searching Neural Architectures for 3D Point Clouds
Abstract:
Developing neural models that accurately understand objects in 3D point clouds is essential for the success of robotics and autonomous driving. However, arguably due to the higher-dimensional nature of the data (as compared to images), existing neural architectures exhibit a large variety in their designs, including but not limited to the views considered, the format of the neural features, and the neural operations used. Lack of a unified framework and interpretation makes it hard to put these designs in perspective, as well as systematically explore new ones. In this paper, we begin by proposing a unified framework of such, with the key idea being factorizing the neural networks into a series of view transforms and neural layers. We demonstrate that this modular framework can reproduce a variety of existing works while allowing a fair comparison of backbone designs. Then, we show how this framework can easily materialize into a concrete neural architecture search (NAS) space, allowing a principled NAS-for-3D exploration. In performing evolutionary NAS on the 3D object detection task on the Waymo Open Dataset, not only do we outperform the state-of-the-art models, but also report the interesting finding that NAS tends to discover the same macro-level architecture concept for both the vehicle and pedestrian classes.



Paperid:856
Authors:Lei Wang; Piotr Koniusz
Title: Uncertainty-DTW for Time Series and Sequences
Abstract:
Dynamic Time Warping (DTW) is used for matching pairs of sequences and celebrated in applications such as forecasting the evolution of time series, clustering time series or even matching sequence pairs in few-shot action recognition. The transportation plan of DTW contains a set of paths; each path matches frames between two sequences under a varying degree of time warping, to account for varying temporal intra-class dynamics of actions. However, as DTW is the smallest distance among all paths, it may be affected by the feature uncertainty which varies across time steps/frames. Thus, in this paper, we propose to model the so-called aleatoric uncertainty of a differentiable (soft) version of DTW. To this end, we model the heteroscedastic aleatoric uncertainty of each path by the product of likelihoods from Normal distributions, each capturing variance of pair of frames. (The path distance is the sum of base distances between features of pairs of frames of the path.) The Maximum Likelihood Estimation (MLE) applied to a path yields two terms: (i) a sum of Euclidean distances weighted by the variance inverse, and (ii) a sum of log-variance regularization terms. Thus, our uncertainty-DTW is the smallest weighted path distance among all paths, and the regularization term (penalty for the high uncertainty) is the aggregate of log-variances along the path. The distance and the regularization term can be used in various objectives. We showcase forecasting the evolution of time series, estimating the Fréchet mean of time series, and supervised/unsupervised few-shot action recognition of the articulated human 3D body joints.



Paperid:857
Authors:Dang Nguyen; Sunil Gupta; Kien Do; Svetha Venkatesh
Title: Black-Box Few-Shot Knowledge Distillation
Abstract:
Knowledge distillation (KD) is an efficient approach to transfer the knowledge from a large “teacher” network to a smaller “student” network. Traditional KD methods require lots of labeled training samples and a white-box teacher (parameters are accessible) to train a good student. However, these resources are not always available in real-world applications. The distillation process often happens at an external party side where we do not have access to much data, and the teacher does not disclose its parameters due to security and privacy concerns. To overcome these challenges, we propose a black-box few-shot KD method to train the student with few unlabeled training samples and a black-box teacher. Our main idea is to expand the training set by generating a diverse set of out-of-distribution synthetic images using MixUp and a conditional variational auto-encoder. These synthetic images along with their labels obtained from the teacher are used to train the student. We conduct extensive experiments to show that our method significantly outperforms recent SOTA few/zero-shot KD methods on image classification tasks.



Paperid:858
Authors:Jim Davis; Logan Frank
Title: Revisiting Batch Norm Initialization
Abstract:
Batch normalization (BN) is comprised of a normalization component followed by an affine transformation and has become essential for training deep neural networks. Standard initialization of each BN in a network sets the affine transformation scale and shift to 1 and 0, respectively. However, after training we have observed that these parameters do not alter much from their initialization. Furthermore, we have noticed that the normalization process can still yield overly large values, which is undesirable for training. We revisit the BN formulation and present a new initialization method and update approach for BN to address the aforementioned issues. Experiments are designed to emphasize and demonstrate the positive influence of proper BN scale initialization on performance, and use rigorous statistical significance tests for evaluation. The approach can be used with existing implementations at no additional computational cost. Source code is available at https://github.com/osu-cvl/revisiting-bn-init.



Paperid:859
Authors:Ho Man Kwan; Shenghui Song
Title: SSBNet: Improving Visual Recognition Efficiency by Adaptive Sampling
Abstract:
Downsampling is widely adopted to achieve a good trade-off between accuracy and latency for visual recognition. However, the commonly used pooling layers are not learned, which causes possible loss of information. As another dimension reduction method, adaptive sampling weights and processes regions that are relevant to the task, which can better preserve useful information. However, the use of adaptive sampling has been limited to certain layers. In this paper, we show that using adaptive sampling as the main component in a deep neural network can improve network efficiency. In particular, we propose SSBNet which is built by inserting sampling layers into existing networks like ResNet. The proposed SSBNet achieved competitive results in the ImageNet and COCO datasets. For example, the SSB-ResNet-RS-200 achieved 82.6% accuracy in the ImageNet dataset, which is 0.6% higher than the baseline ResNet-RS-152 with similar complexity. Visualization shows the advantage of SSBNet in allowing different layers to focus on different positions, and ablation studies further validate the advantage of adaptive sampling over uniform methods.



Paperid:860
Authors:Zhiqiang He; Yaguan Qian; Yuqi Wang; Bin Wang; Xiaohui Guan; Zhaoquan Gu; Xiang Ling; Shaoning Zeng; Haijiang Wang; Wujie Zhou
Title: Filter Pruning via Feature Discrimination in Deep Neural Networks
Abstract:
Filter pruning is one of the most effective methods to compress deep convolutional networks (CNNs). In this paper, as a key component in filter pruning, We first propose a feature discrimination based filter importance criterion, namely Receptive Field Criterion (RFC). It turns the maximum activation responses that characterize the receptive field into probabilities, then measure the filter importance by the distribution of these probabilities from a new perspective of feature discrimination. However, directly applying RFC to global threshold pruning may lead to some problems, because global threshold pruning neglects the differences between different layers. Hence, we propose Distinguishing Layer Pruning based on RFC (DLRFC), i.e., discriminately prune the filters in different layers, which avoids measuring filters between different layers directly against filter criteria. Specifically, our method first selects relatively redundant layers by hard and soft changes of the network output, and then prunes only at these layers. The whole process dynamically adjusts redundant layers through iterations. Extensive experiments conducted on CIFAR-10/100 and ImageNet show that our method achieves state-of-the-art performance in several benchmarks.



Paperid:861
Authors:Mingjun Zhao; Shan Lu; Zixuan Wang; Xiaoli Wang; Di Niu
Title: LA3: Efficient Label-Aware AutoAugment
Abstract:
Automated augmentation is an emerging and effective technique to search for data augmentation policies to improve generalizability of deep neural network training. Most existing work focuses on constructing a unified policy applicable to all data samples in a given dataset, without considering sample or class variations. In this paper, we propose a novel two-stage data augmentation algorithm, named Label-Aware AutoAugment (LA3), which takes advantage of the label information, and learns augmentation policies separately for samples of different labels. LA3 consists of two learning stages, where in the first stage, individual augmentation methods are evaluated and ranked for each label via Bayesian Optimization aided by a neural predictor, which allows us to identify effective augmentation techniques for each label under a low search cost. And in the second stage, a composite augmentation policy is constructed out of a selection of effective as well as complementary augmentations, which produces significant performance boost and can be easily deployed in typical model training. Extensive experiments demonstrate that LA3 achieves excellent performance matching or surpassing existing methods on CIFAR-10 and CIFAR-100, and achieves a new state-of-the-art ImageNet accuracy of 79.97% on ResNet-50 among auto-augmentation methods, while maintaining a low computational cost.



Paperid:862
Authors:Alireza Ganjdanesh; Shangqian Gao; Heng Huang
Title: Interpretations Steered Network Pruning via Amortized Inferred Saliency Maps
Abstract:
Convolutional Neural Networks (CNNs) compression is crucial to deploying these models in edge devices with limited resources. Existing channel pruning algorithms for CNNs have achieved plenty of success on complex models. They approach the pruning problem from various perspectives and use different metrics to guide the pruning process. However, these metrics mainly focus on the model’s ‘outputs’ or ‘weights’ and neglect its ‘interpretations’ information. To fill in this gap, we propose to address the channel pruning problem from a novel perspective by leveraging the interpretations of a model to steer the pruning process, thereby utilizing information from both inputs and outputs of the model. However, existing interpretation methods cannot get deployed to achieve our goal as either they are inefficient for pruning or may predict non-coherent explanations. We tackle this challenge by introducing a selector model that predicts real-time smooth saliency masks for pruned models. We parameterize the distribution of explanatory masks by Radial Basis Function (RBF)-like functions to incorporate geometric prior of natural images in our selector model’s inductive bias. Thus, we can obtain compact representations of explanations to reduce the computational costs of our pruning method. We leverage our selector model to steer the network pruning by maximizing the similarity of explanatory representations for the pruned and original models. Extensive experiments on CIFAR-10 and ImageNet benchmark datasets demonstrate the efficacy of our proposed method. Our implementations are available at \url{https://github.com/Alii-Ganjj/InterpretationsSteeredPruning}"



Paperid:863
Authors:Yue Zhao; Junzhou Chen; Zirui Zhang; Ronghui Zhang
Title: BA-Net: Bridge Attention for Deep Convolutional Neural Networks
Abstract:
In attention mechanism research, most existing methods are hard to utilize well the information of the neural network with high computing efficiency due to heavy feature compression in the attention layer. This paper proposes a simple and general approach named Bridge Attention to address this issue. As a new idea, BA-Net straightforwardly integrates features from previous layers and effectively promotes information interchange. Only simple strategies are employed for the model implementation, similar to the SENet. Moreover, after extensively investigating the effectiveness of different previous features, we discovered a simple and exciting insight that bridging all the convolution outputs inside each block with BN can obtain better attention to enhance the performance of neural networks. BA-Net is effective, stable, and easy to use. A comprehensive evaluation of computer vision tasks demonstrates that the proposed approach achieves better performance than the existing channel attention methods regarding accuracy and computing efficiency.



Paperid:864
Authors:Koushik Biswas; Sandeep Kumar; Shilpak Banerjee; Ashish Kumar Pandey
Title: SAU: Smooth Activation Function Using Convolution with Approximate Identities
Abstract:
Well-known activation functions like ReLU or Leaky ReLU are non-differentiable at the origin. Over the years, many smooth approximations of ReLU have been proposed using various smoothing techniques. We propose new smooth approximations of a non-differentiable activation function by convolving it with approximate identities. In particular, we present smooth approximations of Leaky ReLU and show that they outperform several well-known activation functions in various datasets and models. We call this function Smooth Activation Unit (SAU). Replacing ReLU by SAU, we get 5.63%, 2.95%, and 2.50% improvement with ShuffleNet V2 (2.0x), PreActResNet 50 and ResNet 50 models respectively on the CIFAR100 dataset and 2.31% improvement with ShuffleNet V2 (1.0x) model on ImageNet-1k dataset.



Paperid:865
Authors:Alexandros Kouris; Stylianos I. Venieris; Stefanos Laskaridis; Nicholas Lane
Title: Multi-Exit Semantic Segmentation Networks
Abstract:
Semantic segmentation arises as the backbone of many vision systems, spanning from self-driving cars and robot navigation to augmented reality and teleconferencing. Frequently operating under stringent latency constraints within a limited resource envelope, optimising for efficient execution becomes important. At the same time, the heterogeneous capabilities of the target platforms and diverse constraints of different applications require the design and training of multiple target-specific segmentation models, leading to excessive maintenance costs. To this end, we propose a framework for converting state-of-the-art segmentation CNNs to Multi-Exit Semantic Segmentation (MESS) networks: specially trained models that employ parametrised early exits along their depth to i) dynamically save computation during inference on easier samples and ii) save training and maintenance cost by offering a post-training customisable speed-accuracy trade-off. Designing and training such networks naively can hurt performance. Thus, we propose novel two-staged training scheme for multi-exit networks. Furthermore, the parametrisation of MESS enables co-optimising the number, placement and architecture of the attached segmentation heads along with the exit policy, upon deployment via exhaustive search in <1GPUh. This allows MESS to rapidly adapt to the device capabilities and application requirements for each target use-case, offering a train-once-deploy-everywhere solution. MESS variants achieve latency gains of up to 2.83x with the same accuracy, or 5.33 pp higher accuracy for the same computational budget, compared to the original backbone network. Lastly, MESS delivers orders of magnitude faster architectural customisation, compared to state-of-the-art techniques.



Paperid:866
Authors:Bernd Prach; Christoph H. Lampert
Title: Almost-Orthogonal Layers for Efficient General-Purpose Lipschitz Networks
Abstract:
It is a highly desirable property for deep networks to be robust against small input changes. One popular way to achieve this property is by designing networks with a small Lipschitz constant. In this work, we propose a new technique for constructing such Lipschitz networks that has a number of desirable properties: it can be applied to any linear network layer (fully-connected or convolutional), it provides formal guarantees on the Lipschitz constant, it is easy to implement and efficient to run, and it can be combined with any training objective and optimization method. In fact, our technique is the first one in the literature that achieves all of these properties simultaneously. Our main contribution is a rescaling-based weight matrix parametrization that guarantees each network layer to have a Lipschitz constant of at most 1 and results in the learned weight matrices to be close to orthogonal. Hence we call such layers almost-orthogonal Lipschitz (AOL). Experiments and ablation studies in the context of image classification with certified robust accuracy confirm that AOL layers achieve results that are on par with most existing methods. Yet, they are simpler to implement and more broadly applicable, because they do not require computationally expensive matrix orthogonalization or inversion steps as part of the network architecture. We provide code at https://github.com/berndprach/AOL.



Paperid:867
Authors:Dong Wang; Zhao Zhang; Ziwei Zhao; Yuhang Liu; Yihong Chen; Liwei Wang
Title: \texttt{\textbf{PointScatter}}: Point Set Representation for Tubular Structure Extraction
Abstract:
This paper explores the point set representation for tubular structure extraction tasks. Compared with the traditional mask representation, the point set representation enjoys its flexibility and representation ability, which would not be restricted by the fixed grid as the mask. Inspired by this, we propose PointScatter, an alternative to the segmentation models for the tubular structure extraction task. PointScatter splits the image into scatter regions and parallelly predicts points for each scatter region. We further propose the greedy-based region-wise bipartite matching algorithm to train the network end-to-end and efficiently. We benchmark the PointScatter on four public tubular datasets, and the extensive experiments on tubular structure segmentation and centerline extraction task demonstrate the effectiveness of our approach. Code is available at https://github.com/zhangzhao2022/pointscatter.



Paperid:868
Authors:Ziwei Zhao; Dong Wang; Yihong Chen; Ziteng Wang; Liwei Wang
Title: Check and Link: Pairwise Lesion Correspondence Guides Mammogram Mass Detection
Abstract:
Detecting mass in mammogram is significant due to the high occurrence and mortality of breast cancer. In mammogram mass detection, modeling pairwise lesion correspondence explicitly is particularly important. However, most of the existing methods build relatively coarse correspondence and have not utilized correspondence supervision. In this paper, we propose a new transformer-based framework CL-Net to learn lesion detection and pairwise correspondence in an end-to-end manner. In CL-Net, View-Interactive Lesion Detector is proposed to achieve dynamic interaction across candidates of cross views, while Lesion Linker employs the correspondence supervision to guide the interaction process more accurately. The combination of these two designs accomplishes precise understanding of pairwise lesion correspondence for mammograms. Experiments show that CL-Net yields state-of-the-art performance on the public DDSM dataset and our in-house dataset. Moreover, it outperforms previous methods by a large margin in low FPI regime.



Paperid:869
Authors:Simon Reiß; Constantin Seibold; Alexander Freytag; Erik Rodner; Rainer Stiefelhagen
Title: Graph-Constrained Contrastive Regularization for Semi-Weakly Volumetric Segmentation
Abstract:
Semantic volume segmentation suffers from the requirement of having voxel-wise annotated ground-truth data, which requires immense effort to obtain. In this work, we investigate how models can be trained from sparsely annotated volumes, i.e. volumes with only individual slices annotated. By formulating the scenario as a semi-weakly supervised problem where only some regions in the volume are annotated, we obtain surprising results: expensive dense volumetric annotations can be replaced by cheap, partially labeled volumes with limited impact on accuracy if the hypothesis space of valid models gets properly constrained during training. With our Contrastive Constrained Regularization (Con2R), we demonstrate that 3D convolutional models can be trained with less than 4% of only two dimensional ground-truth labels and still reach up to 88% accuracy of fully supervised baseline models with dense volumetric annotations. To get insights into Con2Rs success, we study how strong semi-supervised algorithms transfer to our new volumetric semi-weakly supervised setting. In this manner, we explore retinal fluid and brain tumor segmentation and give a detailed look into accuracy progression for scenarios with extremely scarce labels.



Paperid:870
Authors:Ziqi Zhou; Lei Qi; Yinghuan Shi
Title: Generalizable Medical Image Segmentation via Random Amplitude Mixup and Domain-Specific Image Restoration
Abstract:
For medical image analysis, segmentation models trained on one or several domains lack generalization ability to unseen domains due to discrepancies between different data acquisition policies. We argue that the degeneration in segmentation performance is mainly attributed to overfitting to source domains and domain shift. To this end, we present a novel generalizable medical image segmentation method. To be specific, we design our approach as a multi-task paradigm by combining the segmentation model with a self-supervision domain-specific image restoration (DSIR) module for model regularization. We also design a random amplitude mixup (RAM) module, which incorporates low-level frequency information of different domain images to synthesize new images. To guide our model be resistant to domain shift, we introduce a semantic consistency loss. We demonstrate the performance of our method on two public generalizable segmentation benchmarks in medical images, which validates our method could achieve the state-of-the-art performance.



Paperid:871
Authors:Pengfei Guo; Dong Yang; Ali Hatamizadeh; An Xu; Ziyue Xu; Wenqi Li; Can Zhao; Daguang Xu; Stephanie Harmon; Evrim Turkbey; Baris Turkbey; Bradford Wood; Francesca Patella; Elvira Stellato; Gianpaolo Carrafiello; Vishal M. Patel; Holger R. Roth
Title: Auto-FedRL: Federated Hyperparameter Optimization for Multi-Institutional Medical Image Segmentation
Abstract:
Federated learning (FL) is a distributed machine learning technique that enables collaborative model training while avoiding explicit data sharing. The inherent privacy-preserving property of FL algorithms makes them especially attractive to the medical field. However, in case of heterogeneous client data distributions, standard FL methods are unstable and require intensive hyperparameter tuning to achieve optimal performance. Conventional hyperparameter optimization algorithms are impractical in real-world FL applications as they involve numerous training trials, which are often not affordable with limited compute budgets. In this work, we propose an efficient reinforcement learning (RL)-based federated hyperparameter optimization algorithm, termed Auto-FedRL, in which an online RL agent can dynamically adjust hyperparameters of each client based on the current training progress. Extensive experiments are conducted to investigate different search strategies and RL agents. The effectiveness of the proposed method is validated on a heterogeneous data split of the CIFAR-10 dataset as well as two real-world medical image segmentation datasets for COVID-19 lesion segmentation in chest CT and pancreas segmentation in abdominal CT..



Paperid:872
Authors:Jiacheng Wang; Yueming Jin; Liansheng Wang
Title: Personalizing Federated Medical Image Segmentation via Local Calibration
Abstract:
Medical image segmentation under federated learning (FL) is a promising direction by allowing multiple clinical sites to collaboratively learn a global model without centralizing datasets. However, using a single model to adapt to various data distributions from different sites is extremely challenging. Personalized FL tackles this issue by only utilizing partial model parameters shared from global server, while keeping the rest to adapt to its own data distribution in the local training of each site. However, most existing methods concentrate on the partial parameter splitting, while do not consider the inter-site in-consistencies during the local training, which in fact can facilitate the knowledge communication over sites to benefit the model learning for improving the local accuracy. In this paper, we propose a personalized federated framework with Local Calibration (LC-Fed), to leverage the inter-site in-consistencies in both feature- and prediction- levels to boost the segmentation. Concretely, as each local site has its alternative attention on the various features, we first design the contrastive site embedding coupled with channel selection operation to calibrate the encoded features. Moreover, we propose to exploit the knowledge of prediction-level in-consistency to guide the personalized modeling on the ambiguous regions, e.g., anatomical boundaries. It is achieved by computing a disagreement-aware map to calibrate the prediction. Effectiveness of our method has been verified on three medical image segmentation tasks with different modalities, where our method consistently shows superior performance to the state-of-the-art personalized FL methods. Code is available at https://github.com/jcwang123/FedLC.



Paperid:873
Authors:Zihao Yin; Ping Gong; Chunyu Wang; Yizhou Yu; Yizhou Wang
Title: One-Shot Medical Landmark Localization by Edge-Guided Transform and Noisy Landmark Refinement
Abstract:
As an important upstream task for many medical applications, supervised landmark localization still requires non-negligible annotation costs to achieve desirable performance. Besides, due to cumbersome collection procedures, the limited size of medical landmark datasets impacts the effectiveness of large-scale self-supervised pre-training methods. To address these challenges, we propose a two-stage framework for one-shot medical landmark localization, which first infers landmarks by unsupervised registration from the labeled exemplar to unlabeled targets, and then utilizes these noisy pseudo labels to train robust detectors. To handle the significant structure variations, we learn an end-to-end cascade of global alignment and local deformations, under the guidance of novel loss functions which incorporate edge information. In stage \uppercase\expandafter{\romannumeral2}, we explore self-consistency for selecting reliable pseudo labels and cross-consistency for semi-supervised learning. Our method achieves state-of-the-art performances on public datasets of different body parts, which demonstrates its general applicability. Code will be publicly available.



Paperid:874
Authors:Ming-Yang Ho; Min-Sheng Wu; Che-Ming Wu
Title: Ultra-High-Resolution Unpaired Stain Transformation via Kernelized Instance Normalization
Abstract:
While hematoxylin and eosin (H&E) is a standard staining procedure, immunohistochemistry (IHC) staining further serves as a diagnostic and prognostic method. However, acquiring special staining results requires substantial costs. Hence, we proposed a strategy for ultra-high-resolution unpaired image-to-image translation: Kernelized Instance Normalization (KIN), which preserves local information and successfully achieves seamless stain transformation with constant GPU memory usage. Given a patch, corresponding position, and a kernel, KIN computes local statistics using convolution operation. In addition, KIN can be easily plugged into most currently developed frameworks without re-training. We demonstrate that KIN achieves state-of-the-art stain transformation by replacing instance normalization (IN) layers with KIN layers in three popular frameworks and testing on two histopathological datasets. Furthermore, we manifest the generalizability of KIN with high-resolution natural images. Finally, human evaluation and several objective metrics are used to compare the performance of different approaches. Overall, this is the first successful study for the ultra-high-resolution unpaired image-to-image translation with constant space complexity. Code is available at: https://github.com/Kaminyou/URUST.



Paperid:875
Authors:Wenxuan Wang; Chen Chen; Jing Wang; Sen Zha; Yan Zhang; Jiangyun Li
Title: Med-DANet: Dynamic Architecture Network for Efficient Medical Volumetric Segmentation
Abstract:
For 3D medical image (e.g. CT and MRI) segmentation, the difficulty of segmenting each slice in a clinical case varies greatly. Previous research on volumetric medical image segmentation in a slice-by-slice manner conventionally use the identical 2D deep neural network to segment all the slices of the same case, ignoring the data heterogeneity among image slices. In this paper, we focus on multi-modal 3D MRI brain tumor segmentation and propose a dynamic architecture network named Med-DANet based on adaptive model selection to achieve effective accuracy and efficiency trade-off. For each slice of the input 3D MRI volume, our proposed method learns a slice-specific decision by the Decision Network to dynamically select a suitable model from the predefined Model Bank for the subsequent 2D segmentation task. Extensive experimental results on both BraTS 2019 and 2020 datasets show that our proposed method achieves comparable or better results than previous state-of-the-art methods for 3D MRI brain tumor segmentation with much less model complexity. Compared with the state-of-the-art 3D method TransBTS, the proposed framework improves the model efficiency by up to 3.5x without sacrificing the accuracy. Our code will be publicly available at https://github.com/Wenxuan-1119/Med-DANet.



Paperid:876
Authors:Jiawei Yang; Hanbo Chen; Yuan Liang; Junzhou Huang; Lei He; Jianhua Yao
Title: ConCL: Concept Contrastive Learning for Dense Prediction Pre-training in Pathology Images
Abstract:
Detecting and segmenting objects within whole slide images is essential in computational pathology workflow. Self-supervised learning (SSL) is appealing to such annotation-heavy tasks. Despite the extensive benchmarks in natural images for dense tasks, such studies are, unfortunately, absent in current works for pathology. Our paper in- tends to narrow this gap. We first benchmark representative SSL methods for dense prediction tasks in pathology images. Then, we propose concept contrastive learning (ConCL), an SSL framework for dense pre-training. We explore how ConCL performs with concepts provided by different sources and end up with proposing a simple dependency-free concept generating method that does not rely on external segmentation algorithms or saliency detection models. Extensive experiments demonstrate the superiority of ConCL over previous state-of-the-art SSL methods across different settings. Along our exploration, we distill several important and intriguing components contributing to the success of dense pre-training for pathology images. We hope this work could provide useful data points and encourage the community to conduct ConCL pre-training for problems of interest. Code is available.



Paperid:877
Authors:Axel Levy; Frédéric Poitevin; Julien Martel; Youssef Nashed; Ariana Peck; Nina Miolane; Daniel Ratner; Mike Dunne; Gordon Wetzstein
Title: CryoAI: Amortized Inference of Poses for Ab Initio Reconstruction of 3D Molecular Volumes from Real Cryo-EM Images
Abstract:
Cryo-electron microscopy (cryo-EM) has become a tool of fundamental importance in structural biology, helping us understand the basic building blocks of life. The algorithmic challenge of cryo-EM is to jointly estimate the unknown 3D poses and the 3D electron scattering potential of a biomolecule from millions of extremely noisy 2D images. Existing reconstruction algorithms, however, cannot easily keep pace with the rapidly growing size of cryo-EM datasets due to their high computational and memory cost. We introduce cryoAI, an ab initio reconstruction algorithm for homogeneous conformations that uses direct gradient-based optimization of particle poses and the electron scattering potential from single-particle cryo-EM data. CryoAI combines a learned encoder that predicts the poses of each particle image with a physics-based decoder to aggregate each particle image into an implicit representation of the scattering potential volume. This volume is stored in the Fourier domain for computational efficiency and leverages a modern coordinate network architecture for memory efficiency. Combined with a symmetrized loss function, this framework achieves results of a quality on par with state-of-the-art cryo-EM solvers for both simulated and experimental data, one order of magnitude faster for large datasets and with significantly lower memory requirements than existing methods.



Paperid:878
Authors:Yutong Xie; Jianpeng Zhang; Yong Xia; Qi Wu
Title: UniMiSS: Universal Medical Self-Supervised Learning via Breaking Dimensionality Barrier
Abstract:
Self-supervised learning (SSL) opens up huge opportunities for medical image analysis that is well known for its lack of annotations. However, aggregating massive (unlabeled) 3D medical images like computerized tomography (CT) remains challenging due to its high imaging cost and privacy restrictions. In this paper, we advocate bringing a wealth of 2D images like chest X-rays as compensation for the lack of 3D data, aiming to build a universal medical self-supervised representation learning framework, called UniMiSS. The following problem is how to break the dimensionality barrier, i.e., making it possible to perform SSL with both 2D and 3D images? To achieve this, we design a pyramid U-like medical Transformer (MiT). It is composed of the switchable patch embedding (SPE) module and Transformers. The SPE module adaptively switches to either 2D or 3D patch embedding, depending on the input dimension. The embedded patches are converted into a sequence regardless of their original dimensions. The Transformers model the long-term dependencies in a sequence-to-sequence manner, thus enabling UniMiSS to learn representations from both 2D and 3D images. With the MiT as the backbone, we perform the UniMiSS in a self-distillation manner. We conduct expensive experiments on six 3D/2D medical image analysis tasks, including segmentation and classification. The results show that the proposed UniMiSS achieves promising performance on various downstream tasks, outperforming the ImageNet pre-training and other advanced SSL counterparts substantially. Code is available at https://github.com/YtongXie/UniMiSS-code.



Paperid:879
Authors:Zelin Zang; Siyuan Li; Di Wu; Ge Wang; Kai Wang; Lei Shang; Baigui Sun; Hao Li; Stan Z. Li
Title: DLME: Deep Local-Flatness Manifold Embedding
Abstract:
Manifold learning (ML) aims to seek low-dimensional embedding from high-dimensional data. The problem is challenging on real-world datasets, especially with under-sampling data, and we find that previous methods perform poorly in this case. Generally, ML methods first transform input data into a low-dimensional embedding space to maintain the data’s geometric structure and subsequently perform downstream tasks therein. The poor local connectivity of under-sampling data in the former step and inappropriate optimization objectives in the latter step leads to two problems: structural distortion and underconstrained embedding. This paper proposes a novel ML framework named Deep Local-flatness Manifold Embedding (DLME) to solve these problems. The proposed DLME constructs semantic manifolds by data augmentation and overcomes the structural distortion problem using a smoothness constrained based on a local flatness assumption about the manifold. To overcome the underconstrained embedding problem, we design a loss and theoretically demonstrate that it leads to a more suitable embedding based on the local flatness. Experiments on three types of datasets (toy, biological, and image) for various downstream tasks (classification, clustering, and visualization) show that our proposed DLME outperforms state-of-the-art ML and contrastive learning methods.



Paperid:880
Authors:Jiazhen Liu; Xirong Li; Qijie Wei; Jie Xu; Dayong Ding
Title: Semi-Supervised Keypoint Detector and Descriptor for Retinal Image Matching
Abstract:
For retinal image matching (RIM), we propose SuperRetina, the first end-to-end method with jointly trainable keypoint detector and descriptor. SuperRetina is trained in a novel semi-supervised manner. A small set of (nearly 100) images are incompletely labeled and used to supervise the network to detect keypoints on the vascular tree. To attack the incompleteness of manual labeling, we propose Progressive Keypoint Expansion to enrich the keypoint labels at each training epoch. By utilizing a keypoint-based improved triplet loss as its description loss, SuperRetina produces highly discriminative descriptors at full input image size. Extensive experiments on multiple real-world datasets justify the viability of SuperRetina. Even with manual labeling replaced by auto labeling and thus making the training process fully manual-annotation free, SuperRetina compares favorably against a number of strong baselines for two RIM tasks, i.e. image registration and identity verification.



Paperid:881
Authors:Tal Ben-Haim; Tammy Riklin Raviv
Title: Graph Neural Network for Cell Tracking in Microscopy Videos
Abstract:
We present a novel graph neural network (GNN) approach for cell tracking in high-throughput microscopy videos. By modeling the entire time-lapse sequence as a direct graph where cell instances are represented by its nodes and their associations by its edges, we extract the entire set of cell trajectories by looking for the maximal paths in the graph. This is accomplished by several key contributions incorporated into an end-to-end deep learning framework. We exploit a deep metric learning algorithm to extract cell feature vectors that distinguish between instances of different biological cells and assemble same cell instances. We introduce a new GNN block type which enables a mutual update of node and edge feature vectors, thus facilitating the underlying message passing process. The message passing concept, whose extent is determined by the number of GNN blocks, is of fundamental importance as it enables the ‘flow’ of information between nodes and edges much behind their neighbors in consecutive frames. Finally, we solve an edge classification problem and use the identified active edges to construct the cells’ tracks and lineage trees. We demonstrate the strengths of the proposed cell tracking approach by applying it to 2D and 3D datasets of different cell types, imaging setups, and experimental conditions. We show that our framework outperforms current state-of-the-art methods on most of the evaluated datasets. The code is available at our repository: https://github.com/talbenha/cell-tracker-gnn.



Paperid:882
Authors:Yujin Oh; Jong Chul Ye
Title: CXR Segmentation by AdaIN-Based Domain Adaptation and Knowledge Distillation
Abstract:
As segmentation labels are scarce, extensive researches have been conducted to train segmentation networks with domain adaptation, semi-supervised or self-supervised learning techniques to utilize abun- dant unlabeled dataset. However, these approaches appear different from each other, so it is not clear how these approaches can be combined for better performance. Inspired by recent multi-domain image translation approaches, here we propose a novel segmentation framework using adap- tive instance normalization (AdaIN), so that a single generator is trained to perform both domain adaptation and semi-supervised segmentation tasks via knowledge distillation by simply changing task-specific AdaIN codes. Specifically, our framework is designed to deal with difficult situ- ations in chest X-ray radiograph (CXR) segmentation, where labels are only available for normal data, but trained model should be applied to both normal and abnormal data. The proposed network demonstrates great generalizability under domain shift and achieves the state-of-the- art performance for abnormal CXR segmentation.



Paperid:883
Authors:Qinwen Huang; Ye Zhou; Hsuan-Fu Liu; Alberto Bartesaghi
Title: Accurate Detection of Proteins in Cryo-Electron Tomograms from Sparse Labels
Abstract:
Cryo-electron tomography (CET) combined with sub-volume averaging (SVA), is currently the only imaging technique capable of determining the structure of proteins imaged inside cells at molecular resolution. To obtain high-resolution reconstructions, sub-volumes containing randomly distributed copies of the protein of interest need be identified, extracted and subjected to SVA, making accurate particle detection a critical step in the CET processing pipeline. Classical template-based methods have high false-positive rates due to the very low signal-to-noise ratios (SNR) typical of CET volumes, while more recent neural-network based detection algorithms require extensive labeling, are very slow to train and can take days to run. To address these issues, we propose a novel particle detection framework that uses positive-unlabeled learning and exploits the unique properties of 3D tomograms to improve detection performance. Our end-to-end framework is able to identify particles within minutes when trained using a single partially labeled tomogram. We conducted extensive validation experiments on two challenging CET datasets representing different experimental conditions, and observed more than 10% improvement in mAP and F1 scores compared to existing particle picking methods used in CET. Ultimately, the proposed framework will facilitate the structural analysis of challenging biomedical targets imaged within the native environment of cells.



Paperid:884
Authors:Minkyu Jeon; Hyeonjin Park; Hyunwoo J. Kim; Michael Morley; Hyunghoon Cho
Title: K-SALSA: K-Anonymous Synthetic Averaging of Retinal Images via Local Style Alignment
Abstract:
The application of modern machine learning to retinal image analyses offers valuable insights into a broad range of human health conditions beyond ophthalmic diseases. Additionally, data sharing is key to fully realizing the potential of machine learning models by providing a rich and diverse collection of training data. However, the personally-identifying nature of retinal images, encompassing the unique vascular structure of each individual, often prevents this data from being shared openly. While prior works have explored image de-identification strategies based on synthetic averaging of images in other domains (e.g. facial images), existing techniques face difficulty in preserving both privacy and clinical utility in retinal images, as we demonstrate in our work. We therefore introduce k-SALSA, a generative adversarial network (GAN)-based framework for synthesizing retinal fundus images that summarize a given private dataset while satisfying the privacy notion of k-anonymity. k-SALSA brings together state-of-the-art techniques for training and inverting GANs to achieve practical performance on retinal images. Furthermore, k-SALSA leverages a new technique, called local style alignment, to generate a synthetic average that maximizes the retention of fine-grain visual patterns in the source images, thus improving the clinical utility of the generated images. On two benchmark datasets of diabetic retinopathy (EyePACS and APTOS), we demonstrate our improvement upon existing methods with respect to image fidelity, classification performance, and mitigation of membership inference attacks. Our work represents a step toward broader sharing of retinal images for scientific collaboration. Keywords : Medical image privacy, k-anonymity, generative adversarial networks, fundus imaging, synthetic data generation, style transfer"



Paperid:885
Authors:Moinak Bhattacharya; Shubham Jain; Prateek Prasanna
Title: RadioTransformer: A Cascaded Global-Focal Transformer for Visual Attention-Guided Disease Classification
Abstract:
In this work, we present RadioTransformer, a novel visual attention-driven transformer framework, that leverages radiologists’ gaze patterns and models their visuo-cognitive behavior for disease diagnosis on chest radiographs. Domain experts, such as radiologists, rely on visual information for medical image interpretation. On the other hand, deep neural networks have demonstrated significant promise in similar tasks even where visual interpretation is challenging. Eye-gaze tracking has been used to capture the viewing behavior of domain experts, lending insights into the complexity of visual search. However, deep learning frameworks, even those that rely on attention mechanisms, do not leverage this rich domain information. RadioTransformer fills this critical gap by learning from radiologists’ visual search patterns, encoded as ‘human visual attention regions’ in a cascaded global-focal transformer framework. The overall ‘global’ image characteristics and the more detailed ‘local’ features are captured by the proposed global and focal modules, respectively. We experimentally validate the efficacy of our student-teacher approach for 8 datasets involving different disease classification tasks where eye-gaze data is not available during the inference phase. Code: https://github.com/bmi-imaginelab/radiotransformer.



Paperid:886
Authors:Kevin Thandiackal; Boqi Chen; Pushpak Pati; Guillaume Jaume; Drew F. K. Williamson; Maria Gabrani; Orcun Goksel
Title: Differentiable Zooming for Multiple Instance Learning on Whole-Slide Images
Abstract:
Multiple Instance Learning (MIL) methods have become increasingly popular for classifying gigapixel-sized Whole-Slide Images (WSIs) in digital pathology. Most MIL methods operate at a single WSI magnification, by processing all the tissue patches. Such a formulation induces high computational requirements and constrains the contextualization of the WSI-level representation to a single scale. Certain MIL methods extend to multiple scales, but they are computationally more demanding. In this paper, inspired by the pathological diagnostic process, we propose ZoomMIL, a method that learns to perform multi-level zooming in an end-to-end manner. ZoomMIL builds WSI representations by aggregating tissue-context information from multiple magnifications. The proposed method outperforms the state-of-the-art MIL methods in WSI classification on two large datasets, while significantly reducing computational demands with regard to Floating-Point Operations (FLOPs) and processing time by 40-50x. Our code is available at: https://github.com/histocartography/zoommil.



Paperid:887
Authors:Chongyang Zhong; Lei Hu; Zihao Zhang; Shihong Xia
Title: Learning Uncoupled-Modulation CVAE for 3D Action-Conditioned Human Motion Synthesis
Abstract:
Motion capture data is largely needed in the movie and game industry in recent years. Since the motion capture system is expensive and requires manual post-processing, motion synthesis is a plausible solution to acquire more motion data. However, generating the action-conditioned, realistic, and diverse 3D human motions given the semantic action labels is still challenging because the mapping from semantic labels to real motion sequences is hard to depict. Previous work made some positive attempts like appending label tokens to pose encoding and performing action bias on latent space, however, how to synthesize diverse motions that accurately match the given label is still not fully explored. In this paper, we propose the Uncoupled-Modulation Conditional Variational AutoEncoder(UM-CVAE) to generate action-conditioned motions from scratch in an uncoupled manner. The main idea is twofold: (i)training an action-agnostic encoder to eliminate the action-related information to learn the easy-modulated latent representation; (ii)strengthening the action-conditioned process with FiLM-based action-aware modulation. We conduct extensive experiments on the HumanAct12, UESTC, and BABEL datasets, demonstrating that our method achieves state-of-the-art performance both qualitatively and quantitatively with potential applications.



Paperid:888
Authors:Bin Yan; Yi Jiang; Peize Sun; Dong Wang; Zehuan Yuan; Ping Luo; Huchuan Lu
Title: Towards Grand Unification of Object Tracking
Abstract:
We present a unified method, termed Unicorn, that can simultaneously solve four tracking problems (SOT, MOT, VOS, MOTS) with a single network using the same model parameters. Due to the fragmented definitions of the object tracking problem itself, most existing trackers are developed to address a single or part of tasks and overspecialize on the characteristics of specific tasks. By contrast, Unicorn provides a unified solution, adopting the same input, backbone, embedding, and head across all tracking tasks. For the first time, we accomplish the great unification of the tracking network architecture and learning paradigm. Unicorn performs on-par or better than its task-specific counterparts in 8 tracking datasets, including LaSOT, TrackingNet, MOT17, BDD100K, DAVIS16-17, MOTS20, and BDD100K MOTS. We believe that Unicorn will serve as a solid step towards the general vision model. Code is available at https://github.com/MasterBin-IIAU/Unicorn.



Paperid:889
Authors:Yifu Zhang; Peize Sun; Yi Jiang; Dongdong Yu; Fucheng Weng; Zehuan Yuan; Ping Luo; Wenyu Liu; Xinggang Wang
Title: ByteTrack: Multi-Object Tracking by Associating Every Detection Box
Abstract:
Multi-object tracking (MOT) aims at estimating bounding boxes and identities of objects in videos. Most methods obtain identities by associating detection boxes whose scores are higher than a threshold. The objects with low detection scores, e.g. occluded objects, are simply thrown away, which brings non-negligible true object missing and fragmented trajectories. To solve this problem, we present a simple, effective and generic association method, tracking by associating almost every detection box instead of only the high score ones. For the low score detection boxes, we utilize their similarities with tracklets to recover true objects and filter out the background detections. When applied to 9 different state-of-the-art trackers, our method achieves consistent improvement on IDF1 score ranging from 1 to 10 points. To put forwards the state-of-the-art performance of MOT, we design a simple and strong tracker, named ByteTrack. For the first time, we achieve 80.3 MOTA, 77.3 IDF1 and 63.1 HOTA on the test set of MOT17 with 30 FPS running speed on a single V100 GPU. ByteTrack also achieves state-of-the-art performance on MOT20, HiEve and BDD100K tracking benchmarks. The source code, pre-trained models with deploy versions and tutorials of applying to other trackers are released at https://github.com/ifzhang/ByteTrack.



Paperid:890
Authors:Yifu Zhang; Chunyu Wang; Xinggang Wang; Wenjun Zeng; Wenyu Liu
Title: Robust Multi-Object Tracking by Marginal Inference
Abstract:
Multi-object tracking in videos requires to solve a fundamental problem of one-to-one assignment between objects in adjacent frames. Most methods address the problem by first discarding impossible pairs whose feature distances are larger than a threshold, followed by linking objects using Hungarian algorithm to minimize the overall distance. However, we find that the distribution of the distances computed from Re-ID features may vary significantly for different videos. So there isn’t a single optimal threshold which allows us to safely discard impossible pairs. To address the problem, we present an efficient approach to compute a marginal probability for each pair of objects in real time. The marginal probability can be regarded as a normalized distance which is significantly more stable than the original feature distance. As a result, we can use a single threshold for all videos. The approach is general and can be applied to the existing trackers to obtain about one point improvement in terms of IDF1 metric. It achieves competitive results on MOT17 and MOT20 benchmarks. In addition, the computed probability is more interpretable which facilitates subsequent post-processing operations.



Paperid:891
Authors:Aleksandr Kim; Guillem Brasó; Aljoša Ošep; Laura Leal-Taixé
Title: PolarMOT: How Far Can Geometric Relations Take Us in 3D Multi-Object Tracking?
Abstract:
Most (3D) multi-object tracking methods rely on appearance-based cues for data association. By contrast, we investigate how far we can get by only encoding geometric relationships between objects in 3D space as cues for data-driven data association. We encode 3D detections as nodes in a graph, where spatial and temporal pairwise relations among objects are encoded via localized polar coordinates on graph edges. This representation makes our geometric relations invariant to global transformations and smooth trajectory changes, especially under non-holonomic motion. This allows our graph neural network to learn to effectively encode temporal and spatial interactions and fully leverage contextual and motion cues to obtain final scene interpretation by posing data association as edge classification. We establish a new state-of-the-art on nuScenes dataset and, more importantly, show that our method, PolarMOT, generalizes remarkably well across different locations (Boston, Singapore, Karlsruhe) and datasets (nuScenes and KITTI).



Paperid:892
Authors:Adam W. Harley; Zhaoyuan Fang; Katerina Fragkiadaki
Title: Particle Video Revisited: Tracking through Occlusions Using Point Trajectories
Abstract:
Tracking pixels in videos is typically studied as an optical flow estimation problem, where every pixel is described with a displacement vector that locates it in the next frame. Even though wider temporal context is freely available, prior efforts to take this into account have yielded only small gains over 2-frame methods. In this paper, we revisit Sand and Teller’s ""particle video"" approach, and study pixel tracking as a long-range motion estimation problem, where every pixel is described with a trajectory that locates it in multiple future frames. We re-build this classic approach using components that drive the current state-of-the-art in flow and object tracking, such as dense cost maps, iterative optimization, and learned appearance updates. We train our models using long-range amodal point trajectories mined from existing optical flow data that we synthetically augment with multi-frame occlusions. We test our approach in trajectory estimation benchmarks and in keypoint label propagation tasks, and compare favorably against state-of-the-art optical flow and feature tracking methods.



Paperid:893
Authors:Zelin Zhao; Ze Wu; Yueqing Zhuang; Boxun Li; Jiaya Jia
Title: Tracking Objects As Pixel-Wise Distributions
Abstract:
Multi-object tracking (MOT) requires detecting and associating objects through frames. Unlike tracking via detected bounding boxes or center points, we propose tracking objects as pixel-wise distributions. We instantiate this idea on a transformer-based architecture named P3AFormer, with pixel-wise propagation, prediction, and association. P3AFormer propagates pixel-wise features guided by flow information to pass messages between frames. Further, P3AFormer adopts a meta-architecture to produce multi-scale object feature maps. During inference, a pixel-wise association procedure is proposed to recover object connections through frames based on the pixel-wise prediction. P3AFormer yields 81.2\% in terms of MOTA on the MOT17 benchmark -- highest among all transformer networks to reach 80\% MOTA in literature. P3AFormer also outperforms state-of-the-arts on the MOT20 and KITTI benchmarks. The code is at https://github.com/dvlab-research/ECCV22-P3AFormer-Tracking-Objects-as-Pixel-wise-Distributions.



Paperid:894
Authors:Zhiyang Guo; Yunyao Mao; Wengang Zhou; Min Wang; Houqiang Li
Title: CMT: Context-Matching-Guided Transformer for 3D Tracking in Point Clouds
Abstract:
How to effectively match the target template features with the search area is the core problem in point-cloud-based 3D single object tracking. However, in the literature, most of the methods focus on devising sophisticated matching modules at point-level, while overlooking the rich spatial context information of points. To this end, we propose Context-Matching-Guided Transformer (CMT), a Siamese tracking paradigm for 3D single object tracking. In this work, we first leverage the local distribution of points to construct a horizontally rotation-invariant contextual descriptor for both the template and the search area. Then, a novel matching strategy based on shifted windows is designed for such descriptors to effectively measure the template-search contextual similarity. Furthermore, we introduce a target-specific transformer and a spatial-aware orientation encoder to exploit the target-aware information in the most contextually relevant template points, thereby enhancing the search feature for a better target proposal. We conduct extensive experiments to verify the merits of our proposed CMT and report a series of new state-of-the-art records on three widely-adopted datasets.



Paperid:895
Authors:Jinyu Yang; Zhongqun Zhang; Zhe Li; Hyung Jin Chang; Aleš Leonardis; Feng Zheng
Title: Towards Generic 3D Tracking in RGBD Videos: Benchmark and Baseline
Abstract:
Tracking in 3D scenes is gaining momentum because of its numerous applications in robotics, autonomous driving, and scene understanding. Currently, 3D tracking is limited to specific model-based approaches involving point clouds, which impedes 3D trackers from applying in natural 3D scenes. RGBD sensors provide a more reasonable and acceptable solution for 3D object tracking due to their readily available synchronised color and depth information. Thus, in this paper, we investigate a novel problem: is it possible to track a generic (class-agnostic) 3D object in RGBD videos and predict 3D bounding boxes of the object of interest? To inspire further research on this topic, we newly construct a standard benchmark for generic 3D object tracking, ‘Track-it-in-3D’, which contains 300 RGBD video sequences with dense 3D annotations and corresponding evaluation protocols. Furthermore, we propose an effective tracking baseline to estimate 3D bounding boxes for arbitrary objects in RGBD videos, by fusing appearance and spatial information effectively. The dataset and codes will be publicly available.



Paperid:896
Authors:Dooseop Choi; KyoungWook Min
Title: Hierarchical Latent Structure for Multi-modal Vehicle Trajectory Forecasting
Abstract:
Variational autoencoder (VAE) has widely been utilized for modeling data distributions because it is theoretically elegant, easy to train, and has nice manifold representations. However, when applied to image reconstruction and synthesis tasks, VAE shows the limitation that the generated sample tends to be blurry. We observe that a similar problem, in which the generated trajectory is located between adjacent lanes, often arises in VAE-based trajectory forecasting models. To mitigate this problem, we introduce a hierarchical latent structure into the VAE-based forecasting model. Based on the assumption that the trajectory distribution can be approximated as a mixture of simple distributions (or modes), the low-level latent variable is employed to model each mode of the mixture and the high-level latent variable is employed to represent the weights for the modes. To model each mode accurately, we condition the low-level latent variable using two lane-level context vectors computed in novel ways, one corresponds to vehicle-lane interaction and the other to vehicle-vehicle interaction. The context vectors are also used to model the weights via the proposed mode selection network. To evaluate our forecasting model, we use two large-scale real-world datasets. Experimental results show that our model is not only capable of generating clear multi-modal trajectory distributions but also outperforms the state-of-the-art (SOTA) models in terms of prediction accuracy.



Paperid:897
Authors:Shenyuan Gao; Chunluan Zhou; Chao Ma; Xinggang Wang; Junsong Yuan
Title: AiATrack: Attention in Attention for Transformer Visual Tracking
Abstract:
Transformer trackers have achieved impressive advancements recently, where the attention mechanism plays an important role. However, the independent correlation computation in the attention mechanism could result in noisy and ambiguous attention weights, which inhibits further performance improvement. To address this issue, we propose an attention in attention (AiA) module, which enhances appropriate correlations and suppresses erroneous ones by seeking consensus among all correlation vectors. Our AiA module can be readily applied to both self-attention blocks and cross-attention blocks to facilitate feature aggregation and information propagation for visual tracking. Moreover, we propose a streamlined Transformer tracking framework, dubbed AiATrack, by introducing efficient feature reuse and target-background embeddings to make full use of temporal references. Experiments show that our tracker achieves state-of-the-art performance on six tracking benchmarks while running at a real-time speed.



Paperid:898
Authors:Deqing Sun; Charles Herrmann; Fitsum Reda; Michael Rubinstein; David J. Fleet; William T. Freeman
Title: Disentangling Architecture and Training for Optical Flow
Abstract:
How important are training details and datasets to recent optical flow models like RAFT? And do they generalize? To explore these questions, rather than develop a new model, we revisit three prominent models, PWC-Net, IRR-PWC and RAFT, with a common set of modern training techniques, and observe significantly better performance, demonstrating the importance and generality of these training details. Our newly trained PWC-Net and IRR-PWC models show surprisingly large improvements, up to 30% versus original published results on Sintel and KITTI 2015 benchmarks. They outperform the more recent Flow1D on KITTI 2015 while being 3× faster during inference. Our newly trained RAFT achieves an Fl-all score of 4.31% on KITTI 2015, more accurate than all published optical flow methods. Our results demonstrate the benefits of separating the contributions of models, training techniques and datasets when analyzing performance gains of optical flow methods. Our source code will be publicly available.



Paperid:899
Authors:Jenny Schmalfuss; Philipp Scholze; Andrés Bruhn
Title: A Perturbation-Constrained Adversarial Attack for Evaluating the Robustness of Optical Flow
Abstract:
Recent optical flow methods are almost exclusively judged in terms of accuracy, while their robustness is often neglected. Although adversarial attacks offer a useful tool to perform such an analysis, current attacks on optical flow methods focus on real-world attacking scenarios rather than a worst case robustness assessment. Hence, in this work, we propose a novel adversarial attack - the Perturbation-Constrained Flow Attack (PCFA) - that emphasizes destructivity over applicability as a real-world attack. PCFA is a global attack that optimizes adversarial perturbations to shift the predicted flow towards a specified target flow, while keeping the L2 norm of the perturbation below a chosen bound. Our experiments demonstrate PCFA’s applicability in white- and black-box settings, and show it finds stronger adversarial samples than previous attacks. Based on these strong samples, we provide the first joint ranking of optical flow methods considering both prediction quality and adversarial robustness, which reveals state-of-the-art methods to be particularly vulnerable. Code is available at https://github.com/cv-stuttgart/PCFA.



Paperid:900
Authors:Luojie Huang; Yikang Liu; Li Chen; Eric Z. Chen; Xiao Chen; Shanhui Sun
Title: Robust Landmark-Based Stent Tracking in X-Ray Fluoroscopy
Abstract:
In clinical procedures of angioplasty (i.e., open clogged coronary arteries), devices such as balloons and stents need to be placed and expanded in arteries under the guidance of X-ray fluoroscopy. Due to the limitation of X-ray dose, the resulting images are often noisy. To check the correct placement of these devices, typically multiple motion-compensated frames are averaged to enhance the view. Therefore, device tracking is a necessary procedure for this purpose. Even though angioplasty devices are designed to have radiopaque markers for the ease of tracking, current methods struggle to deliver satisfactory results due to the small marker size and complex scenes in angioplasty. In this paper, we propose an end-to-end deep learning framework for single stent tracking, which consists of three hierarchical modules: a U-Net for landmark detection, a ResNet for stent proposal and feature extraction, and a graph convolutional neural network for stent tracking that temporally aggregates both spatial information and appearance features. The experiments show that our method performs significantly better in detection compared with the state-of-the-art point-based tracking models. In addition, its fast inference speed satisfies clinical requirements.



Paperid:901
Authors:Song Wen; Hao Wang; Dimitris N. Metaxas
Title: Social ODE: Multi-agent Trajectory Forecasting with Neural Ordinary Differential Equations
Abstract:
Multi-agent trajectory forecasting has recently attracted a lot of attention due to its widespread applications including autonomous driving. Most previous methods use RNNs or Transformers to model agent dynamics in the temporal dimension and social pooling or GNNs to model interactions with other agents; these approaches usually fail to learn the underlying continuous temporal dynamics and agent interactions explicitly. To address these problems, we propose Social ODE which explicitly models temporal agent dynamics and agent interactions. Our approach leverages Neural ODEs to model continuous temporal dynamics, and incorporates distance, interaction intensity, and aggressiveness estimation into agent interaction modeling in latent space. We show in extensive experiments that our Social ODE approach compares favorably with state-of-the-art, and more importantly, can successfully avoid sudden obstacles and effectively control the motion of the agent, while previous methods often fail in such cases.



Paperid:902
Authors:Li-Wu Tsao; Yan-Kai Wang; Hao-Siang Lin; Hong-Han Shuai; Lai-Kuan Wong; Wen-Huang Cheng
Title: Social-SSL: Self-Supervised Cross-Sequence Representation Learning Based on Transformers for Multi-agent Trajectory Prediction
Abstract:
Earlier trajectory prediction approaches focus on ways of capturing sequential structures among pedestrians by using recurrent networks, which is known to have some limitations in capturing long sequence structures. To address this limitation, some recent works proposed Transformer-based architectures, which are built with attention mechanisms. However, these Transformer-based networks are trained end-to-end without capitalizing on the value of pre-training. In this work, we propose Social-SSL that captures cross-sequence trajectory structures via self-supervised pre-training, which plays a crucial role in improving both data efficiency and generalizability of Transformer networks for trajectory prediction. Specifically, Social-SSL models the interaction and motion patterns with three pretext tasks: interaction type prediction, closeness prediction, and masked cross-sequence to sequence pre-training. Comprehensive experiments show that Social-SSL outperforms the state-of-the-art methods by at least 12% and 20% on ETH/UCY and SDD datasets in terms of Average Displacement Error and Final Displacement Error.



Paperid:903
Authors:Sirui Xu; Yu-Xiong Wang; Liang-Yan Gui
Title: Diverse Human Motion Prediction Guided by Multi-level Spatial-Temporal Anchors
Abstract:
Predicting diverse human motions given a sequence of historical poses has received increasing attention. Despite rapid progress, existing work captures the multi-modal nature of human motions primarily through likelihood-based sampling, where the mode collapse has been widely observed. In this paper, we propose a simple yet effective approach that disentangles randomly sampled codes with a deterministic learnable component named anchors to promote sample precision and diversity. Anchors are further factorized into spatial anchors and temporal anchors, which provide attractively interpretable control over spatial-temporal disparity. In principle, our spatial-temporal anchor-based sampling (STARS) can be applied to different motion predictors. Here we propose an interaction-enhanced spatial-temporal graph convolutional network (IE-STGCN) that encodes prior knowledge of human motions (e.g., spatial locality), and incorporate the anchors into it. Extensive experiments demonstrate that our approach outperforms state of the art in both stochastic and deterministic prediction, suggesting it as a unified framework for modeling human motions. Our code and pretrained models are available at https://github.com/Sirui-Xu/STARS.



Paperid:904
Authors:Inhwan Bae; Jin-Hwi Park; Hae-Gon Jeon
Title: Learning Pedestrian Group Representations for Multi-modal Trajectory Prediction
Abstract:
Modeling the dynamics of people walking is a problem of long-standing interest in computer vision. Many previous works involving pedestrian trajectory prediction define a particular set of individual actions to implicitly model group actions. In this paper, we present a novel architecture named GP-Graph which has collective group representations for effective pedestrian trajectory prediction in crowded environments, and is compatible with all types of existing approaches. A key idea of GP-Graph is to model both individual-wise and group-wise relations as graph representations. To do this, GP-Graph first learns to assign each pedestrian into the most likely behavior group. Using this assignment information, GP-Graph then forms both intra- and inter-group interactions as graphs, accounting for human-human relations within a group and group-group relations, respectively. To be specific, for the intra-group interaction, we mask pedestrian graph edges out of an associated group. We also propose group pooling&unpooling operations to represent a group with multiple pedestrians as one graph node. Lastly, GP-Graph infers a probability map for socially-acceptable future trajectories from the integrated features of both group interactions. Moreover, we introduce a group-level latent vector sampling to ensure collective inferences over a set of possible future trajectories. Extensive experiments are conducted to validate the effectiveness of our architecture, which demonstrates consistent performance improvements with publicly available benchmarks. Code is publicly available at https://github.com/inhwanbae/GPGraph.



Paperid:905
Authors:Gang Zhang; Xiaoyan Li; Zhenhua Wang
Title: Sequential Multi-View Fusion Network for Fast LiDAR Point Motion Estimation
Abstract:
The LiDAR point motion estimation, including motion state prediction and velocity estimation, is crucial for understanding a dynamic scene in autonomous driving. Recent 2D projection-based methods run in real-time by applying the well-optimized 2D convolution networks on either the bird’s-eye view (BEV) or the range view (RV) but suffer from lower accuracy due to information loss during the 2D projection. Thus, we propose a novel sequential multi-view fusion network (SMVF), composed of a BEV branch and an RV branch, in charge of encoding the motion information and spatial information, respectively. By looking from distinct views and integrating with the original LiDAR point features, the SMVF produces a comprehensive motion prediction, while keeping its efficiency. Moreover, to generalize the motion estimation well to the objects with fewer training samples, we propose a sequential instance copy-paste (SICP) for generating realistic LiDAR sequences for these objects. The experiments on the SemanticKITTI moving object segmentation (MOS) and Waymo scene flow benchmarks demonstrate that our SMVF outperforms all existing methods by a large margin. \keywords{Motion State Prediction, Velocity Estimation, Multi-View Fusion, Generalization of Motion Estimation}"



Paperid:906
Authors:Yanyan Li; Federico Tombari
Title: E-Graph: Minimal Solution for Rigid Rotation with Extensibility Graphs
Abstract:
Minimal solutions for relative rotation and translation estimation tasks have been explored in different scenarios, typically relying on the so-called co-visibility graph. However, how to build direct rotation relationships between two frames without overlap is still an open topic, which, if solved, could greatly improve the accuracy of visual odometry. In this paper, a new minimal solution is proposed to solve relative rotation estimation between two images without overlapping areas by exploiting a new graph structure, which we call Extensibility Graph (E-Graph). Differently from a co-visibility graph, high-level landmarks, including vanishing directions and plane normals, are stored in our E-Graph, which are geometrically extensible. Based on E-Graph, the rotation estimation problem becomes simpler and more elegant, as it can deal with pure rotational motion and requires fewer assumptions, e.g. Manhattan/Atlanta World, planar/vertical motion. Finally, we embed our rotation estimation strategy into a complete camera tracking and mapping system which obtains 6-DoF camera poses and a dense 3D mesh model. Extensive experiments on public benchmarks demonstrate that the proposed method achieves state-of-the-art tracking performance.



Paperid:907
Authors:Sukai Wang; Ming Liu
Title: Point Cloud Compression with Range Image-Based Entropy Model for Autonomous Driving
Abstract:
For autonomous driving systems, the storage cost and transmission speed of the large-scale point clouds become an important bottleneck because of their large volume. In this paper, we propose a range image-based three-stage framework to compress the scanning LiDAR’s point clouds using the entropy model. In our three-stage framework, we refine the coarser range image by converting the regression problem into the limited classification problem to improve the performance of generating accurate point clouds. And in the feature extraction part, we propose a novel attention Conv layer to fuse the voxel-based 3D features in the 2D range image. Compared with the Octree-based compression methods, the range image compression with the entropy model performs better in the autonomous driving scene. Experiments on LiDARs with different lines and in different scenarios show that our proposed compression scheme outperforms the state-of-the-art approaches in reconstruction quality and downstream tasks by a wide margin.



Paperid:908
Authors:Botao Ye; Hong Chang; Bingpeng Ma; Shiguang Shan; Xilin Chen
Title: Joint Feature Learning and Relation Modeling for Tracking: A One-Stream Framework
Abstract:
The current popular two-stream, two-stage tracking framework extracts the template and the search region features separately and then performs relation modeling, thus the extracted features lack the awareness of the target and have limited target-background discriminability. To tackle the above issue, we propose a novel one-stream tracking (OSTrack) framework that unifies feature learning and relation modeling by bridging the template-search image pairs with bidirectional information flows. In this way, discriminative target-oriented features can be dynamically extracted by mutual guidance. Since no extra heavy relation modeling module is needed and the implementation is highly parallelized, the proposed tracker runs at a fast speed. To further improve the inference efficiency, an in-network candidate early elimination module is proposed based on the strong similarity prior calculated in the one-stream framework. As a unified framework, OSTrack achieves state-of-the-art performance on multiple benchmarks, in particular, it shows impressive results on the one-shot tracking benchmark GOT-10k, i.e., achieving 73.7% AO, improving the existing best result (SwinTrack) by 4.3%. Besides, our method maintains a good performance-speed trade-off and shows faster convergence. The code and models are available at https://github.com/botaoye/OSTrack.



Paperid:909
Authors:Guy Tevet; Brian Gordon; Amir Hertz; Amit H. Bermano; Daniel Cohen-Or
Title: MotionCLIP: Exposing Human Motion Generation to CLIP Space
Abstract:
We introduce MotionCLIP, a 3D human motion auto-encoder featuring a latent embedding that is disentangled, well behaved, and supports highly semantic textual descriptions. MotionCLIP gains its unique power by aligning its latent space with that of the Contrastive Language-Image Pre-training (CLIP) model. Aligning the human motion manifold to CLIP space implicitly infuses the extremely rich semantic knowledge of CLIP into the manifold. In particular, it helps continuity by placing semantically similar motions close to one another, and disentanglement, which is inherited from the CLIP-space structure. MotionCLIP comprises a transformer-based motion auto-encoder, trained to reconstruct motion while being aligned to its text label’s position in CLIP-space. We further leverage CLIP’s unique visual understanding and inject an even stronger signal through aligning motion to rendered frames in a self-supervised manner. We show that although CLIP has never seen the motion domain, MotionCLIP offers unprecedented text-to-motion abilities, allowing out-of-domain actions, disentangled editing, and abstract language specification. For example, the text prompt “couch” is decoded into a sitting down motion, due to lingual similarity, and the prompt “Spiderman” results in a web-swinging-like solution that is far from seen during training. In addition, we show how the introduced latent space can be leveraged for motion interpolation, editing and recognition.



Paperid:910
Authors:Boyu Chen; Peixia Li; Lei Bai; Lei Qiao; Qiuhong Shen; Bo Li; Weihao Gan; Wei Wu; Wanli Ouyang
Title: Backbone Is All Your Need: A Simplified Architecture for Visual Object Tracking
Abstract:
Exploiting a general-purpose neural architecture to replace hand-wired designs or inductive biases has recently drawn extensive interest. However, existing tracking approaches rely on customized sub-modules and need prior knowledge for architecture selection, hindering the development of tracking in a more general system. This paper presents a Simplified Tracking architecture (SimTrack) by leveraging a transformer backbone for joint feature extraction and interaction. Unlike existing Siamese trackers, we serialize the input images and concatenate them directly before the one-branch backbone. Feature interaction in the backbone helps to remove well-designed interaction modules and produce a more efficient and effective framework. To reduce the information loss from down-sampling in vision transformers, we further propose a foveal window strategy, providing more diverse input patches with acceptable computational costs. Our SimTrack improves the baseline with 2.5%/2.6% AUC gains on LaSOT/TNL2K and gets results competitive with other specialized tracking algorithms without bells and whistles. The source codes are available at https://github.com/LPXTT/SimTrack.



Paperid:911
Authors:Yiqi Zhong; Zhenyang Ni; Siheng Chen; Ulrich Neumann
Title: Aware of the History: Trajectory Forecasting with the Local Behavior Data
Abstract:
The historical trajectories previously passing through a location may help infer the future trajectory of an agent currently at this location. Despite great improvements in trajectory forecasting with the guidance of high-definition maps, only a few works have explored such local historical information. In this work, we re-introduce this information as a new type of input data for trajectory forecasting systems: the local behavior data, which we conceptualize as a collection of location-specific historical trajectories. Local behavior data helps the systems emphasize the prediction locality and better understand the impact of static map objects on moving agents. We propose a novel local-behavior-aware (LBA) prediction framework that improves forecasting accuracy by fusing information from observed trajectories, HD maps, and local behavior data. Also, where such historical data is insufficient or unavailable, we employ a local-behavior-free (LBF) prediction framework, which adopts a knowledge-distillation-based architecture to infer the impact of missing data. Extensive experiments demonstrate that upgrading existing methods with these two frameworks significantly improves their performances. Especially, the LBA framework boosts the SOTA methods’ performance on the nuScenes dataset by at least 14% for the K=1 metrics.



Paperid:912
Authors:Shuai Yuan; Xian Sun; Hannah Kim; Shuzhi Yu; Carlo Tomasi
Title: Optical Flow Training under Limited Label Budget via Active Learning
Abstract:
Supervised training of optical flow predictors generally yields better accuracy than unsupervised training. However, the improved performance comes at an often high annotation cost. Semi-supervised training trades off accuracy against annotation cost. We use a simple yet effective semi-supervised training method to show that even a small fraction of labels can improve flow accuracy by a significant margin over unsupervised training. In addition, we propose active learning methods based on simple heuristics to further reduce the number of labels required to achieve the same target accuracy. Our experiments on both synthetic and real optical flow datasets show that our semi-supervised networks generally need around 50% of the labels to achieve close to full-label accuracy, and only around 20% with active learning on Sintel. We also analyze and show insights on the factors that may influence active learning performance. Code is available at https://github.com/duke-vision/optical-flow-active-learning-release.



Paperid:913
Authors:Zhixiong Pi; Weitao Wan; Chong Sun; Changxin Gao; Nong Sang; Chen Li
Title: Hierarchical Feature Embedding for Visual Tracking
Abstract:
Features extracted by existing tracking methods may contain instance- and category-level information. However, it usually occurs that either instance- or category-level information uncontrollably dominates the feature embeddings depending on the training data distribution, since the two types of information are not explicitly modeled. A more favorable way is to produce features that emphasize both types of information in visual tracking. To achieve this, we propose a hierarchical feature embedding model which separately learns the instance and category information, and progressively embeds them. We develop the instance-aware and category-aware modules that collaborate from different semantic levels to produce discriminative and robust feature embeddings. The instance-aware module concentrates on the instance level in which the inter-video contrastive learning mechanism is adopted to facilitate inter-instance separability and intra-instance compactness. However, it is challenging to force the intra-instance compactness by using instance-level information alone because of the prevailing appearance changes of the instance in visual tracking. To tackle this problem, the category-aware module is employed to summarize high-level category information which remains robust despite instance-level appearance changes. As such, intra-instance compactness can be effectively improved by jointly leveraging the instance- and category-aware modules. Experimental results on various tracking benchmarks demonstrate that the proposed method performs favorably against the state-of-the-arts.



Paperid:914
Authors:Suhwan Cho; Heansung Lee; Minhyeok Lee; Chaewon Park; Sungjun Jang; Minjung Kim; Sangyoun Lee
Title: Tackling Background Distraction in Video Object Segmentation
Abstract:
Semi-supervised video object segmentation (VOS) aims to densely track certain designated objects in videos. One of the main challenges in this task is the existence of background distractors that appear similar to the target objects. We propose three novel strategies to suppress such distractors: 1) a spatio-temporally diversified template construction scheme to obtain generalized properties of the target objects; 2) a learnable distance-scoring function to exclude spatially-distant distractors by exploiting the temporal consistency between two consecutive frames; 3) swap-and-attach augmentation to force each object to have unique features by providing training samples containing entangled objects. On all public benchmark datasets, our model achieves a comparable performance to contemporary state-of-the-art approaches, even with real-time performance. Qualitative results also demonstrate the superiority of our approach over existing methods. We believe our approach will be widely used for future VOS research.



Paperid:915
Authors:Abduallah Mohamed; Deyao Zhu; Warren Vu; Mohamed Elhoseiny; Christian Claudel
Title: Social-Implicit: Rethinking Trajectory Prediction Evaluation and the Effectiveness of Implicit Maximum Likelihood Estimation
Abstract:
Best-of-N (BoN) Average Displacement Error (ADE)/ Final Displacement Error (FDE) is the most used metric for evaluating trajectory prediction models. Yet, the BoN does not quantify the whole generated samples, resulting in an incomplete view of the model’s prediction quality and performance. We propose a new metric, Average Mahalanobis Distance (AMD) to tackle this issue. AMD is a metric that quantifies how close the whole generated samples are to the ground truth. We also introduce the Average Maximum Eigenvalue (AMV) metric that quantifies the overall spread of the predictions. Our metrics are validated empirically by showing that the ADE/FDE is not sensitive to distribution shifts, giving a biased sense of accuracy, unlike the AMD/AMV metrics. We introduce the usage of Implicit Maximum Likelihood Estimation (IMLE) as a replacement for traditional generative models to train our model, Social-Implicit. IMLE training mechanism aligns with AMD/AMV objective of predicting trajectories that are close to the ground truth with a tight spread. Social-Implicit is a memory efficient deep model with only 5.8K parameters that runs in real time of about 580Hz and achieves competitive results.



Paperid:916
Authors:Mathis Petrovich; Michael J. Black; Gül Varol
Title: TEMOS: Generating Diverse Human Motions from Textual Descriptions
Abstract:
We address the problem of generating diverse 3D human motions from textual descriptions. This challenging task requires joint modeling of both modalities: understanding and extracting useful human-centric information from the text, and then generating plausible and realistic sequences of human poses. In contrast to most previous work which focuses on generating a single, deterministic, motion from a textual description, we design a variational approach that can produce multiple diverse human motions. We propose TEMOS, a text-conditioned generative model leveraging variational autoencoder (VAE) training with human motion data, in combination with a text encoder that produces distribution parameters compatible with the VAE latent space. We show the TEMOS framework can produce both skeleton-based animations as in prior work, as well more expressive SMPL body motions. We evaluate our approach on the KIT Motion-Language benchmark and, despite being relatively straightforward, demonstrate significant improvements over the state of the art. Code and models are available on our webpage.



Paperid:917
Authors:Siyuan Li; Martin Danelljan; Henghui Ding; Thomas E. Huang; Fisher Yu
Title: Tracking Every Thing in the Wild
Abstract:
Current multi-category Multiple Object Tracking (MOT) metrics use class labels to group tracking results for per-class evaluation. Similarly, MOT methods typically only associate objects with the same class predictions. These two prevalent strategies in MOT implicitly assume that the classification performance is near-perfect. However, this is far from the case in recent large-scale MOT datasets, which contain large numbers of classes with many rare or semantically similar categories. Therefore, the resulting inaccurate classification leads to sub-optimal tracking and inadequate benchmarking of trackers. We address these issues by disentangling classification from tracking. We introduce a new metric, Track Every Thing Accuracy (TETA), breaking tracking measurement into three sub-factors: localization, association, and classification, allowing comprehensive benchmarking of tracking performance even under inaccurate classification. TETA also deals with the challenging incomplete annotation problem in large-scale tracking datasets. We further introduce a Track Every Thing tracker (TETer), that performs association using Class Exemplar Matching (CEM). Our experiments show that TETA evaluates trackers more comprehensively, and TETer achieves significant improvements on the challenging large-scale datasets BDD100K and TAO compared to the state-of-the-art.



Paperid:918
Authors:Soshi Shimada; Vladislav Golyanik; Zhi Li; Patrick Pérez; Weipeng Xu; Christian Theobalt
Title: HULC: 3D HUman Motion Capture with Pose Manifold SampLing and Dense Contact Guidance
Abstract:
Marker-less monocular 3D human motion capture (MoCap) with scene interactions is a challenging research topic relevant for extended reality, robotics and virtual avatar generation. Due to the inherent depth ambiguity of monocular settings, 3D motions captured with existing methods often contain severe artefacts such as incorrect body-scene inter-penetrations, jitter and body floating. To tackle these issues, we propose HULC, a new approach for 3D human MoCap which is aware of the scene geometry. HULC estimates 3D poses and dense body-environment surface contacts for improved 3D localisations, as well as the absolute scale of the subject. Furthermore, we introduce a 3D pose trajectory optimisation based on a novel pose manifold sampling that resolves erroneous body-environment inter-penetrations. Although the proposed method requires less structured inputs compared to existing scene-aware monocular MoCap algorithms, it produces more physically plausible poses: HULC significantly and consistently outperforms the existing approaches in various experiments and on different metrics. Project page: https://vcai.mpi-inf.mpg.de/projects/HULC/"



Paperid:919
Authors:Minji Kim; Seungkwan Lee; Jungseul Ok; Bohyung Han; Minsu Cho
Title: Towards Sequence-Level Training for Visual Tracking
Abstract:
Despite the extensive adoption of machine learning on the task of visual object tracking, recent learning-based approaches have largely overlooked the fact that visual tracking is a sequence-level task in its nature; they rely heavily on frame-level training, which inevitably induces inconsistency between training and testing in terms of both data distributions and task objectives. This work introduces a sequence-level training strategy for visual tracking based on reinforcement learning and discusses how a sequence-level design of data sampling, learning objectives, and data augmentation can improve the accuracy and robustness of tracking algorithms. Our experiments on standard benchmarks including LaSOT, TrackingNet, and GOT-10k demonstrate that four representative tracking models, SiamRPN++, SiamAttn, TransT, and TrDiMP, consistently improve by incorporating the proposed methods in training without modifying architectures.



Paperid:920
Authors:Yunwen Zhou; Abhishek Kar; Eric Turner; Adarsh Kowdle; Chao X. Guo; Ryan C. DuToit; Konstantine Tsotsos
Title: Learned Monocular Depth Priors in Visual-Inertial Initialization
Abstract:
Visual-inertial odometry (VIO) is the pose estimation backbone for most AR/VR and autonomous robotic systems today, in both academia and industry. However, these systems are highly sensitive to the initialization of key parameters such as sensor biases, gravity direction, and metric scale. In practical scenarios where high-parallax or variable acceleration assumptions are rarely met (e.g. hovering aerial robot, smartphone AR user not gesticulating with phone), classical visual-inertial initialization formulations often become ill-conditioned and/or fail to meaningfully converge. In this paper we target visual-inertial initialization specifically for these low-excitation scenarios critical to in-the-wild usage. We propose to circumvent the limitations of classical visual-inertial structure-from-motion (SfM) initialization by incorporating a new learning-based measurement as a higher-level input. We leverage learned monocular depth images (mono-depth) to constrain the relative depth of features, and upgrade the mono-depths to metric scale by jointly optimizing for their scales and shifts. Our experiments show a significant improvement in problem conditioning compared to a classical formulation for visual-inertial initialization, and demonstrate significant accuracy and robustness improvements relative to the state-of-the-art on public benchmarks, particularly under low-excitation scenarios. We further extend this improvement to implementation within an existing odometry system to illustrate the impact of our improved initialization method on resulting tracking trajectories.



Paperid:921
Authors:Matthieu Paul; Martin Danelljan; Christoph Mayer; Luc Van Gool
Title: Robust Visual Tracking by Segmentation
Abstract:
Estimating the target extent poses a fundamental challenge in visual object tracking. Typically, trackers are box-centric and fully rely on a bounding box to define the target in the scene. In practice, objects often have complex shapes and are not aligned with the image axis. In these cases, bounding boxes do not provide an accurate description of the target and often contain a majority of background pixels. We propose a segmentation-centric tracking pipeline that not only produces a highly accurate segmentation mask, but also internally works with segmentation masks instead of bounding boxes. Thus, our tracker is able to better learn a target representation that clearly differentiates the target in the scene from background content. In order to achieve the necessary robustness for the challenging tracking scenario, we propose a separate instance localization component that is used to condition the segmentation decoder when producing the output mask. We infer a bounding box from the segmentation mask, validate our tracker on challenging tracking datasets and achieve the new state of the art on LaSOT with a success AUC score of 69.7%. Since most tracking datasets do not contain mask annotations, we cannot use them to evaluate predicted segmentation masks. Instead, we validate our segmentation quality on two popular video object segmentation datasets. The code and trained models are available at https://github.com/visionml/pytracking.



Paperid:922
Authors:Vojtech Panek; Zuzana Kukelova; Torsten Sattler
Title: MeshLoc: Mesh-Based Visual Localization
Abstract:
Visual localization, i.e., the problem of camera pose estimation, is a central component of applications such as autonomous robots and augmented reality systems. A dominant approach in the literature, shown to scale to large scenes and to handle complex illumination and seasonal changes, is based on local features extracted from images. The scene representation is a sparse Structure-from-Motion point cloud that is tied to a specific local feature. Switching to another feature type requires an expensive feature matching step between the database images used to construct the point cloud. In this work, we thus explore a more flexible alternative based on dense 3D meshes that does not require features matching between database images to build the scene representation. We show that this approach can achieve state-of-the-art results. We further show that surprisingly competitive results can be obtained when extracting features on renderings of these meshes, without any neural rendering stage, and even when rendering raw scene geometry without color or texture. Our results show that dense 3D model-based representations are a promising alternative to existing representations and point to interesting and challenging directions for future research.



Paperid:923
Authors:Yu-Wen Chen; Hsuan-Kung Yang; Chu-Chi Chiu; Chun-Yi Lee
Title: S2F2: Single-Stage Flow Forecasting for Future Multiple Trajectories Prediction
Abstract:
In this work, we present a single-stage framework, named S2F2, for forecasting multiple human trajectories from raw video images by predicting future optical flows. S2F2 differs from the previous two-stage approaches in that it performs detection, Re-ID, and forecasting of multiple pedestrians at the same time. The architecture of S2F2 consists of two primary parts: (1) a context feature extractor responsible for extracting a shared latent feature embedding for performing detection and Re-ID, and (2) a forecasting module responsible for extracting a shared latent feature embedding for forecasting. The outputs of the two parts are then processed to generate the final predicted trajectories of pedestrians. Unlike previous approaches, the computational burden of S2F2 remains consistent even if the number of pedestrians grows. In order to fairly compare S2F2 against the other approaches, we designed a StaticMOT dataset that excludes video sequences involving egocentric motions. The experimental results demonstrate that S2F2 is able to outperform two conventional trajectory forecasting algorithms and a recent learning-based two-stage model, while maintaining tracking performance on par with the contemporary MOT models.



Paperid:924
Authors:Xuhui Tian; Xinran Lin; Fan Zhong; Xueying Qin
Title: Large-Displacement 3D Object Tracking with Hybrid Non-local Optimization
Abstract:
Optimization-based 3D object tracking is known to be precise and fast, but sensitive to large inter-frame displacements due to the local minimum, which usually requires large amount of computation to be overcome. In this paper we propose a fast and effective non-local 3D tracking method. Based on the observation that local minimum are mostly due to the out-of-plane rotation, we propose a hybrid approach combining non-local and local optimizations for different parameters, resulting in efficient non-local search in the 6D pose space. In addition, a precomputed robust contour-based tracking method is proposed for the local optimization. By using long search lines with multiple candidate correspondences, it can better adaptive to different frame displacements without the need of coarse-to-fine search. After the pre-computation, pose updates can be conducted very fast, enabling the non-local optimization in real time. Our method outperforms all previous methods for both small and large displacements. For large displacements, the accuracy is greatly improved (81.7% v.s.19.4%). At the same time, real-time speed (>50fps) can be achieved with only CPU.The source code is available at https://github.com/cvbubbles/nonlocal-3dtracking.



Paperid:925
Authors:Vasyl Borsuk; Roman Vei; Orest Kupyn; Tetiana Martyniuk; Igor Krashenyi; Jiři Matas
Title: "FEAR: Fast, Efficient, Accurate and Robust Visual Tracker"
Abstract:
We present FEAR, a family of fast, efficient, accurate, and robust Siamese visual trackers. We present a novel and efficient way to benefit from dual-template representation for object model adaption, which incorporates temporal information with only a single learnable parameter. We further improve the tracker architecture with a pixel-wise fusion block. By plugging-in sophisticated backbones with the abovementioned modules, FEAR-M and FEAR-L trackers surpass most Siamese trackers on several academic benchmarks in both accuracy and efficiency. Employed with the lightweight backbone, the optimized version FEAR-XS offers more than 10 times faster tracking than current Siamese trackers while maintaining near state-of-the-art results. FEAR-XS tracker is 2.4x smaller and 4.3x faster than LightTrack with superior accuracy. In addition, we expand the definition of the model efficiency by introducing FEAR benchmark that assesses energy consumption and execution speed. We show that energy consumption is a limiting factor for trackers on mobile devices. Source code, pretrained models, and evaluation protocol are available at https://github.com/PinataFarms/FEARTracker.



Paperid:926
Authors:Liangchen Song; Xuan Gong; Benjamin Planche; Meng Zheng; David Doermann; Junsong Yuan; Terrence Chen; Ziyan Wu
Title: PREF: Predictability Regularized Neural Motion Fields
Abstract:
Knowing the 3D motions in a dynamic scene is essential to many vision applications. Recent progress is mainly focused on estimating the activity of some specific elements like humans. In this paper, we leverage a neural motion field for estimating the motion of all points in a multiview setting. Modeling the motion from a dynamic scene with multiview data is challenging due to the ambiguities in points of similar color and points with time-varying color. We propose to regularize the estimated motion to be predictable. If the motion from previous frames is known, then the motion in the near future should be predictable. Therefore, we introduce a predictability regularization by first conditioning the estimated motion on latent embeddings, then by adopting a predictor network to enforce predictability on the embeddings. The proposed framework PREF (Predictability REgularized Fields) achieves on par or better results than state-of-the-art neural motion field-based dynamic scene representation methods, while requiring no prior knowledge of the scene.



Paperid:927
Authors:Conghao Wong; Beihao Xia; Ziming Hong; Qinmu Peng; Wei Yuan; Qiong Cao; Yibo Yang; Xinge You
Title: View Vertically: A Hierarchical Network for Trajectory Prediction via Fourier Spectrums
Abstract:
Understanding and forecasting future trajectories of agents are critical for behavior analysis, robot navigation, autonomous cars, and other related applications. Previous methods mostly treat trajectory prediction as time sequence generation. Different from them, this work studies agents’ trajectories in a ""vertical"" view, i.e., modeling and forecasting trajectories from the spectral domain. Different frequency bands in the trajectory spectrums could hierarchically reflect agents’ motion preferences at different scales. The low-frequency and high-frequency portions could represent their coarse motion trends and fine motion variations, respectively. Accordingly, we propose a hierarchical network V$^2$-Net, which contains two sub-networks, to hierarchically model and predict agents’ trajectories with trajectory spectrums. The coarse-level keypoints estimation sub-network first predicts the ""minimal"" spectrums of agents’ trajectories on several ""key"" frequency portions. Then the fine-level spectrum interpolation sub-network interpolates the spectrums to reconstruct the final predictions. Experimental results display the competitiveness and superiority of V$^2$-Net on both ETH-UCY benchmark and the Stanford Drone Dataset.



Paperid:928
Authors:Haoxian Zhang; Yonggen Ling
Title: "HVC-Net: Unifying Homography, Visibility, and Confidence Learning for Planar Object Tracking"
Abstract:
Robust and accurate planar tracking over a whole video sequence is vitally important for many vision applications. The key to planar object tracking is to find object correspondences, modeled by homography, between the reference image and the tracked image. Existing methods tend to obtain wrong correspondences with changing appearance variations, camera-object relative motions and occlusions. To alleviate this problem, we present a unified convolutional neural network (CNN) model that jointly considers homography, visibility, and confidence. First, we introduce correlation blocks that explicitly account for the local appearance changes and camera-object relative motions as the base of our model. Second, we jointly learn the homography and visibility that links camera-object relative motions with occlusions. Third, we propose a confidence module that actively monitors the estimation quality from the pixel correlation distributions obtained in correlation blocks. All these modules are plugged into a Lucas-Kanade (LK) tracking pipeline to obtain both accurate and robust planar object tracking. Our approach outperforms the state-of-the-art methods on public POT and TMT datasets. Its superior performance is also verified on a real-world application, synthesizing high-quality in-video advertisements.



Paperid:929
Authors:Jianfeng Xiang; Junliang Chen; Wenshuang Liu; Xianxu Hou; Linlin Shen
Title: RamGAN: Region Attentive Morphing GAN for Region-Level Makeup Transfer
Abstract:
In this paper, we propose a region adaptive makeup transfer GAN, called RamGAN, for precise region-level makeup transfer. Compared to face-level transfer methods, our RamGAN uses spatial-aware Region Attentive Morphing Module (RAMM) to encode Region Attentive Matrices (RAMs) for local regions like lips, eye shadow and skin. After that, the Region Style Injection Module (RSIM) is applied to RAMs produced by RAMM to obtain two Region Makeup Tensors, gamma and beta, which are subsequently added to the feature map of source image to transfer the makeup. As attention and makeup styles are calculated for each region, RamGAN can achieve better disentangled makeup transfer for different facial regions. When there are significant pose and expression variations between source and reference, RamGAN can also achieve better transfer results, due to the integration of spatial information and region-level correspondence. Experimental results are conducted on public datasets like MT, M-Wild and Makeup datasets, both visual and quantitative results and user study suggest that our approach achieves better transfer results than state-of-the-art methods like BeautyGAN, BeautyGlow, DMT, CPM and PSGAN.



Paperid:930
Authors:Dejia Xu; Yifan Jiang; Peihao Wang; Zhiwen Fan; Humphrey Shi; Zhangyang Wang
Title: SinNeRF: Training Neural Radiance Fields on Complex Scenes from a Single Image
Abstract:
Despite the rapid development of Neural Radiance Field (NeRF), the necessity of dense covers largely prohibits its wider applications. While several recent works have attempted to address this issue, they either operate with sparse views (yet still, a few of them) or on simple objects/scenes. In this work, we consider a more ambitious task: training neural radiance field, over realistically complex visual scenes, by “looking only once”, i.e., using only a single view. To attain this goal, we present a Single View NeRF (SinNeRF) framework consisting of thoughtfully designed semantic and geometry regularizations. Specifically, SinNeRF constructs a semi-supervised learning process, where we introduce and propagate geometry pseudo labels and semantic pseudo labels to guide the progressive training process. Extensive experiments are conducted on complex scene benchmarks, including NeRF synthetic dataset, Local Light Field Fusion dataset, and DTU dataset. We show that even without pre-training on multi-view datasets, SinNeRF can yield photo-realistic novel-view synthesis results. Under the single image setting, SinNeRF significantly outperforms the current state-of-the-art NeRF baselines in all cases. Project page: https://vita-group.github.io/SinNeRF/"



Paperid:931
Authors:Guangcong Zheng; Shengming Li; Hui Wang; Taiping Yao; Yang Chen; Shouhong Ding; Xi Li
Title: Entropy-Driven Sampling and Training Scheme for Conditional Diffusion Generation
Abstract:
Denoising Diffusion Probabilistic Model (DDPM) is able to make flexible conditional image generation from prior noise to real data, by introducing an independent noise-aware classifier to provide conditional gradient guidance at each time step of denoising process. However, due to the ability of the classifier to easily discriminate an incompletely generated image only with high-level structure, the gradient, which is a kind of class information guidance, tends to vanish early, leading to the collapse from conditional generation process into the unconditional process. To address this problem, we propose two simple but effective approaches from two perspectives. For sampling procedure, we introduce the entropy of predicted distribution as the measure of guidance vanishing level and propose an entropy-aware scaling method to adaptively recover the conditional semantic guidance. For the training stage, we propose the entropy-aware optimization objectives to alleviate the overconfident prediction for noisy data. On ImageNet1000 256x256, with our proposed sampling scheme and trained classifier, the pretrained conditional and unconditional DDPM model can achieve 10.89% (4.59 to 4.09) and 43.5% (12 to 6.78) FID improvement, respectively. Code is available at https://github.com/ZGCTroy/ED-DPM.



Paperid:932
Authors:Hengyuan Ma; Li Zhang; Xiatian Zhu; Jianfeng Feng
Title: Accelerating Score-Based Generative Models with Preconditioned Diffusion Sampling
Abstract:
Score-based generative models (SGMs) have recently emerged as a promising class of generative models. However, a fundamental limitation is that their inference is very slow due to a need for many (e.g., 2000) iterations of sequential computations. An intuitive acceleration method is to reduce the sampling iterations which however causes severe performance degradation. We investigate this problem by viewing the diffusion sampling process as a Metropolis adjusted Langevin algorithm, which helps reveal the underlying cause to be ill-conditioned curvature. Under this insight, we propose a model-agnostic preconditioned diffusion sampling (PDS) method that leverages matrix preconditioning to alleviate the aforementioned problem. Crucially, PDS is proven theoretically to converge to the original target distribution of a SGM, no need for retraining. Extensive experiments on three image datasets with a variety of resolutions and diversity validate that PDS consistently accelerates off-the-shelf SGMs whilst maintaining the synthesis quality. In particular, PDS can accelerate by up to 29x on more challenging high resolution (1024x1024) image generation.



Paperid:933
Authors:Vlas Zyrianov; Xiyue Zhu; Shenlong Wang
Title: Learning to Generate Realistic LiDAR Point Clouds
Abstract:
We present LiDARGen, a novel, effective, and controllable generative model that produces realistic LiDAR point cloud sensory readings. Our method leverages the powerful score-matching energy-based model and formulates the point cloud generation process as a stochastic denoising process in the equirectangular view. This model allows us to sample diverse and high-quality point cloud samples with guaranteed physical feasibility and controllability. We validate the effectiveness of our method on the challenging KITTI-360 and NuScenes datasets. The quantitative and qualitative results show that our approach produces more realistic samples than other generative models. Furthermore, LiDARGen can sample point clouds conditioned on inputs without retraining. We demonstrate that our proposed generative model could be directly used to densify LiDAR point clouds.



Paperid:934
Authors:Tuan-Anh Vu; Thanh Nguyen; Binh-Son Hua; Quang-Hieu Pham; Sai-Kit Yeung
Title: RFNet-4D: Joint Object Reconstruction and Flow Estimation from 4D Point Clouds
Abstract:
Object reconstruction from 3D point clouds has achieved impressive progress in the computer vision and computer graphics research field. However, reconstruction from time-varying point clouds (a.k.a. 4D point clouds) is generally overlooked. In this paper, we propose a new network architecture, namely RFNet-4D, that jointly reconstruct objects and their motion flows from 4D point clouds. The key insight is that simultaneously performing both tasks via learning spatial and temporal features from a sequence of point clouds can leverage individual tasks, leading to improved overall performance. To prove this ability, we design a temporal vector field learning module using unsupervised learning approach for flow estimation, leveraged by supervised learning of spatial structures for object reconstruction. Extensive experiments and analyses on benchmark dataset validated the effectiveness and efficiency of our method. As shown in experimental results, our method achieves state-of-the-art performance on both flow estimation and object reconstruction while performing much faster than existing methods in both training and inference. Our code and data are available at https://github.com/hkust-vgd/RFNet-4D.



Paperid:935
Authors:Cairong Wang; Yiming Zhu; Chun Yuan
Title: Diverse Image Inpainting with Normalizing Flow
Abstract:
Image Inpainting is an ill-posed problem since there are diverse possible counterparts for the missing areas. The challenge of inpainting is to keep the ""corrupted region"" content consistent with the background and generate a variety of reasonable texture details. However, existing one-stage methods that directly output the inpainting results have to make a trade-off between diversity and consistency. The two-stage methods as the current trend can circumvent such shortcomings. These methods predict diverse structural priors in the first stage and focus on rich texture details generation in the second stage. However, all two-stage methods require autoregressive models to predict the probability distribution of the structural priors, which significantly limits the inference speed. In addition, their discretization assumption of prior distribution reduces the diversity of the inpainting results. We propose Flow-Fill, a novel two-stage image inpainting framework that utilizes a conditional normalizing flow model to generate diverse structural priors in the first stage. Flow-Fill can directly estimate the joint probability density of the missing regions as a flow-based model without reasoning pixel by pixel. Hence it achieves real-time inference speed and eliminates discretization assumptions. In addition, as a reversible model, Flow-Fill can invert the latent variables for a specified region, which allows us to make the inference process as semantic image editing. Experiments on benchmark datasets validate that Flow-Fill achieves superior diversity and fidelity in image inpainting qualitatively and quantitatively.



Paperid:936
Authors:José Lezama; Huiwen Chang; Lu Jiang; Irfan Essa
Title: Improved Masked Image Generation with Token-Critic
Abstract:
Non-autoregressive generative transformers recently demonstrated impressive image generation performance, and orders of magnitude faster sampling than their autoregressive counterparts. However, optimal parallel sampling from the true joint distribution of visual tokens remains an open challenge. In this paper we introduce Token-Critic, an auxiliary model to guide the sampling of a non-autoregressive generative transformer. Given a masked-and-reconstructed real image, the Token-Critic model is trained to distinguish which visual tokens belong to the original image and which were sampled by the generative transformer. During non-autoregressive iterative sampling, Token-Critic is used to select which tokens to accept and which to reject and resample. Coupled with Token-Critic, a state-of-the-art generative transformer significantly improves its performance, and outperforms recent diffusion models and GANs in terms of the trade-off between generated image quality and diversity, in the challenging class-conditional ImageNet generation.



Paperid:937
Authors:Junghyuk Lee; Jong-Seok Lee
Title: TREND: Truncated Generalized Normal Density Estimation of Inception Embeddings for GAN Evaluation
Abstract:
Evaluating image generation models such as generative adversarial networks (GANs) is a challenging problem. A common approach is to compare the distributions of the set of ground truth images and the set of generated test images. The Frechet Inception distance is one of the most widely used metrics for evaluation of GANs, which assumes that the features from a trained Inception model for a set of images follow a normal distribution. In this paper, we argue that this is an over-simplified assumption, which may lead to unreliable evaluation results, and more accurate density estimation can be achieved using a truncated generalized normal distribution. Based on this, we propose a novel metric for accurate evaluation of GANs, named TREND (TRuncated gEneralized Normal Density estimation of inception embeddings). We demonstrate that our approach significantly reduces errors of density estimation, which consequently eliminates the risk of faulty evaluation results. Furthermore, we show that the proposed metric significantly improves robustness of evaluation results against variation of the number of image samples.



Paperid:938
Authors:Zikun Chen; Ruowei Jiang; Brendan Duke; Han Zhao; Parham Aarabi
Title: Exploring Gradient-Based Multi-directional Controls in GANs
Abstract:
Generative Adversarial Networks (GANs) have been widely applied in modeling diverse image distributions. However, despite its impressive applications, the structure of the latent space in GANs largely remains as a black-box, leaving its controllable generation an open problem, especially when spurious correlations between different semantic attributes exist in the image distributions. To address this problem, previous methods typically learn linear directions or individual channels that control semantic attributes in the image space. However, they often suffer from imperfect disentanglement, or are unable to obtain multi-directional controls. In this work, in light of the above challenges, we propose a novel approach that discovers nonlinear controls, which enables multi-directional manipulation as well as effective disentanglement, based on gradient information in the learned GAN latent space. More specifically, we first learn interpolation directions by following the gradients from classification networks trained separately on the attributes, and then navigate the latent space by exclusively controlling channels activated for the target attribute in the learned directions. Empirically, with small training data, our approach is able to gain fine-grained controls over a diverse set of bi-directional and multi-directional attributes, and we showcase its ability to achieve disentanglement significantly better than state-of-the-art methods both qualitatively and quantitatively.



Paperid:939
Authors:Tianyu Wang; Miaomiao Liu; Kee Siong Ng
Title: Spatially Invariant Unsupervised 3D Object-Centric Learning and Scene Decomposition
Abstract:
We tackle the problem of object-centric learning on point clouds, which is crucial for high-level relational reasoning and scalable machine intelligence. In particular, we introduce a framework, SPAIR3D, to factorize a 3D point cloud into a spatial mixture model where each component corresponds to one object. To model the spatial mixture model on point clouds, we derive the Chamfer Mixture Loss, which fits naturally into our variational training pipeline. Moreover, we adopt an object-specification scheme that describes each object’s location relative to its local voxel grid cell. Such a scheme allows SPAIR3D to model scenes with an arbitrary number of objects. We evaluate our method on the task of unsupervised scene decomposition. Experimental results demonstrate that SPAIR3D has strong scalability and is capable of detecting and segmenting an unknown number of objects from a point cloud in an unsupervised manner.



Paperid:940
Authors:Hong-Wing Pang; Yingshu Chen; Phuoc-Hieu Le; Binh-Son Hua; Thanh Nguyen; Sai-Kit Yeung
Title: Neural Scene Decoration from a Single Photograph
Abstract:
Furnishing and rendering indoor scenes has been a long-standing task for interior design, where artists create a conceptual design for the space, build a 3D model of the space, decorate, and then perform rendering. Although the task is important, it is tedious and requires tremendous effort. In this paper, we introduce a new problem of domain-specific indoor scene image synthesis, namely neural scene decoration. Given a photograph of an empty indoor space and a list of decorations with layout determined by user, we aim to synthesize a new image of the same space with desired furnishing and decorations. Neural scene decoration can be applied to create conceptual interior designs in a simple yet effective manner. Our attempt to this research problem is a novel scene generation architecture that transforms an empty scene and an object layout into a realistic furnished scene photograph. We demonstrate the performance of our proposed method by comparing it with conditional image synthesis baselines built upon prevailing image translation approaches both qualitatively and quantitatively. We conduct extensive experiments to further validate the plausibility and aesthetics of our generated scenes.



Paperid:941
Authors:Kai Yao; Penglei Gao; Xi Yang; Jie Sun; Rui Zhang; Kaizhu Huang
Title: Outpainting by Queries
Abstract:
Image outpainting, which is well studied with Convolution Neural Network (CNN) based framework, has recently drawn more attention in computer vision. However, CNNs rely on inherent inductive biases to achieve effective sample learning, which may degrade the performance ceiling. In this paper, motivated by the flexible self-attention mechanism with minimal inductive biases in transformer architecture, we reframe the generalised image outpainting problem as a patch-wise sequence-to-sequence autoregression problem, enabling query-based image outpainting. Specifically, we propose a novel hybrid vision-transformer-based encoder-decoder framework, named Query Outpainting TRansformer (QueryOTR), for extrapolating visual context all-side around a given image. Patch-wise mode’s global modeling capacity allows us to extrapolate images from the attention mechanism’s query standpoint. A novel Query Expansion Module (QEM) is designed to integrate information from the predicted queries based on the encoder’s output, hence accelerating the convergence of the pure transformer even with a relatively small dataset. To further enhance connectivity between each patch, the proposed Patch Smoothing Module (PSM) re-allocates and averages the overlapped regions, thus providing seamless predicted images. We experimentally show that QueryOTR could generate visually appealing results smoothly and realistically against the state-of-the-art image outpainting approaches.



Paperid:942
Authors:Sam Bond-Taylor; Peter Hessey; Hiroshi Sasaki; Toby P. Breckon; Chris G. Willcocks
Title: Unleashing Transformers: Parallel Token Prediction with Discrete Absorbing Diffusion for Fast High-Resolution Image Generation from Vector-Quantized Codes
Abstract:
Whilst diffusion probabilistic models can generate high quality image content, key limitations remain in terms of both generating high-resolution imagery and their associated high computational requirements. Recent Vector-Quantized image models have overcome this limitation of image resolution but are prohibitively slow and unidirectional as they generate tokens via element-wise autoregressive sampling from the prior. By contrast, in this paper we propose a novel discrete diffusion probabilistic model prior which enables parallel prediction of Vector-Quantized tokens by using an unconstrained Transformer architecture as the backbone. During training, tokens are randomly masked in an order-agnostic manner and the Transformer learns to predict the original tokens. This parallelism of Vector-Quantized token prediction in turn facilitates unconditional generation of globally consistent high-resolution and diverse imagery at a fraction of the computational expense. In this manner, we can generate image resolutions exceeding that of the original training set samples whilst additionally provisioning per-image likelihood estimates (in a departure from generative adversarial approaches). Our approach achieves state-of-the-art results in terms of the manifold overlap metrics Coverage (LSUN Bedroom: 0.83; LSUN Churches: 0.73; FFHQ: 0.80) and Density (LSUN Bedroom: 1.51; LSUN Churches: 1.12; FFHQ: 1.20), and performs competitively on FID (LSUN Bedroom: 3.64; LSUN Churches: 4.07; FFHQ: 6.11) whilst offering advantages in terms of both computation and reduced training set requirements.



Paperid:943
Authors:Adéla Šubrtová; David Futschik; Jan Čech; Michal Lukáč; Eli Shechtman; Daniel Sýkora
Title: ChunkyGAN: Real Image Inversion via Segments
Abstract:
We present ChunkyGAN-a novel paradigm for modeling and editing images using generative adversarial networks. Unlike previous techniques seeking a global latent representation of the input image, our approach subdivides the input image into a set of smaller components (chunks) specified either manually or automatically using a pre-trained segmentation network. For each chunk, the latent code of a generative network is estimated locally with greater accuracy thanks to a smaller number of constraints. Moreover, during the optimization of latent codes, segmentation can further be refined to improve matching quality. This process enables high-quality projection of the original image with spatial disentanglement that previous methods would find challenging to achieve. To demonstrate the advantage of our approach, we evaluated it quantitatively and also qualitatively in various image editing scenarios that benefit from the higher reconstruction quality and local nature of the approach. Our method is flexible enough to manipulate even out-of-domain images that would be hard to reconstruct using global techniques.



Paperid:944
Authors:Omri Avrahami; Dani Lischinski; Ohad Fried
Title: GAN Cocktail: Mixing GANs without Dataset Access
Abstract:
Today’s generative models are capable of synthesizing high-fidelity images, but each model specializes on a specific target domain. This raises the need for model merging: combining two or more pretrained generative models into a single unified one. In this work we tackle the problem of model merging, given two constraints that often come up in the real world: (1) no access to the original training data, and (2) without increasing the size of the neural network. To the best of our knowledge, model merging under these constraints has not been studied thus far. We propose a novel, two-stage solution. In the first stage, we transform the weights of all the models to the same parameter space by a technique we term model rooting. In the second stage, we merge the rooted models by averaging their weights and fine-tuning them for each specific domain, using only data generated by the original trained models. We demonstrate that our approach is superior to baseline methods and to existing transfer learning techniques, and investigate several applications.



Paperid:945
Authors:Mingfei Chen; Jianfeng Zhang; Xiangyu Xu; Lijuan Liu; Yujun Cai; Jiashi Feng; Shuicheng Yan
Title: Geometry-Guided Progressive NeRF for Generalizable and Efficient Neural Human Rendering
Abstract:
In this work we develop a generalizable and efficient Neural Radiance Field (NeRF) pipeline for high-fidelity free-viewpoint human body synthesis under settings with sparse camera views. Though existing NeRF-based methods can synthesize rather realistic details for human body, they tend to produce poor results when the input has self-occlusion, especially for unseen humans under sparse views. Moreover, these methods often require a large number of sampling points for rendering, which leads to low efficiency and limits their real-world applicability. To address these challenges, we propose a Geometry-guided Progressive NeRF (GP-NeRF). In particular, to better tackle self-occlusion, we devise a geometry-guided multi-view feature integration approach that utilizes the estimated geometry prior to integrate the incomplete information from input views and construct a complete geometry volume for the target human body. Meanwhile, for achieving higher rendering efficiency, we introduce a geometry-guided progressive rendering pipeline, which leverages the geometric feature volume and the predicted density values to progressively reduce the number of sampling points and speed up the rendering process. Experiments on the ZJU-MoCap and THUman datasets show that our method outperforms the state-of-the-arts significantly across multiple generalization settings, while the time cost is reduced via >70% via applying our efficient progressive rendering pipeline.



Paperid:946
Authors:Yichen Sheng; Yifan Liu; Jianming Zhang; Wei Yin; A. Cengiz Oztireli; He Zhang; Zhe Lin; Eli Shechtman; Bedrich Benes
Title: Controllable Shadow Generation Using Pixel Height Maps
Abstract:
Shadows are essential for realistic image compositing. Physics based shadow rendering methods require 3D geometries, which are not always available. Deep learning-based shadow synthesis methods learn a mapping from the light information to an object’s shadow without explicitly modeling the shadow geometry. Still, they lack control and are prone to visual artifacts. We introduce “Pixel Height”, a novel geometry representation that encodes the correlations between objects, ground, and camera pose. The Pixel Height can be calculated from 3D geometries, manually annotated on 2D images, and it can also be predicted from a single-view RGB image by a supervised approach. It can be used to calculate hard shadows in a 2D image based on the projective geometry, providing precise control of the shadows’ direction and shape. Furthermore, we propose a data-driven soft shadow generator to apply softness to a hard shadow based on a softness input parameter. Qualitative and quantitative evaluations demonstrate that the proposed Pixel Height significantly improves the quality of the shadow generation while allowing for controllability.



Paperid:947
Authors:Jovita Lukasik; Steffen Jung; Margret Keuper
Title: Learning Where to Look – Generative NAS Is Surprisingly Efficient
Abstract:
The efficient, automated search for well-performing neural architectures (NAS) has drawn increasing attention in the recent past. Thereby, the predominant research objective is to reduce the necessity of costly evaluations of neural architectures while efficiently exploring large search spaces. To this aim, surrogate models embed architectures in a latent space and predict their performance, while generative models for neural architectures enable optimization-based search within the latent space the generator draws from. Both, surrogate and generative models, have the aim of facilitating query-efficient search in a well-structured latent space. In this paper, we further improve the trade-off between query-efficiency and promising architecture generation by leveraging advantages from both, efficient surrogate models and generative design. To this end, we propose a generative model, paired with a surrogate predictor, that iteratively learns to generate samples from increasingly promising latent subspaces. This approach leads to very effective and efficient architecture search, while keeping the query amount low. In addition, our approach allows in a straightforward manner to jointly optimize for multiple objectives such as accuracy and hardware latency. We show the benefit of this approach not only w.r.t. the optimization of architectures for highest classification accuracy but also in the context of hardware constraints and outperform state-of-the-art methods on several NAS benchmarks for single and multiple objectives. We also achieve state-of-the-art performance on ImageNet. The code is available at https://github.com/jovitalukasik/AG-Net.



Paperid:948
Authors:Bowen Jing; Gabriele Corso; Renato Berlinghieri; Tommi Jaakkola
Title: Subspace Diffusion Generative Models
Abstract:
Score-based models generate samples by mapping noise to data (and vice versa) via a high-dimensional diffusion process. We question whether it is necessary to run this entire process at high dimensionality and incur all the inconveniences thereof. Instead, we restrict the diffusion via projections onto subspaces as the data distribution evolves toward noise. When applied to state-of-the-art models, our framework simultaneously improves sample quality---reaching an FID of 2.17 on unconditional CIFAR-10---and reduces the computational cost of inference for the same number of denoising steps. Our framework is fully compatible with continuous-time diffusion and retains its flexible capabilities, including exact log-likelihoods and controllable generation. Code is available at https://github.com/bjing2016/subspace-diffusion.



Paperid:949
Authors:Jiaheng Wei; Minghao Liu; Jiahao Luo; Andrew Zhu; James Davis; Yang Liu
Title: DuelGAN: A Duel between Two Discriminators Stabilizes the GAN Training
Abstract:
In this paper, we introduce DuelGAN, a generative adversarial network (GAN) solution to improve the stability of the generated samples and to mitigate mode collapse. Built upon the Vanilla GAN’s two-player game between the discriminator D_1 and the generator G, we introduce a peer discriminator D_2 to the min-max game. Similar to previous work using two discriminators, the first role of both D_1, D_2 is to distinguish between generated samples and real ones, while the generator tries to generate high-quality samples which are able to fool both discriminators. Different from existing methods, we introduce a duel between D_1 and D_2 to discourage their agreement and therefore increase the level of diversity of the generated samples. This property alleviates the issue of early mode collapse by preventing D_1 and D_2 from converging too fast. We provide theoretical analysis for the equilibrium of the min-max game formed among G, D_1, D_2. We offer convergence behavior of DuelGAN as well as stability of the min-max game. It’s worth mentioning that DuelGAN operates in the unsupervised setting, and the duel between D_1 and D_2 does not need any label supervision. Experiments results on a synthetic dataset and on real-world image datasets (MNIST, Fashion MNIST, CIFAR-10, STL-10, CelebA, VGG) demonstrate that DuelGAN outperforms competitive baseline work in generating diverse and high-quality samples, while only introduces negligible computation cost. Our code is publicly available at https://github.com/UCSC-REAL/DuelGAN.



Paperid:950
Authors:Vishwanath Saragadam; Jasper Tan; Guha Balakrishnan; Richard G. Baraniuk; Ashok Veeraraghavan
Title: MINER: Multiscale Implicit Neural Representation
Abstract:
We introduce a new neural signal model designed for efficient high-resolution representation of large-scale signals. The key innovation in our multiscale implicit neural representation (MINER) is an internal representation via a Laplacian pyramid, which provides a sparse multiscale decomposition of the signal that captures orthogonal parts of the signal across scales. We leverage the advantages of the Laplacian pyramid by representing small disjoint patches of the pyramid at each scale with a small MLP. This enables the capacity of the network to adaptively increase from coarse to fine scales, and only represent parts of the signal with strong signal energy. The parameters of each MLP are optimized from coarse-to-fine scale which results in faster approximations at coarser scales, thereby ultimately an extremely fast training process. We apply MINER to a range of large-scale signal representation tasks, including gigapixel images and very large point clouds, and demonstrate that it requires fewer than 25% of the parameters, 33% of the memory footprint, and 10% of the computation time of competing techniques such as ACORN to reach the same representation accuracy.



Paperid:951
Authors:Hongwei Yong; Lei Zhang
Title: An Embedded Feature Whitening Approach to Deep Neural Network Optimization
Abstract:
Compared with the feature normalization methods that are widely used in deep neural network (DNN) training, feature whitening methods take the correlation of features into consideration, which can help to learn more effective features. However, existing feature whitening methods have a few limitations, such as the large computation and memory cost, incapability to adopt pre-trained DNN models, the introduction of additional parameters, etc., making them impractical to use in optimizing DNNs. To overcome these drawbacks, we propose a novel Embedded Feature Whitening (EFW) approach to DNN optimization. EFW only adjusts the gradient of weight by using the whitening matrix without changing any part of the network so that it can be easily adopted to optimize pre-trained and well-defined DNN architectures. We consequently develop the associated momentum, adaptive damping and gradient norm recovery techniques w.r.t. EFW, which can be implemented efficiently with acceptable extra computation and memory cost. We apply EFW to the two most commonly used DNN optimizers, i.e., SGDM and Adam, and name them W-SGDM and W-Adam. Extensive experimental results on various vision tasks, including image classification, object detection, segmentation and person ReID, demonstrate the superiority of W-SGDM and W-Adam to their original counterparts.



Paperid:952
Authors:Alp Yurtsever; Tolga Birdal; Vladislav Golyanik
Title: Q-FW: A Hybrid Classical-Quantum Frank-Wolfe for Quadratic Binary Optimization
Abstract:
We present a hybrid classical-quantum framework based on the Frank-Wolfe algorithm, Q-FW, for solving quadratic, linearly-constrained, binary optimization problems on quantum annealers (QA). The computational premise of quantum computers has cultivated the re-design of various existing vision problems into quantum-friendly forms. Experimental QA realisations can solve a particular non-convex problem known as the quadratic unconstrained binary optimization (QUBO). Yet a naive-QUBO cannot take into account the restrictions on the parameters. To introduce additional structure in the parameter space, researchers have crafted ad-hoc solutions incorporating (linear) constraints in the form of regularizers. However, this comes at the expense of a hyper-parameter, balancing the impact of regularization. To date, a true constrained solver of quadratic binary optimization (QBO) problems has lacked. Q-FW first reformulates constrained-QBO as a copositive program (CP), then employs Frank-Wolfe iterations to solve CP while satisfying linear (in)equality constraints. This procedure unrolls the original constrained-QBO into a set of unconstrained QUBOs all of which are solved, in a sequel, on a QA. We use D-Wave Advantage QA to conduct synthetic and real experiments on two important computer vision problems, graph matching and permutation synchronization, which demonstrate that our approach is effective in alleviating the need for an explicit regularization coefficient.



Paperid:953
Authors:Chang Liu; Shaofeng Zhang; Xiaokang Yang; Junchi Yan
Title: Self-Supervised Learning of Visual Graph Matching
Abstract:
Despite the rapid progress made by existing graph matching methods, expensive or even unrealistic node-level correspondence labels are often required. Inspired by recent progress in self-supervised contrastive learning, we propose an end-to-end label-free self-supervised contrastive graph matching framework (SCGM). Unlike in vision tasks like classification and segmentation, where the backbone is often forced to extract object instance-level or pixel-level information, we design an extra objective function at node-level on graph data which also considers both the visual appearance and graph structure by node embedding. Further, we propose two-stage augmentation functions on both raw images and extracted graphs to increase the variance, which has been shown effective in self-supervised learning. We conduct experiments on standard graph matching benchmarks, where our method boosts previous state-of-the-arts under both label-free self-supervised and fine-tune settings. Without the ground truth labels for node matching nor the graph/image-level category information, our proposed framework SCGM outperforms several deep graph matching methods. By proper fine-tuning, SCGM can surpass the state-of-the-art supervised deep graph matching methods.



Paperid:954
Authors:Xuxi Chen; Tianlong Chen; Yu Cheng; Weizhu Chen; Ahmed Awadallah; Zhangyang Wang
Title: Scalable Learning to Optimize: A Learned Optimizer Can Train Big Models
Abstract:
Learning to optimize (L2O) has gained increasing attention since it demonstrates a promising path to automating and accelerating the optimization of complicated problems. Unlike manually crafted classical optimizers, L2O parameterizes and learns optimization rules in a data-driven fashion. However, the primary barrier, scalability, persists for this paradigm: as the typical L2O models create massive memory overhead due to unrolled computational graphs, it disables L2O’s applicability to large-scale tasks. To overcome this core challenge, we propose a new scalable learning to optimize (SL2O) framework which (i) first constrains the network updates in a tiny subspace and (ii) then explores learning rules on top of it. Thanks to substantially reduced trainable parameters, learning optimizers for large-scale networks with a single GPU become feasible for the first time, showing that the scalability roadblock of applying L2O to training large models is now removed. Comprehensive experiments on various network architectures (i.e., ResNets, VGGs, ViTs) and datasets (i.e., CIFAR, ImageNet, E2E) across vision and language tasks, consistently validate that SL2O can achieve significantly faster convergence speed and competitive performance compared to analytical optimizers. For example, our approach converges 3.41 4.60 times faster on CIFAR-10/100 with ResNet-18, and 1.24 times faster on ViTs, at nearly no performance loss. Codes are included in the supplement.



Paperid:955
Authors:Gang-Xuan Lin; Shih-Wei Hu; Chun-Shien Lu
Title: QISTA-ImageNet: A Deep Compressive Image Sensing Framework Solving $\ell_q$-Norm Optimization Problem
Abstract:
In this paper, we study how to reconstruct the original images from the given sensed samples/measurements by proposing a so-called deep compressive image sensing framework. This framework, dubbed QISTA-ImageNet, is built upon a deep neural network to realize our optimization algorithm QISTA (Lq-ISTA) in solving image recovery problem. The unique characteristics of QISTA-ImageNet are that we (1) introduce a generalized proximal operator and present learning-based proximal gradient descent (PGD) together with an iterative algorithm in reconstructing images, (2) analyze how QISTA-ImageNet can exhibit better solutions compared to state-of-the-art methods and interpret clearly the insight of proposed method, and (3) conduct empirical comparisons with state-of-the-art methods to demonstrate that QISTA-ImageNet exhibits the best performance in terms of image reconstruction quality to solve the Lq-norm optimization problem.



Paperid:956
Authors:Qiankun Gao; Chen Zhao; Bernard Ghanem; Jian Zhang
Title: R-DFCIL: Relation-Guided Representation Learning for Data-Free Class Incremental Learning
Abstract:
Class-Incremental Learning (CIL) struggles with catastrophic forgetting when learning new knowledge, and Data-Free CIL (DFCIL) is even more challenging without access to the training data of previously learned classes. Though recent DFCIL works introduce techniques such as model inversion to synthesize data for previous classes, they fail to overcome forgetting due to the severe domain gap between the synthetic and real data. To address this issue, this paper proposes relation-guided representation learning (RRL) for DFCIL, dubbed R-DFCIL. In RRL, we introduce relational knowledge distillation to flexibly transfer the structural relation of new data from the old model to the current model. Our RRL-boosted DFCIL can guide the current model to learn representations of new classes better compatible with representations of previous classes, which greatly reduces forgetting while improving plasticity. To avoid the mutual interference between representation and classifier learning, we employ local rather than global classification loss during RRL. After RRL, the classification head is refined with global class-balanced classification loss to address the data imbalance issue as well as learn the decision boundaries between new and previous classes. Extensive experiments on CIFAR100, Tiny-ImageNet200, and ImageNet100 demonstrate that our R-DFCIL significantly surpasses previous approaches and achieves a new state-of-the-art performance for DFCIL.



Paperid:957
Authors:Junbum Cha; Kyungjae Lee; Sungrae Park; Sanghyuk Chun
Title: Domain Generalization by Mutual-Information Regularization with Pre-trained Models
Abstract:
Domain generalization (DG) aims to learn a generalized model to an unseen target domain using only limited source domains. Previous attempts to DG fail to learn domain-invariant representations only from the source domains due to the significant domain shifts between training and test domains. Instead, we re-formulate the DG objective using mutual information with the oracle model, a model generalized to any possible domain. We derive a tractable variational lower bound via approximating the oracle model by a pre-trained model, called Mutual Information Regularization with Oracle (MIRO). Our extensive experiments show that MIRO significantly improves the out-of-distribution performance. Furthermore, our scaling experiments show that the larger the scale of the pre-trained model, the greater the performance improvement of MIRO. Code is available at https://github.com/kakaobrain/miro.



Paperid:958
Authors:Damien Teney; Maxime Peyrard; Ehsan Abbasnejad
Title: Predicting Is Not Understanding: Recognizing and Addressing Underspecification in Machine Learning
Abstract:
Machine learning models are typically designed for maximum accuracy on validation data. This predictive criterion rarely captures all desirable properties, in particular how a model matches a domain expert’s \emph{understanding} of the task. In this situation, known as underspecification, two models with similar validation accuracy may rely on different features (e.g. shape or texture in image recognition) and make very different predictions on out-of-distribution (OOD) data. Identifying underspecification is important as a warning against unexpected behaviour of deployed models, and as an indication of the need for additional task-specific knowledge. In this paper, we formalize the notion of underspecification and propose a method to identify and address the issue. We train multiple models with an independence constraint that forces them to discover distinct predictive features, most of which are missed by standard training. The number of models trainable under this constraint characterize the degree of underspecification of a task. Moreover, we show that an optimal set of these features can be combined to obtain a global predictor with superior OOD performance. We demonstrate the method on existing benchmarks and discuss important implications of underspecification. In particular, in-domain validation performance cannot serve for OOD model selection without additional assumptions.



Paperid:959
Authors:Yunhao Ge; Harkirat Behl; Jiashu Xu; Suriya Gunasekar; Neel Joshi; Yale Song; Xin Wang; Laurent Itti; Vibhav Vineet
Title: Neural-Sim: Learning to Generate Training Data with NeRF
Abstract:
Traditional approaches for training a computer vision models requires collecting and labelling vast amounts of imagery under a diverse set of scene configurations and properties. This process is incredibly time-consuming, and it is challenging to ensure that the captured data distribution maps well to the target domain of an application scenario. In recent years, synthetic data has emerged as a way to address both of these issues. However, current approaches either require human experts to manually tune each scene property or use automatic methods that provide little to no control; this requires rendering large amounts random data variations, which is slow and is often suboptimal for the target domain. We present the first fully differentiable synthetic data generation pipeline that uses Neural Radiance Fields (NeRFs) in a closed-loop with a target application’s loss function to generate data, on demand, with no human labor, to maximise accuracy for a target task. We illustrate the effectiveness of our method with synthetic and real-world object detection experiments. In addition, we evaluate on a new ""YCB-in-the-Wild"" dataset that provides a test scenario for object detection with varied pose in real-world environments.



Paperid:960
Authors:Hanwei Fan; Jiandong Mu; Wei Zhang
Title: Bayesian Optimization with Clustering and Rollback for CNN Auto Pruning
Abstract:
Pruning is an effective technique for convolutional neural networks (CNNs) model compression, but it is difficult to find the optimal pruning policy due to the large design space. To improve the usability of pruning, many auto pruning methods have been developed. Recently, Bayesian optimization (BO) has been considered to be a competitive algorithm for auto pruning due to its solid theoretical foundation and high sampling efficiency. However, BO suffers from the curse of dimensionality. The performance of BO deteriorates when pruning deep CNNs, since the dimension of the design spaces increase. We propose a novel clustering algorithm that reduces the dimension of the design space to speed up the searching process. Subsequently, a rollback algorithm is proposed to recover the high-dimensional design space so that higher pruning accuracy can be obtained. We validate our proposed method on ResNet, MobileNetV1, and MobileNetV2 models. Experiments show that the proposed method significantly improves the convergence rate of BO when pruning deep CNNs with no increase in running time. The source code is available at https://github.com/fanhanwei/BOCR.



Paperid:961
Authors:Markus Hofinger; Erich Kobler; Alexander Effland; Thomas Pock
Title: Learned Variational Video Color Propagation
Abstract:
In this paper, we propose a novel method for color propagation that is used to recolor gray-scale videos (e.g. historic movies). Our energy-based model combines deep learning with a variational formulation. At its core, the method optimizes over a set of plausible color proposals that are extracted from motion and semantic feature matches, together with a learned regularizer that resolves color ambiguities by enforcing spatial color smoothness. Our approach allows interpreting intermediate results and to incorporate extensions like using multiple reference frames even after training. We achieve state-of-the-art results on a number of standard benchmark datasets with multiple metrics and also provide convincing results on real historical videos - even though such types of video are not present during training. Moreover, a user evaluation shows that our method propagates initial colors more faithfully and temporally consistent.



Paperid:962
Authors:Fei Ye; Adrian G. Bors
Title: Continual Variational Autoencoder Learning via Online Cooperative Memorization
Abstract:
Due to their inference, data representation and reconstruction properties, Variational Autoencoders (VAE) have been successfully used in continual learning classification tasks. However, their ability to generate images with specifications corresponding to the classes and databases learned during Continual Learning (CL) is not well understood and catastrophic forgetting remains a significant challenge. In this paper, we firstly analyze the forgetting behaviour of VAEs by developing a new theoretical framework that formulates CL as a dynamic optimal transport problem. This framework proves approximate bounds to the data likelihood without requiring the task information and explains how the prior knowledge is lost during the training process. We then propose a novel memory buffering approach, namely the Online Cooperative Memorization (OCM) framework, which consists of a Short-Term Memory (STM) that continually stores recent samples to provide future information for the model, and a Long-Term Memory (LTM) aiming to preserve a wide diversity of samples. The proposed OCM transfers certain samples from STM to LTM according to the information diversity selection criterion without requiring any supervised signals. The OCM framework is then combined with a dynamic VAE expansion mixture network for further enhancing its performance.



Paperid:963
Authors:Yuanhao Xiong; Cho-Jui Hsieh
Title: Learning to Learn with Smooth Regularization
Abstract:
Recent decades have witnessed great advances of deep learning in tackling various problems such as classification and decision making. The rapid development gave rise to a novel framework, Learning-to-Learn (L2L), in which an automatic optimization algorithm (optimizer) modeled by neural networks is expected to learn rules for updating the target objective function (optimizee). Despite its advantages for specific problems, L2L still cannot replace classic methods due to its instability. Unlike hand-engineered algorithms, neural optimizers may suffer from the instability issue---under distinct but similar states, the same neural optimizer can produce quite different updates. Motivated by the stability property that should be satisfied by an ideal optimizer, we propose a regularization term that can enforce the smoothness and stability of the learned optimizers. Comprehensive experiments on the neural network training tasks demonstrate that the proposed regularization consistently improve the learned neural optimizers even when transferring to tasks with different architectures and datasets. Furthermore, we show that our smoothness-inducing regularizer can improve the performance of neural optimizers on few-shot learning tasks. Code can be found at https://github.com/xyh97/SmoothedOptimizer.



Paperid:964
Authors:Rakib Hyder; Ken Shao; Boyu Hou; Panos Markopoulos; Ashley Prater-Bennette; M. Salman Asif
Title: Incremental Task Learning with Incremental Rank Updates
Abstract:
Incremental Task learning (ITL) is a category of continual learning that seeks to train a single network for multiple tasks (one after another), where training data for each task is only available during the training of that task. Neural networks tend to forget older tasks when they are trained for the newer tasks; this property is often known as catastrophic forgetting. To address this issue, ITL methods use episodic memory, parameter regularization, masking and pruning, or extensible network structures. In this paper, we propose a new incremental task learning framework based on low-rank factorization. In particular, we represent the network weights for each layer as a linear combination of several rank-1 matrices. To update the network for a new task, we learn a rank-1 (or low-rank) matrix and add that to the weights of every layer. We also introduce an additional selector vector that assigns different weights to the low-rank matrices learned for the previous tasks. We show that our approach performs better than the current state-of-the-art methods in terms of accuracy and forgetting. Our method also offers better memory efficiency compared to episodic memory- and mask-based approaches. Our code will be available at https://github.com/CSIPlab/task-increment-rank-update.git.



Paperid:965
Authors:Yue Song; Nicu Sebe; Wei Wang
Title: Batch-Efficient EigenDecomposition for Small and Medium Matrices
Abstract:
EigenDecomposition (ED) is at the heart of many computer vision algorithms and applications. One crucial bottleneck limiting its usage is the expensive computation cost, particularly for a mini-batch of matrices in the deep neural networks. In this paper, we propose a QR-based ED method dedicated to the application scenarios of computer vision. Our proposed method performs the ED entirely by batched matrix/vector multiplication, which processes all the matrices simultaneously and thus fully utilizes the power of GPUs. Our technique is based on the explicit QR iterations by Givens rotation with double Wilkinson shifts. With several acceleration techniques, the time complexity of QR iterations is reduced from $O{(}n^5{)}$ to $O{(}n^3{)}$. The numerical test shows that for small and medium batched matrices (\emph{e.g.,} $dim{<}32$) our method can be much faster than the Pytorch SVD function. Experimental results on visual recognition and image generation demonstrate that our methods also achieve competitive performances.



Paperid:966
Authors:Chengshuai Yang; Shiyu Zhang; Xin Yuan
Title: Ensemble Learning Priors Driven Deep Unfolding for Scalable Video Snapshot Compressive Imaging
Abstract:
Snapshot compressive imaging (SCI) can record the 3D datacube by a 2D measurement and from this 2D measurement to reconstruct the desired 3D information by algorithms. The reconstruction algorithm thus plays a vital role in SCI. Recently, deep learning (DL) has demonstrated outstanding performance in reconstruction, leading to better results than conventional optimization based methods. Therefore, it is desired to improve DL reconstruction performance for SCI. Existing DL algorithms are limited by two bottlenecks: 1) a high accuracy network is usually large and requires a long running time; 2) DL algorithms are limited by scalability, i.e., a well trained network in general can not be applied to new systems. Towards this end, this paper proposes to use ensemble learning priors in DL to keep high reconstruction speed and accuracy in a single network. Furthermore, we develop the scalable learning approach during training to empower DL to handle data of different sizes without additional training. Extensive results on both simulation and real datasets demonstrate the superiority of our proposed algorithm. The code and model will be released.



Paperid:967
Authors:Dongsheng An; Na Lei; Xianfeng Gu
Title: Approximate Discrete Optimal Transport Plan with Auxiliary Measure Method
Abstract:
Optimal transport (OT) between two measures plays an essential role in many fields, ranging from economy, biology to machine learning and artificial intelligence. Conventional discrete OT problem can be solved using linear programming (LP). Unfortunately, due to the large scale and the intrinsic non-linearity, achieving discrete OT plan with adequate accuracy and efficiency is challenging. Generally speaking, the OT plan is highly sparse. This work proposes an auxiliary measure method to use the semi-discrete OT maps to estimate the sparsity of the discrete OT plan with squared Euclidean cost. Although obtaining the accurate semi-discrete OT maps is difficult, we can find the sparsity information through computing the approximate semi-discrete OT maps by convex optimization. The sparsity information can be further incorporated into the downstream LP optimization to greatly reduce the computational complexity and improve the accuracy. We also give a theoretic error bound between the estimated transport plan and the OT plan in terms of Wasserstein distance. Experiments on both synthetic data and color transfer tasks demonstrate the accuracy and efficiency of the proposed method.



Paperid:968
Authors:Stefan Haller; Lorenz Feineis; Lisa Hutschenreiter; Florian Bernard; Carsten Rother; Dagmar Kainmüller; Paul Swoboda; Bogdan Savchynskyy
Title: A Comparative Study of Graph Matching Algorithms in Computer Vision
Abstract:
The graph matching optimization problem is an essential component for many tasks in computer vision, such as bringing two deformable objects in correspondence. Naturally, a wide range of applicable algorithms have been proposed in the last decades. Since a common standard benchmark has not been developed, their performance claims are often hard to verify as evaluation on differing problem instances and criteria make the results incomparable. To address these shortcomings, we present a comparative study of graph matching algorithms. We create a uniform benchmark where we collect and categorize a large set of existing and publicly available computer vision graph matching problems in a common format. At the same time we collect and categorize the most popular open-source implementations of graph matching algorithms. Their performance is evaluated in a way that is in line with the best practices for comparing optimization algorithms. The study is designed to be reproducible and extensible to serve as a valuable resource in the future. Our study provides three notable insights: (i) popular problem instances are exactly solvable in substantially less than 1 second, and, therefore, are insufficient for future empirical evaluations; (ii) the most popular baseline methods are highly inferior to the best available methods; (iii) despite the NP-hardness of the problem, instances coming from vision applications are often solvable in a few seconds even for graphs with more than 500 vertices.



Paperid:969
Authors:Debora Caldarola; Barbara Caputo; Marco Ciccone
Title: Improving Generalization in Federated Learning by Seeking Flat Minima
Abstract:
Models trained in federated settings often suffer from degraded performances and fail at generalizing, especially when facing heterogeneous scenarios. In this work, we investigate such behavior through the lens of geometry of the loss and Hessian eigenspectrum, linking the model’s lack of generalization capacity to the sharpness of the solution. Motivated by prior studies connecting the sharpness of the loss surface and the generalization gap, we show that i) training clients locally with Sharpness-Aware Minimization (SAM) or its adaptive version (ASAM) and ii) averaging stochastic weights (SWA) on the server-side can substantially improve generalization in Federated Learning and help bridging the gap with centralized models. By seeking parameters in neighborhoods having uniform low loss, the model converges towards flatter minima and its generalization significantly improves in both homogeneous and heterogeneous scenarios. Empirical results demonstrate the effectiveness of those optimizers across a variety of benchmark vision datasets (e.g. CIFAR, Landmarks-User-160k, IDDA) and tasks (large scale classification, semantic segmentation, domain generalization).



Paperid:970
Authors:Liangzu Peng; Mahyar Fazlyab; René Vidal
Title: Semidefinite Relaxations of Truncated Least-Squares in Robust Rotation Search: Tight or Not
Abstract:
The rotation search problem aims to find a 3D rotation that best aligns a given number of point pairs. To induce robustness against outliers for rotation search, prior work considers truncated least-squares (TLS), which is a non-convex optimization problem, and its semidefinite relaxation (SDR) as a tractable alternative. Whether or not this SDR is theoretically tight in the presence of noise, outliers, or both has remained largely unexplored. We derive conditions that characterize the tightness of this SDR, showing that the tightness depends on the noise level, the truncation parameters of TLS, and the outlier distribution (random or clustered). In particular, we give a short proof for the tightness in the noiseless and outlier-free case, as opposed to the lengthy analysis of prior work.



Paperid:971
Authors:Matteo Boschini; Lorenzo Bonicelli; Angelo Porrello; Giovanni Bellitto; Matteo Pennisi; Simone Palazzo; Concetto Spampinato; Simone Calderara
Title: Transfer without Forgetting
Abstract:
This work investigates the entanglement between Continual Learning (CL) and Transfer Learning (TL). In particular, we shed light on the widespread application of network pretraining, highlighting that it is itself subject to catastrophic forgetting. Unfortunately, this issue leads to the under-exploitation of knowledge transfer during later tasks. On this ground, we propose Transfer without Forgetting (TwF), a hybrid approach building upon a fixed pretrained sibling network, which continuously propagates the knowledge inherent in the source domain through a layer-wise loss term. Our experiments indicate that TwF steadily outperforms other CL methods across a variety of settings, averaging a 4.81% gain in Class-Incremental accuracy over a variety of datasets and different buffer sizes.



Paperid:972
Authors:Farshid Varno; Marzie Saghayi; Laya Rafiee Sevyeri; Sharut Gupta; Stan Matwin; Mohammad Havaei
Title: AdaBest: Minimizing Client Drift in Federated Learning via Adaptive Bias Estimation
Abstract:
In Federated Learning (FL), a number of clients or devices collaborate to train a model without sharing their data. Models are optimized locally at each client and further communicated to a central hub for aggregation. While FL is an appealing decentralized training paradigm, heterogeneity among data from different clients can cause the local optimization to drift away from the global objective. In order to estimate and therefore remove this drift, variance reduction techniques have been incorporated into FL optimization recently. However, these approaches inaccurately estimate the clients’ drift and ultimately fail to remove it properly. In this work, we propose an adaptive algorithm that accurately estimates drift across clients. In comparison to previous works, our approach necessitates less storage and communication bandwidth, as well as lower compute costs. Additionally, our proposed methodology induces stability by constraining the norm of estimates for client drift, making it more practical for large scale FL. Experimental findings demonstrate that the proposed algorithm converges significantly faster and achieves higher accuracy than the baselines across various FL benchmarks.



Paperid:973
Authors:Xiao Gu; Yao Guo; Zeju Li; Jianing Qiu; Qi Dou; Yuxuan Liu; Benny Lo; Guang-Zhong Yang
Title: Tackling Long-Tailed Category Distribution under Domain Shifts
Abstract:
Machine learning models fail to perform well on real-world applications when 1) the category distribution P(Y) of the training dataset suffers from long-tailed distribution and 2) the test data is drawn from different conditional distributions P(X|Y). Existing approaches cannot handle the scenario where both issues exist, which however is common for real-world applications. In this study, we took a step forward and looked into the problem of long-tailed classification under domain shifts. We designed three novel core functional blocks including Distribution Calibrated Classification Loss, Visual-Semantic Mapping and Semantic-Similarity Guided Augmentation. Furthermore, we adopted a meta-learning framework which integrates these three blocks to improve domain generalization on unseen target domains. Two new datasets were proposed for this problem, named AWA2-LTS and ImageNet-LTS. We evaluated our method on the two datasets and extensive experimental results demonstrate that our proposed method can achieve superior performance over state-of-the-art long-tailed/domain generalization approaches and the combinations. Source codes and datasets can be found at our project page https://xiaogu.site/LTDS.



Paperid:974
Authors:Li Gao; Dong Nie; Bo Li; Xiaofeng Ren
Title: Doubly-Fused ViT: Fuse Information from Vision Transformer Doubly with Local Representation
Abstract:
Vision Transformer (ViT) has recently emerged as a new paradigm for computer vision tasks, but is not as efficient as convolutional neural networks (CNN). In this paper, we propose an efficient ViT architecture, named Doubly-Fused ViT (DFvT), where we feed low-resolution feature maps to self-attention (SA) to achieve larger context with efficiency (by moving downsampling prior to SA), and enhance it with fine-detailed spatial information. SA is a powerful mechanism that extracts rich context information, thus could and should operate at a low spatial resolution. To make up for the loss of details, convolutions are fused into the main ViT pipeline, without incurring high computational costs. In particular, a Context Module (CM), consisting of fused downsampling operator and subsequent SA, is introduced to effectively capture global features with high efficiency. A Spatial Module (SM) is proposed to preserve fine-grained spatial information. To fuse the heterogeneous features, we specially design a Dual AtteNtion Enhancement (DANE) module to selectively fuse low-level and high-level features. Experiments demonstrate that DFvT achieves state-of-the-art accuracy with much higher efficiency across a spectrum of different model sizes. Ablation study validates the effectiveness of our designed components.



Paperid:975
Authors:Jiawang Bai; Li Yuan; Shu-Tao Xia; Shuicheng Yan; Zhifeng Li; Wei Liu
Title: Improving Vision Transformers by Revisiting High-Frequency Components
Abstract:
The transformer models have shown promising effectiveness in dealing with various vision tasks. However, compared with training Convolutional Neural Network (CNN) models, training Vision Transformer (ViT) models is more difficult and relies on the large-scale training set. To explain this observation we make a hypothesis that \textit{ViT models are less effective in capturing the high-frequency components of images than CNN models}, and verify it by a frequency analysis. Inspired by this finding, we first investigate the effects of existing techniques for improving ViT models from a new frequency perspective, and find that the success of some techniques (e.g., RandAugment) can be attributed to the better usage of the high-frequency components. Then, to compensate for this insufficient ability of ViT models, we propose HAT, which directly augments high-frequency components of images via adversarial training. We show that HAT can consistently boost the performance of various ViT models (e.g., +1.2% for ViT-B, +0.5% for Swin-B), and especially enhance the advanced model VOLO-D5 to 87.3% that only uses ImageNet-1K data, and the superiority can also be maintained on out-of-distribution data and transferred to downstream tasks. The code is available at: https://github.com/jiawangbai/HAT.



Paperid:976
Authors:Sheng Xu; Yanjing Li; Tiancheng Wang; Teli Ma; Baochang Zhang; Peng Gao; Yu Qiao; Jinhu Lü; Guodong Guo
Title: Recurrent Bilinear Optimization for Binary Neural Networks
Abstract:
Binary Neural Networks (BNNs) show great promise for real-world embedded devices. As one of the critical steps to achieve a powerful BNN, the scale factor calculation plays an essential role in reducing the performance gap to their real-valued counterparts. However, existing BNNs neglect the intrinsic bilinear relationship of real-valued weights and scale factors, resulting in a sub-optimal model caused by an insufficient training process. To address this issue, Recurrent Bilinear Optimization is proposed to improve the learning process of BNNs (RBONNs) by associating the intrinsic bilinear variables in the back propagation process. Our work is the first attempt to optimize BNNs from the bilinear perspective. Specifically, we employ a recurrent optimization and Density-ReLU to sequentially backtrack the sparse real-valued weight filters, which will be sufficiently trained and reach their performance limits based on a controllable learning process. We obtain robust RBONNs, which show impressive performance over state-of-the-art BNNs on various models and datasets. Particularly, on the task of object detection, RBONNs have great generalization performance. Our code is open-sourced on https://github.com/SteveTsui/RBONN.



Paperid:977
Authors:Youngeun Kim; Yuhang Li; Hyoungseob Park; Yeshwanth Venkatesha; Priyadarshini Panda
Title: Neural Architecture Search for Spiking Neural Networks
Abstract:
Spiking Neural Networks (SNNs) have gained huge attention as a potential energy-efficient alternative to conventional Artificial Neural Networks (ANNs) due to their inherent high-sparsity activation. However, most prior SNN methods use ANN-like architectures (e.g., VGG-Net or ResNet), which could provide sub-optimal performance for temporal sequence processing of binary information in SNNs. To address this, in this paper, we introduce a novel Neural Architecture Search (NAS) approach for finding better SNN architectures. Inspired by recent NAS approaches that find the optimal architecture from activation patterns at initialization, we select the architecture that can represent diverse spike activation patterns across different data samples without training. Moreover, to further leverage the temporal information among the spikes, we search for feed forward connections as well as backward connections (i.e., temporal feedback connections) between layers. Interestingly, SNASNet found by our search algorithm achieves higher performance with backward connections, demonstrating the importance of designing SNN architecture for suitably using temporal information. We conduct extensive experiments on three image recognition benchmarks where we show that SNASNet achieves state-of-the-art performance with significantly lower timesteps (5 timesteps). Code is available at Github.



Paperid:978
Authors:Yang Liu; Lei Zhou; Pengcheng Zhang; Xiao Bai; Lin Gu; Xiaohan Yu; Jun Zhou; Edwin R. Hancock
Title: Where to Focus: Investigating Hierarchical Attention Relationship for Fine-Grained Visual Classification
Abstract:
Object categories are often grouped into a multi-granularity taxonomic hierarchy. Classifying objects at coarser-grained hierarchy requires global and common characteristics, while finer-grained hierarchy classification relies on local and discriminative features. Therefore, humans should also subconsciously focus on different object regions when classifying different hierarchies. This granularity-wise attention is confirmed by our collected human real-time gaze data on different hierarchy classifications. To leverage this mechanism, we propose a Cross-Hierarchical Region Feature (CHRF) learning framework. Specifically, we first design a region feature mining module that imitates humans to learn different granularity-wise attention regions with multi-grained classification tasks. To explore how human attention shifts from one hierarchy to another, we further present a cross-hierarchical orthogonal fusion module to enhance the region feature representation by blending the original feature and an orthogonal component extracted from adjacent hierarchies. Experiments on five hierarchical fine-grained datasets demonstrate the effectiveness of CHRF compared with the state-of-the-art methods. Ablation study and visualization results also consistently verify the advantages of our human attention-oriented modules. The code and dataset are available at https://github.com/visiondom/CHRF.



Paperid:979
Authors:Mingyu Ding; Bin Xiao; Noel Codella; Ping Luo; Jingdong Wang; Lu Yuan
Title: DaViT: Dual Attention Vision Transformers
Abstract:
In this work, we introduce Dual Attention Vision Transformers (DaViT), a simple yet effective vision transformer architecture that is able to capture global context while maintaining computational efficiency. We propose approaching the problem from an orthogonal angle: exploiting self-attention mechanisms with both ""spatial tokens"" and ""channel tokens"". With spatial tokens, the spatial dimension defines the token scope, and the channel dimension defines the token feature dimension. With channel tokens, we have the inverse: the channel dimension defines the token scope, and the spatial dimension defines the token feature dimension. We further group tokens along the sequence direction for both spatial and channel tokens to maintain the linear complexity of the entire model. We show that these two self-attentions complement each other: (i) since each channel token contains an abstract representation of the entire image, the channel attention naturally captures global interactions and representations by taking all spatial positions into account when computing attention scores between channels; (ii) the spatial attention refines the local representations by performing fine-grained interactions across spatial locations, which in turn helps the global information modeling in channel attention. Extensive experiments show our DaViT achieves state-of-the-art performance on four different tasks with efficient computations. Without extra data, DaViT-Tiny, DaViT-Small, and DaViT-Base achieve 82.8%, 84.2%, and 84.6% top-1 accuracy on ImageNet-1K with 28.3M, 49.7M, and 87.9M parameters, respectively. When we further scale up DaViT with 1.5B weakly supervised image and text pairs, DaViT-Gaint reaches 90.4% top-1 accuracy on ImageNet-1K.



Paperid:980
Authors:Jiangming Wang; Zhizhong Zhang; Mingang Chen; Yi Zhang; Cong Wang; Bin Sheng; Yanyun Qu; Yuan Xie
Title: Optimal Transport for Label-Efficient Visible-Infrared Person Re-identification
Abstract:
Visible-infrared person re-identification (VI-ReID) has been a key enabler for night intelligent monitoring system. However, the extensive laboring efforts significantly limit its applications. In this paper, we raise a new label-efficient training pipeline for VI-ReID. Our observation is: RGB ReID datasets have rich annotation information and annotating infrared images is expensive due to the lack of color information. In our approach, it includes two key steps: 1) We utilize the standard unsupervised domain adaptation technique to generate the pseudo labels for visible subset with the help of well-annotated RGB datasets; 2) We propose an optimal-transport strategy trying to assign pseudo labels from visible to infrared modality. In our framework, each infrared sample owns a label assignment choice, and each pseudo label requires unallocated images. By introducing uniform sample-wise and label-wise prior, we achieve a desirable assignment plan that allows us to find matched visible and infrared samples, and thereby facilitates cross-modality learning. Besides, a prediction alignment loss is designed to eliminate the negative effects brought by the incorrect pseudo labels. Extensive experimental results on benchmarks demonstrate the effectiveness of our approach. Code will be released at https://github.com/wjm-wjm/OTLA-ReID.



Paperid:981
Authors:Kehan Li; Runyi Yu; Zhennan Wang; Li Yuan; Guoli Song; Jie Chen
Title: Locality Guidance for Improving Vision Transformers on Tiny Datasets
Abstract:
While the Vision Transformer (VT) architecture is becoming trendy in computer vision, pure VT models perform poorly on tiny datasets. To address this issue, this paper proposes the locality guidance for improving the performance of VTs on tiny datasets. We first analyze that the local information, which is of great importance for understanding images, is hard to be learned with limited data due to the high flexibility and intrinsic globality of the self-attention mechanism in VTs. To facilitate local information, we realize the locality guidance for VTs by imitating the features of an already trained convolutional neural network (CNN), inspired by the built-in local-to-global hierarchy of CNN. Under our dual-task learning paradigm, the locality guidance provided by a lightweight CNN trained on low-resolution images is adequate to accelerate the convergence and improve the performance of VTs to a large extent. Therefore, our locality guidance approach is very simple and efficient, and can serve as a basic performance enhancement method for VTs on tiny datasets. Extensive experiments demonstrate that our method can significantly improve VTs when training from scratch on tiny datasets and is compatible with different kinds of VTs and datasets. For example, our proposed method can boost the performance of various VTs on tiny datasets (e.g., 13.07% for DeiT, 8.98% for T2T and 7.85% for PVT), and enhance even stronger baseline PVTv2 by 1.86% to 79.30%, showing the potential of VTs on tiny datasets. The code is available at https://github.com/lkhl/tiny-transformers.



Paperid:982
Authors:Jichang Li; Guanbin Li; Feng Liu; Yizhou Yu
Title: Neighborhood Collective Estimation for Noisy Label Identification and Correction
Abstract:
Learning with noisy labels (LNL) aims at designing strategies to improve model performance and generalization by mitigating the effects of model overfitting to noisy labels. The key success of LNL lies in identifying as many clean samples as possible from massive noisy data, while rectifying the wrongly assigned noisy labels. Recent advances employ the predicted label distributions of individual samples to perform noise verification and noisy label correction, easily giving rise to confirmation bias. To mitigate this issue, we propose Neighborhood Collective Estimation, in which the predictive reliability of a candidate sample is re-estimated by contrasting it against its feature-space nearest neighbors. Specifically, our method is divided into two steps: 1) Neighborhood Collective Noise Verification to separate all training samples into a clean or noisy subset, 2) Neighborhood Collective Label Correction to relabel noisy samples, and then auxiliary techniques are used to assist further model optimization. Extensive experiments on four commonly used benchmark datasets, i.e., CIFAR-10, CIFAR-100, Clothing-1M and Webvision-1.0, demonstrate that our proposed method considerably outperforms state-of-the-art methods.



Paperid:983
Authors:Huan Liu; Li Gu; Zhixiang Chi; Yang Wang; Yuanhao Yu; Jun Chen; Jin Tang
Title: Few-Shot Class-Incremental Learning via Entropy-Regularized Data-Free Replay
Abstract:
Few-shot class-incremental learning (FSCIL) has been proposed aiming to enable a deep learning system to incrementally learn new classes with limited data. Recently, a pioneer claims that the commonly used replay-based method in class-incremental learning (CIL) is ineffective and thus not preferred for FSCIL. This has, if truth, a significant influence on the fields of FSCIL. In this paper, we show through empirical results that adopting the data replay is surprisingly favorable. However, storing and replaying old data can lead to a privacy concern. To address this issue, we alternatively propose using data-free replay that can synthesize data by a generator without accessing real data. In observing the the effectiveness of uncertain data for knowledge distillation, we impose entropy regularization in the generator training to encourage more uncertain examples. Moreover, we propose to relabel the generated data with one-hot-like labels. This modification allows the network to learn by solely minimizing the cross-entropy loss, which mitigates the problem of balancing different objectives in the conventional knowledge distillation approach. Finally, we show extensive experimental results and analysis on CIFAR-100, miniImageNet and CUB-200 to demonstrate the effectiveness of our proposed one.



Paperid:984
Authors:Runqi Wang; Yuxiang Bao; Baochang Zhang; Jianzhuang Liu; Wentao Zhu; Guodong Guo
Title: Anti-Retroactive Interference for Lifelong Learning
Abstract:
Humans can continuously learn new knowledge. However, machine learning models suffer from drastic dropping in performance on previous tasks after learning new tasks. Cognitive science points out that the competition of similar knowledge is an important cause of forgetting. In this paper, we design a paradigm for lifelong learning based on meta-learning and associative mechanism of the brain. It tackles the problem from two aspects: extracting knowledge and memorizing knowledge. First, we disrupt the sample’s background distribution through a background attack, which strengthens the model to extract the key features of each task. Second, according to the similarity between incremental knowledge and base knowledge, we design an adaptive fusion of incremental knowledge, which helps the model allocate capacity to the knowledge of different difficulties. It is theoretically analyzed that the proposed learning paradigm can make the models of different tasks converge to the same optimum. The proposed method is validated on the MNIST, CIFAR100, CUB200 and ImageNet100 datasets. The code is available at https://github.com/bhrqw/ARI.



Paperid:985
Authors:Hualiang Wang; Siming Fu; Xiaoxuan He; Hangxiang Fang; Zuozhu Liu; Haoji Hu
Title: Towards Calibrated Hyper-Sphere Representation via Distribution Overlap Coefficient for Long-Tailed Learning
Abstract:
Long-tailed learning aims to tackle the crucial challenge that head classes dominate the training procedure under severe class imbalance in real-world scenarios. However, little attention has been given to how to quantify the dominance severity of head classes in the representation space. Motivated by this, we generalize the cosine-based classifiers to a von Mises-Fisher (vMF) mixture model, denoted as vMF classifier, which enables to quantitatively measure representation quality upon the hyper-sphere space via calculating distribution overlap coefficient. To our knowledge, this is the first work to measure the representation quality of classifiers and features from the perspective of the distribution overlap coefficient. On top of it, we formulate the inter-class discrepancy and class feature consistency loss terms to alleviate the interference among the classifier weights and align features with classifier weights. Furthermore, a novel post-training calibration algorithm is devised to zero-costly boost the performance via inter-class overlap coefficients. Our models outperform previous work with a large margin and achieve state-of-the-art performance on long-tailed image classification, semantic segmentation, and instance segmentation tasks (e.g., we achieve 55.0% overall accuracy with ResNetXt-50 in ImageNet-LT). Our code is available at https://github.com/VipaiLab/vMF_OP.



Paperid:986
Authors:Wenzhao Zheng; Yuanhui Huang; Borui Zhang; Jie Zhou; Jiwen Lu
Title: Dynamic Metric Learning with Cross-Level Concept Distillation
Abstract:
A good similarity metric should be consistent with the human perception of similarities: a sparrow is more similar to an owl if compared to a dog but is more similar to a dog if compared to a car. It depends on the semantic levels to determine if two images are from the same class. As most existing metric learning methods push away interclass samples and pull closer intraclass samples, it seems contradictory if the labels cross semantic levels. The core problem is that a negative pair on a finer semantic level can be a positive pair on a coarser semantic level, so pushing away this pair damages the class structure on the coarser semantic level. We identify the negative repulsion as the key obstacle in existing methods since a positive pair is always positive for coarser semantic levels but not for negative pairs. Our solution, cross-level concept distillation (CLCD), is simple in concept: we only pull closer positive pairs. To facilitate the cross-level semantic structure of the image representations, we propose a hierarchical concept refiner to construct multiple levels of concept embeddings of an image and then pull closer the distance of the corresponding concepts. Extensive experiments demonstrate that the proposed CLCD method outperforms all other competing methods on the hierarchically labeled datasets.



Paperid:987
Authors:Linhui Sun; Yifan Zhang; Ke Cheng; Jian Cheng; Hanqing Lu
Title: MENet: A Memory-Based Network with Dual-Branch for Efficient Event Stream Processing
Abstract:
Event cameras are bio-inspired sensors that asynchronously capture per-pixel brightness change and trigger a stream of events instead of frame-based images. Each event stream is generally split into multiple sliding windows for subsequent processing. However, most existing event-based methods ignore the motion continuity between adjacent spatiotemporal windows, which will result in the loss of dynamic information and additional computational costs. To efficiently extract strong features for event streams containing dynamic information, this paper proposes a novel memory-based network with dual-branch, namely MENet. It contains a base branch with a full-sized event point-wise processing structure to extract the base features and an incremental branch equipped with a light-weighted network to capture the temporal dynamics between two adjacent spatiotemporal windows. For enhancing the features, especially in the incremental branch, a point-wise memory bank is designed, which sketches the representative information of event feature space. Compared with the base branch, the incremental branch reduces the computational complexity up to 5 times and improves the speed by 19 times. Experiments show that MENet significantly reduces the computational complexity compared with previous methods while achieving state-of-the-art performance on gesture recognition and object recognition.



Paperid:988
Authors:Sen Pei; Xin Zhang; Bin Fan; Gaofeng Meng
Title: Out-of-Distribution Detection with Boundary Aware Learning
Abstract:
There is an increasing need to determine whether inputs are out-of-distribution (OOD) for safely deploying machine learning models in the open world scenario. Typical neural classifiers are based on the closed-world assumption, where the training data and the test data are drawn i.i.d. from the same distribution, and as a result, give over-confident predictions even faced with OOD inputs. For tackling this problem, previous studies either use real outliers for training or generate synthetic OOD data under strong assumptions, which are either costly or intractable to generalize. In this paper, we propose boundary aware learning (BAL), a novel framework that can learn the distribution of OOD features adaptively. The key idea of BAL is to generate OOD features from trivial to hard progressively with a generator, meanwhile, a discriminator is trained for distinguishing these synthetic OOD features and in-distribution (ID) features. Benefiting from the adversarial training scheme, the discriminator can well separate ID and OOD features, allowing more robust OOD detection. The proposed BAL achieves state-of-the-art performance on classification benchmarks, reducing up to 13.9% FPR95 compared with previous methods.



Paperid:989
Authors:Ashima Garg; Depanshu Sani; Saket Anand
Title: Learning Hierarchy Aware Features for Reducing Mistake Severity
Abstract:
Label hierarchies are often available apriori as part of biological taxonomy or language datasets WordNet. Several works exploit these to learn hierarchy aware features in order to improve the classifier to make semantically meaningful mistakes while maintaining or reducing the overall error. In this paper, we propose a novel approach for learning Hierarchy Aware Features (HAF) that leverages classifiers at each level of the hierarchy that are constrained to generate predictions consistent with the label hierarchy. The classifiers are trained by minimizing a Jensen-Shannon Divergence with target soft labels obtained from the fine-grained classifiers. Additionally, we employ a simple geometric loss that constrains the feature space geometry to capture the semantic structure of the label space. HAF is a training time approach that improves the mistakes while maintaining top-1 error, thereby, addressing the problem of cross-entropy loss that treats all mistakes as equal. We evaluate HAF on three hierarchical datasets and achieve state-of-the-art results on the iNaturalist-19 and CIFAR-100 datasets.



Paperid:990
Authors:Kuniaki Saito; Ping Hu; Trevor Darrell; Kate Saenko
Title: Learning to Detect Every Thing in an Open World
Abstract:
Many open-world applications require the detection of novel objects, yet state-of-the-art object detection and instance segmentation networks do not excel at this task. The key issue lies in their assumption that regions without any annotations should be suppressed as negatives, which teaches the model to treat any unannotated (hidden) objects as background. To address this issue, we propose a simple yet surprisingly powerful data augmentation and training scheme we call Learning to Detect Every Thing (LDET). To avoid suppressing hidden objects, we develop a new data augmentation method, BackErase, which pastes annotated objects on a background image sampled from a small region of the original image. Since training solely on such synthetically-augmented images suffers from domain shift, we propose a multi-domain training strategy that allows the model to generalize to real images. LDET leads to significant improvements on many datasets in the open-world instance segmentation task, outperforming baselines on cross-category generalization on COCO, as well as cross-dataset evaluation on UVO, Objects365, and Cityscapes.



Paperid:991
Authors:Pichao Wang; Xue Wang; Fan Wang; Ming Lin; Shuning Chang; Hao Li; Rong Jin
Title: KVT: $k$-NN Attention for Boosting Vision Transformers
Abstract:
Convolutional Neural Networks (CNNs) have dominated computer vision for years, due to its ability in capturing locality and translation invariance. Recently, many vision transformer architectures have been proposed and they show promising performance. A key component in vision transformers is the fully-connected self-attention which is more powerful than CNNs in modelling long range dependencies. However, since the current dense self-attention uses all image patches (tokens) to compute attention matrix, it may neglect locality of images patches and involve noisy tokens (e.g., clutter background and occlusion), leading to a slow training process and potential degradation of performance. To address these problems, we propose the $k$-NN attention for boosting vision transformers. Specifically, instead of involving all the tokens for attention matrix calculation, we only select the top-$k$ similar tokens from the keys for each query to compute the attention map. The proposed $k$-NN attention naturally inherits the local bias of CNNs without introducing convolutional operations, as nearby tokens tend to be more similar than others. In addition, the $k$-NN attention allows for the exploration of long range correlation and at the same time filters out irrelevant tokens by choosing the most similar tokens from the entire image. Despite its simplicity, we verify, both theoretically and empirically, that $k$-NN attention is powerful in speeding up training and distilling noise from input tokens. Extensive experiments are conducted by using 11 different vision transformer architectures to verify that the proposed $k$-NN attention can work with any existing transformer architectures to improve its prediction performance. The codes are available at \url{https://github.com/damo-cv/KVT}.



Paperid:992
Authors:Chaoqin Huang; Haoyan Guan; Aofan Jiang; Ya Zhang; Michael Spratling; Yan-Feng Wang
Title: Registration Based Few-Shot Anomaly Detection
Abstract:
This paper considers few-shot anomaly detection (FSAD), a practical yet under-studied setting for anomaly detection (AD), where only a limited number of normal images are provided for each category at training. So far, existing FSAD studies follow the one-model-per-category learning paradigm used for standard AD, and the inter-category commonality has not been explored. Inspired by how humans detect anomalies, i.e., comparing an image in question to normal images, we here leverage registration, an image alignment task that is inherently generalizable across categories, as the proxy task, to train a category-agnostic anomaly detection model. During testing, the anomalies are identified by comparing the registered features of the test image and its corresponding support (normal) images. As far as we know, this is the first FSAD method that trains a single generalizable model and requires no re-training or parameter fine-tuning for new categories. Experimental results have shown that the proposed method outperforms the state-of-the-art FSAD methods by 3%-8% in AUC on the MVTec and MPDD benchmarks. Source code is available at: https://github.com/MediaBrain-SJTU/RegAD"



Paperid:993
Authors:Yong Guo; David Stutz; Bernt Schiele
Title: Improving Robustness by Enhancing Weak Subnets
Abstract:
Despite their success, deep networks have been shown to be highly susceptible to perturbations, often causing significant drops in accuracy. In this paper, we investigate model robustness on perturbed inputs by studying the performance of internal sub-networks (subnets). Interestingly, we observe that most subnets show particularly poor robustness against perturbations. More importantly, these weak subnets are correlated with the overall lack of robustness. Tackling this phenomenon, we propose a new training procedure that identifies and enhances weak subnets (EWS) to improve robustness. Specifically, we develop a search algorithm to find particularly weak subnets and explicitly strengthen them via knowledge distillation from the full network. We show that EWS greatly improves both robustness against corrupted images as well as accuracy on clean data. Being complementary to popular data augmentation methods, EWS consistently improves robustness when combined with these approaches. To highlight the flexibility of our approach, we combine EWS also with popular adversarial training methods resulting in improved adversarial robustness.



Paperid:994
Authors:Tian Zhang; Kongming Liang; Ruoyi Du; Xian Sun; Zhanyu Ma; Jun Guo
Title: Learning Invariant Visual Representations for Compositional Zero-Shot Learning
Abstract:
Compositional Zero-Shot Learning (CZSL) aims to recognize novel compositions using knowledge learned from seen attribute-object compositions in the training set. Previous works mainly project an image and a composition into a common embedding space to measure their compatibility score. However, both attributes and objects share the visual representations learned above, leading the model to exploit spurious correlations and bias towards seen pairs. Instead, we reconsider CZSL as an out-of-distribution generalization problem. If an object is treated as a domain, we can learn object-invariant features to recognize the attributes attached to any object reliably. Similarly, attribute-invariant features can also be learned when recognizing the objects with attributes as domains. Specifically, we propose an invariant feature learning framework to align different domains at the representation and gradient levels to capture the intrinsic characteristics associated with the tasks. Experiments on two CZSL benchmarks demonstrate that the proposed method significantly outperforms the previous state-of-the-art.



Paperid:995
Authors:Yue Song; Nicu Sebe; Wei Wang
Title: Improving Covariance Conditioning of the SVD Meta-Layer by Orthogonality
Abstract:
Inserting an SVD meta-layer into neural networks is prone to make the covariance ill-conditioned, which could harm the model in the training stability and generalization abilities. In this paper, we systematically study how to improve the covariance conditioning by enforcing orthogonality to the Pre-SVD layer. Existing orthogonal treatments on the weights are first investigated. However, these techniques can improve the conditioning but would hurt the performance. To avoid such a side effect, we propose the Nearest Orthogonal Gradient (NOG) and Optimal Learning Rate (OLR). The effectiveness of our methods is validated in two applications: decorrelated Batch Normalization (BN) and Global Covariance Pooling (GCP). Extensive experiments on visual recognition demonstrate that our methods can simultaneously improve the covariance conditioning and generalization. Moreover, the combinations with orthogonal weight can further boost the performances.



Paperid:996
Authors:Yijun Yang; Ruiyuan Gao; Qiang Xu
Title: Out-of-Distribution Detection with Semantic Mismatch under Masking
Abstract:
This paper proposes a novel out-of-distribution (OOD) detection framework named MOODCat for image classifiers. MOODCat masks a random portion of the input image and uses a generative model to synthesize the masked image to a new image conditioned on the classification result. It then calculates the semantic difference between the original image and the synthesized one for OOD detection. Compared to existing solutions, MOODCat naturally learns the semantic information of the in-distribution data with the proposed mask and conditional synthesis strategy, which is critical to identify OODs. Experimental results demonstrate that MOODCat outperforms state-of-the-art OOD detection solutions by a large margin. Our code is available at https://github.com/cure-lab/MOODCat.



Paperid:997
Authors:Zechun Liu; Zhiqiang Shen; Yun Long; Eric Xing; Kwang-Ting Cheng; Chas Leichner
Title: Data-Free Neural Architecture Search via Recursive Label Calibration
Abstract:
This paper aims to explore the feasibility of neural architecture search (NAS) given only a pre-trained model without using any original training data. This is an important circumstance for privacy protection, bias avoidance, etc., in real-world scenarios. To achieve this, we start by synthesizing usable data through recovering the knowledge from a pre-trained deep neural network. Then we use the synthesized data and their predicted soft labels to guide NAS. We identify that the quality of the synthesized data will substantially affect the NAS results. Particularly, we find NAS requires the synthesized images to possess enough semantics, diversity, and a minimal domain gap from the natural images. To meet these requirements, we propose recursive label calibration to encode more relative semantics in images, as well as regional update strategy to enhance the diversity. Further, we use input and feature-level regularization to mimic the original data distribution in latent space and reduce the domain gap. We instantiate our proposed framework with three popular NAS algorithms: DARTS, ProxylessNAS and SPOS. Surprisingly, our results demonstrate that the architectures discovered by searching with our synthetic data achieve accuracy that is comparable to, or even higher than, architectures discovered by searching from the original ones, for the first time, deriving the conclusion that NAS can be done effectively with no need of access to the original or called natural data if the synthesis method is well designed. Code and models are availabel at: https://github.com/liuzechun/Data-Free-NAS.



Paperid:998
Authors:Zhengqi Gao; Fan-Keng Sun; Mingran Yang; Sucheng Ren; Zikai Xiong; Marc Engeler; Antonio Burazer; Linda Wildling; Luca Daniel; Duane S. Boning
Title: Learning from Multiple Annotator Noisy Labels via Sample-Wise Label Fusion
Abstract:
Data lies at the core of modern deep learning. The impressive performance of supervised learning is built upon a base of massive accurately labeled data. However, in some real-world applications, accurate labeling might not be viable; instead, multiple noisy labels (instead of one accurate label) are provided by several annotators for each data sample. Learning a classifier on such a noisy training dataset is a challenging task. Previous approaches usually assume that all data samples share the same set of parameters related to annotator errors, while we demonstrate that label error learning should be both annotator and data sample dependent. Motivated by this observation, we propose a novel learning algorithm. The proposed method displays superiority compared with several state-of-the-art baseline methods on MNIST, CIFAR-100, and ImageNet-100. Our code is available at: https://github.com/zhengqigao/Learning-from-Multiple-Annotator-Noisy-Labels.



Paperid:999
Authors:Donghao Zhou; Pengfei Chen; Qiong Wang; Guangyong Chen; Pheng-Ann Heng
Title: Acknowledging the Unknown for Multi-Label Learning with Single Positive Labels
Abstract:
Due to the difficulty of collecting exhaustive multi-label annotations, multi-label datasets often contain partial labels. We consider an extreme of this weakly supervised learning problem, called single positive multi-label learning (SPML), where each multi-label training image has only one positive label. Traditionally, all unannotated labels are assumed as negative labels in SPML, which introduces false negative labels and causes model training to be dominated by assumed negative labels. In this work, we choose to treat all unannotated labels from an alternative perspective, i.e. acknowledging they are unknown. Hence, we propose entropy-maximization (EM) loss to attain a special gradient regime for providing proper supervision signals. Moreover, we propose asymmetric pseudo-labeling (APL), which adopts asymmetric-tolerance strategies and a self-paced procedure, to cooperate with EM loss and then provide more precise supervision. Experiments show that our method significantly improves performance and achieves state-of-the-art results on all four benchmarks. Code is available at https://github.com/Correr-Zhou/SPML-AckTheUnknown.



Paperid:1000
Authors:Zicheng Liu; Siyuan Li; Di Wu; Zihan Liu; Zhiyuan Chen; Lirong Wu; Stan Z. Li
Title: AutoMix: Unveiling the Power of Mixup for Stronger Classifiers
Abstract:
Data mixing augmentation have proved to be effective for improving the generalization ability of deep neural networks. While early methods mix samples by hand-crafted policies (\textit{e.g.}, linear interpolation), recent methods utilize saliency information to match the mixed samples and labels via complex offline optimization. However, there arises a trade-off between precise mixing policies and optimization complexity. To address this challenge, we propose a novel automatic mixup (AutoMix) framework, where the mixup policy is parameterized and serves the ultimate classification goal directly. Specifically, AutoMix reformulates the mixup classification into two sub-tasks (\textit{i.e.}, mixed sample generation and mixup classification) with corresponding sub-networks and solves them in a bi-level optimization framework. For the generation, a learnable lightweight mixup generator, Mix Block, is designed to generate mixed samples by modeling patch-wise relationships under the direct supervision of the corresponding mixed labels. To prevent the degradation and instability of bi-level optimization, we further introduce a momentum pipeline to train AutoMix in an end-to-end manner. Extensive experiments on nine image benchmarks prove the superiority of AutoMix compared with state-of-the-arts in various classification scenarios and downstream tasks.



Paperid:1001
Authors:Zhengzhong Tu; Hossein Talebi; Han Zhang; Feng Yang; Peyman Milanfar; Alan Bovik; Yinxiao Li
Title: MaxViT: Multi-axis Vision Transformer
Abstract:
Transformers have recently gained significant attention in the computer vision community. However, the lack of scalability of self-attention mechanisms with respect to image size has limited their wide adoption in state-of-the-art vision backbones. In this paper we introduce an efficient and scalable attention model we call multi-axis attention, which consists of two aspects: blocked local and dilated global attention. These design choices allow global-local spatial interactions on arbitrary input resolutions with only linear complexity. We also present a new architectural element by effectively blending our proposed attention model with convolutions, and accordingly propose a simple hierarchical vision backbone, dubbed MaxViT, by simply repeating the basic building block over multiple stages. Notably, MaxViT is able to ”see” globally throughout the entire network, even in earlier, high-resolution stages. We demonstrate the effectiveness of our model on a broad spectrum of vision tasks. On image classification, MaxViT achieves state-of-the-art performance under various settings: without extra data, MaxViT attains 86.5% ImageNet-1K top-1 accuracy; with ImageNet-21K pre-training, our model achieves 88.7% top-1 accuracy. For downstream tasks, MaxViT as a backbone delivers favorable performance on object detection as well as visual aesthetic assessment. We also show that our proposed model expresses strong generative modeling capability on ImageNet, demonstrating the superior potential of MaxViT blocks as a universal vision module. The source code and trained models will be available at https://github.com/google-research/maxvit.



Paperid:1002
Authors:Rui Yang; Hailong Ma; Jie Wu; Yansong Tang; Xuefeng Xiao; Min Zheng; Xiu Li
Title: ScalableViT: Rethinking the Context-Oriented Generalization of Vision Transformer
Abstract:
The vanilla self-attention mechanism inherently relies on pre-defined and steadfast computational dimensions. Such inflexibility restricts it from possessing context-oriented generalization that can bring more contextual cues and graphic representations. To mitigate this issue, we propose a Scalable Self-Attention (SSA) mechanism that leverages two scaling factors to release dimensions of query, key, and value matrices while unbinding them with the input. This scalability fetches context-oriented generalization and enhances object sensitivity, which pushes the whole network into a more effective trade-off state between precision and cost. Furthermore, we propose an Interactive Window-based Self-Attention (IWSA), which establishes interaction between non-overlapping regions by re-merging independent value tokens and aggregating spatial information from adjacent windows. By stacking the SSA and IWSA alternately, the Scalable Vision Transformer (ScalableViT) achieves state-of-the-art performance in general-purpose vision tasks. For example, ScalableViT-S outperforms Twins-SVT-S by 1.4% and Swin-T by 1.8% on ImageNet-1K classification.



Paperid:1003
Authors:Hugo Touvron; Matthieu Cord; Alaaeldin El-Nouby; Jakob Verbeek; Hervé Jégou
Title: Three Things Everyone Should Know about Vision Transformers
Abstract:
After their initial success in natural language processing, transformer architectures have rapidly gained traction in computer vision, providing state-of-the-art results for tasks such as image classification, detection, segmentation, and video analysis. We offer three insights based on simple and easy to implement variants of vision transformers. (1) The residual layers of vision transformers, which are usually processed sequentially, can to some extent be processed efficiently in parallel without noticeably affecting the accuracy. (2) Fine-tuning the weights of the attention layers is sufficient to adapt vision transformers to a higher resolution and to other classification tasks. This saves compute, reduces the peak memory consumption at fine-tuning time, and allows sharing the majority of weights across tasks. (3) Adding MLP-based patch pre-processing layers improves Bert-like self-supervised training based on patch masking. We evaluate the impact of these design choices using the ImageNet-1k dataset, and also report result on the ImageNet-v2 test set. Transfer performance is measured across six smaller datasets.



Paperid:1004
Authors:Hugo Touvron; Matthieu Cord; Hervé Jégou
Title: DeiT III: Revenge of the ViT
Abstract:
A Vision Transformer (ViT) is a simple neural architecture amenable to serve several computer vision tasks. It has limited built-in architectural priors, in contrast to more recent architectures that incorporate priors either about the input data or of specific tasks. Recent works show that ViTs benefit from self-supervised pre-training, in particular BerT-like pre-training like BeiT. In this paper, we revisit the supervised training of ViTs. Our procedure builds upon and simplifies a recipe introduced for training ResNet-50. It includes a new simple data-augmentation procedure with only 3 augmentations, closer to the practice in self-supervised learning. Our evaluations on Image classification (ImageNet-1k with and without pre-training on ImageNet-21k), transfer learning and semantic segmentation show that our procedure outperforms by a large margin previous fully supervised training recipes for ViT. It also reveals that the performance of our ViT trained with supervision is comparable to that of more recent architectures. Our results could serve as better baselines for recent self-supervised approaches demonstrated on ViT.



Paperid:1005
Authors:Chuanguang Yang; Zhulin An; Helong Zhou; Linhang Cai; Xiang Zhi; Jiwen Wu; Yongjun Xu; Qian Zhang
Title: MixSKD: Self-Knowledge Distillation from Mixup for Image Recognition
Abstract:
Unlike the conventional Knowledge Distillation (KD), Self-KD allows a network to learn knowledge from itself without any guidance from extra networks. This paper proposes to perform Self-KD from image Mixture (MixSKD), which integrates these two techniques into a unified framework. MixSKD mutually distills feature maps and probability distributions between the random pair of original images and their mixup images in a meaningful way. Therefore, it guides the network to learn cross-image knowledge by modelling supervisory signals from mixup images. Moreover, we construct a self-teacher network by aggregating multi-stage feature maps for providing soft labels to supervise the backbone classifier, further improving the efficacy of self-boosting. Experiments on image classification and transfer learning to object detection and semantic segmentation demonstrate that MixSKD outperforms other state-of-the-art Self-KD and data augmentation methods. The code is available at https://github.com/winycg/Self-KD-Lib.



Paperid:1006
Authors:Zhou Yang; Weisheng Dong; Xin Li; Jinjian Wu; Leida Li; Guangming Shi
Title: Self-Feature Distillation with Uncertainty Modeling for Degraded Image Recognition
Abstract:
Despite the remarkable performance on high-quality (HQ) data, the accuracy of deep image recognition models degrades rapidly in the presence of low-quality (LQ) images. Both feature de-drifting and quality agnostic models have been developed to make the features extracted from degraded images closer to those of HQ images. In these methods, the L2-norm is usually used as a constraint. It treats each pixel in the feature equally and may result in relatively poor reconstruction performance in some difficult regions. To address this issue, we propose a novel self-feature distillation method with uncertainty modeling for better producing HQ-like features from low-quality observations in this paper. Specifically, in a standard recognition model, we use the HQ features to distill the corresponding degraded ones and conduct uncertainty modeling according to the diversity of degradation sources to adaptively increase the weights of feature regions that are difficult to recover in the distillation loss. Experiments demonstrate that our method can extract HQ-like features better even when the inputs are degraded images, which makes the model more robust than other approaches.



Paperid:1007
Authors:K J Joseph; Sujoy Paul; Gaurav Aggarwal; Soma Biswas; Piyush Rai; Kai Han; Vineeth N Balasubramanian
Title: Novel Class Discovery without Forgetting
Abstract:
Humans possess an innate ability to identify and differentiate instances that they are not familiar with, by leveraging and adapting the knowledge that they have acquired so far. Importantly, they achieve this without deteriorating the performance on their earlier learning. Inspired by this, we identify and formulate a new, pragmatic problem setting of NCDwF: Novel Class Discovery without Forgetting, which tasks a machine learning model to incrementally discover novel categories of instances from unlabeled data, while maintaining its performance on the previously seen categories. We propose 1) a method to generate pseudo-latent representations which act as a proxy for (no longer available) labeled data, thereby alleviating forgetting, 2) a mutual-information based regularizer which enhances unsupervised discovery of novel classes, and 3) a simple Known Class Identifier which aids generalized inference when the testing data contains instances form both seen and unseen categories. We introduce experimental protocols based on CIFAR-10, CIFAR-100 and ImageNet-1000 to measure the trade-off between knowledge retention and novel class discovery. Our extensive evaluations reveal that existing models catastrophically forget previously seen categories while identifying novel categories, while our method is able to effectively balance between the competing objectives. We hope our work will attract further research into this newly identified pragmatic problem setting.



Paperid:1008
Authors:Yan Hong; Jianfu Zhang; Zhongyi Sun; Ke Yan
Title: SAFA: Sample-Adaptive Feature Augmentation for Long-Tailed Image Classification
Abstract:
Imbalanced datasets with long-tailed distribution widely exist in practice, posing great challenges for deep networks on how to handle the biased predictions between head (majority, frequent) classes and tail (minority, rare) classes. Feature space of tail classes learned by deep networks is usually under-represented, causing heterogeneous performance among different classes. Existing methods augment tail-class features to compensate tail classes on feature space, but these methods fail to generalize on test phase. To mitigate this problem, we propose a novel Sample-Adaptive Feature Augmentation (SAFA) to augment features for tail classes resulting in ameliorating the classifier performance. SAFA aims to extract diverse and transferable semantic directions from head classes, and adaptively translate tail-class features along extracted semantic directions for augmentation. SAFA leverages a recycling training scheme ensuring augmented features are sample-specific. Contrastive loss ensures the transferable semantic directions are class-irrelevant and mode seeking loss is adopted to produce diverse tail-class features and enlarge the feature space of tail classes. The proposed SAFA as a plug-in is convenient and versatile to be combined with different methods during training phase without additional computational burden at test time. By leveraging SAFA, we obtain outstanding results on CIFAR-LT-10, CIFAR-LT-100, Places-LT, ImageNet-LT, and iNaturalist2018.



Paperid:1009
Authors:Hyungtae Lee; Sungmin Eum; Heesung Kwon
Title: Negative Samples Are at Large: Leveraging Hard-Distance Elastic Loss for Re-identification
Abstract:
We present a Momentum Re-identification (MoReID) framework that can leverage a very large number of negative samples in training for general re-identification task. The design of this framework is inspired by Momentum Contrast (MoCo), which uses a dictionary to store current and past batches to build a large set of encoded samples. As we find it less effective to use past positive samples which may be highly inconsistent to the encoded feature property formed with the current positive samples, MoReID is designed to use only a large number of negative samples stored in the dictionary. However, if we train the model using the widely used Triplet loss that uses only one sample to represent a set of positive/negative samples, it is hard to effectively leverage the enlarged set of negative samples acquired by the MoReID framework. To maximize the advantage of using the scaled-up negative sample set, we newly introduce Hard-distance Elastic loss (HE loss), which is capable of using more than one hard sample to represent a large number of samples. Our experiments demonstrate that a large number of negative samples provided by MoReID framework can be utilized at full capacity only with the HE loss, achieving the state-of-the-art accuracy on three re-ID benchmarks, VeRi-776, Market-1501, and VeRi-Wild.



Paperid:1010
Authors:Haipeng Xiong; Angela Yao
Title: Discrete-Constrained Regression for Local Counting Models
Abstract:
Local counts, or the number of objects in a local area, is a continuous value by nature. Yet recent state-of-the-art methods show that formulating counting as a classification task performs better than regression. Through a series of experiments on carefully controlled synthetic data, we show that this counter-intuitive result is caused by imprecise ground truth local counts. Factors such as biased dot annotations and incorrectly matched Gaussian kernels used to generate ground truth counts introduce deviations from the true local counts. Standard continuous regression is highly sensitive to these errors, explaining the performance gap between classification and regression. To mitigate the sensitivity, we loosen the regression formulation from a continuous scale to a discrete ordering and propose a novel discrete-constrained (DC) regression. Applied to crowd counting, DC-regression is more accurate than both classification and standard regression on three public benchmarks. A similar advantage also holds for the age estimation task, verifying the overall effectiveness of DC-regression. Code is available at https://github.com/xhp-hust-2018-2011/dcreg.



Paperid:1011
Authors:Bo Liu; Haoxiang Li; Hao Kang; Gang Hua; Nuno Vasconcelos
Title: Breadcrumbs: Adversarial Class-Balanced Sampling for Long-Tailed Recognition
Abstract:
The problem of long-tailed recognition, where the number of examples per class is highly unbalanced, is considered. While training with class-balanced sampling has been shown effective for this problem, it is known to over-fit to few-shot classes. It is hypothesized that this is due to the repeated sampling of examples and can be addressed by feature space augmentation. A new feature augmentation strategy, EMANATE, based on back-tracking of features across epochs during training, is proposed. It is shown that, unlike class-balanced sampling, this is an adversarial augmentation strategy. A new sampling procedure, Breadcrumb, is then introduced to implement adversarial class-balanced sampling without extra computation. Experiments on three popular long-tailed recognition datasets show that Breadcrumb training produces classifiers that outperform existing solutions to the problem.



Paperid:1012
Authors:Guangzhi Wang; Yangyang Guo; Yongkang Wong; Mohan Kankanhalli
Title: Chairs Can Be Stood On: Overcoming Object Bias in Human-Object Interaction Detection
Abstract:
Detecting Human-Object Interaction (HOI) in images is an important step towards high-level visual comprehension. Existing work often shed light on improving either human and object detection, or interaction recognition. However, due to the limitation of datasets, these methods tend to fit well on frequent interactions conditioned on the detected objects, yet largely ignoring the rare ones, which is referred to as the object bias problem in this paper. In this work, we for the first time, uncover the problem from two aspects: unbalanced interaction distribution and biased model learning. To overcome the object bias problem, we propose a novel plug-and-play Object-wise Debiasing Memory (ODM) method for re-balancing the distribution of interactions under detected objects. Equipped with carefully designed read and write strategies, the proposed ODM allows rare interaction instances to be more frequently sampled for training, thereby alleviating the object bias induced by the unbalanced interaction distribution. We apply this method to three advanced baselines and conduct experiments on the HICO-DET and HOI-COCO datasets. To quantitatively study the object bias problem, we advocate a new protocol for evaluating model performance. As demonstrated in the experimental results, our method brings consistent and significant improvements over baselines, especially on rare interactions under each object. In addition, when evaluating under the conventional standard setting, our method achieves new state-of-the-art on the two benchmarks.



Paperid:1013
Authors:Zhiqiang Shen; Eric Xing
Title: A Fast Knowledge Distillation Framework for Visual Recognition
Abstract:
While Knowledge Distillation (KD) has been recognized as a useful tool in many visual tasks, such as supervised classification and self-supervised representation learning, the main drawback of a vanilla KD framework is its mechanism that consumes the majority of the computational overhead on forwarding through the giant teacher networks, making the entire learning procedure inefficient and costly. The recently proposed solution ReLabel suggests creating a label map for the entire image. During training, it receives the cropped region-level label by RoI aligning on a pre-generated entire label map, which allows for efficient supervision generation without having to pass through the teachers repeatedly. However, as the pre-trained teacher employed in ReLabel is from the conventional multi-crop scheme, there are various mismatches between the global label-map and region-level labels in this technique, resulting in performance deterioration compared to the vanilla KD. In this study, we present a Fast Knowledge Distillation (FKD) framework that replicates the distillation training phase and generates soft labels using the multi-crop KD approach, meanwhile training faster than ReLabel since no post-processes such as RoI align and softmax operations are used. When conducting multi-crop in the same image for data loading, our FKD is even more efficient than the traditional image classification framework. On ImageNet-1K, we obtain 80.1% Top-1 accuracy on ResNet-50, outperforming ReLabel by 1.2% while being faster in training and more flexible to use. On the distillation-based self-supervised learning task, we also show that FKD has an efficiency advantage. Code and models are available at: https://github.com/szq0214/FKD.



Paperid:1014
Authors:Yiyou Sun; Yixuan Li
Title: DICE: Leveraging Sparsification for Out-of-Distribution Detection
Abstract:
Detecting out-of-distribution (OOD) inputs is a central challenge for safely deploying machine learning models in the real world. Previous methods commonly rely on an OOD score derived from the overparameterized weight space, while largely overlooking the role of sparsification. In this paper, we reveal important insights that reliance on unimportant weights and units can directly attribute to the brittleness of OOD detection. To mitigate the issue, we propose a sparsification-based OOD detection framework termed DICE. Our key idea is to rank weights based on a measure of contribution, and selectively use the most salient weights to derive the output for OOD detection. We provide both empirical and theoretical insights, characterizing and explaining the mechanism by which DICE improves OOD detection. By pruning away noisy signals, DICE provably reduces the output variance for OOD data, resulting in a sharper output distribution and stronger separability from ID data. We demonstrate the effectiveness of sparsification-based OOD detection on several benchmarks and establish competitive performance.



Paperid:1015
Authors:Kaihua Tang; Mingyuan Tao; Jiaxin Qi; Zhenguang Liu; Hanwang Zhang
Title: Invariant Feature Learning for Generalized Long-Tailed Classification
Abstract:
Existing long-tailed classification (LT) methods only focus on tackling the class-wise imbalance that head classes have more samples than tail classes, but overlook the attribute-wise imbalance. In fact, even if the class is balanced, samples within each class may still be long-tailed due to the varying attributes. Note that the latter is fundamentally more ubiquitous and challenging than the former because attributes are not just implicit for most datasets, but also combinatorially complex, thus prohibitively expensive to be balanced. Therefore, we introduce a novel research problem: Generalized Long-Tailed classification (GLT), to jointly consider both kinds of imbalances. By “generalized”, we mean that a GLT method should naturally solve the traditional LT, but not vice versa. Not surprisingly, we find that most class-wise LT methods degenerate in our proposed two benchmarks: ImageNet-GLT and MSCOCO-GLT. We argue that it is because they over-emphasize the adjustment of class distribution while neglecting to learn attribute-invariant features. To this end, we propose an Invariant Feature Learning (IFL) method as the first strong baseline for GLT. IFL first discovers environments with divergent intra-class distributions from the imperfect predictions, and then learns invariant features across them. Promisingly, as an improved feature backbone, IFL boosts all the LT line-up: one/two-stage re-balance, augmentation, and ensemble. Codes and benchmarks are available on Github: https://github.com/KaihuaTang/Generalized-Long-Tailed-Benchmarks.pytorch"



Paperid:1016
Authors:Zhiqiang Shen; Zechun Liu; Eric Xing
Title: Sliced Recursive Transformer
Abstract:
We present a neat yet effective recursive operation on vision transformers that can improve parameter utilization without involving additional parameters. This is achieved by sharing weights across the depth of transformer networks. The proposed method can obtain a substantial gain ( 2%) simply using naive recursive operation, requires no special or sophisticated knowledge for designing principles of networks, and introduces minimal computational overhead to the training procedure. To reduce the additional computation caused by recursive operation while maintaining the superior accuracy, we propose an approximating method through multiple sliced group self-attentions across recursive layers which can reduce the cost consumption by 10 30% without sacrificing performance. We call our model Sliced Recursive Transformer (SReT), a novel and parameter-efficient vision transformer design that is compatible with a broad range of other designs for efficient ViT architectures. Our best model establishes significant improvement on ImageNet-1K over state-of-the-art methods while containing fewer parameters. The proposed weight sharing mechanism by sliced recursion structure allows us to build a transformer with more than 100 or even 1000 shared layers with ease while keeping a compact size (13 15M), to avoid optimization difficulties when the model is too large. The flexible scalability has shown great potential for scaling up models and constructing extremely deep vision transformers. Code is available at https://github.com/szq0214/SReT.



Paperid:1017
Authors:Kyungmoon Lee; Sungyeon Kim; Suha Kwak
Title: Cross-Domain Ensemble Distillation for Domain Generalization
Abstract:
Domain generalization is the task of learning models that generalize to unseen target domains. We propose a simple yet effective method for domain generalization, named cross-domain ensemble distillation (XDED), that learns domain-invariant features while encouraging the model to converge to flat minima, which recently turned out to be a sufficient condition for domain generalization. To this end, our method generates an ensemble of the output logits from training data with the same label but from different domains and then penalizes each output for the mismatch with the ensemble. Also, we present a de-stylization technique that standardizes features to encourage the model to produce style-consistent predictions even in an arbitrary target domain. Our method greatly improves generalization capability in public benchmarks for cross-domain image classification, cross-dataset person re-ID, and cross-dataset semantic segmentation. Moreover, we show that models learned by our method are robust against adversarial attacks and unseen corruptions.



Paperid:1018
Authors:Ganlong Zhao; Guanbin Li; Yipeng Qin; Feng Liu; Yizhou Yu
Title: Centrality and Consistency: Two-Stage Clean Samples Identification for Learning with Instance-Dependent Noisy Labels
Abstract:
Deep models trained with noisy labels are prone to over-fitting and struggle in generalization. Most existing solutions are based on an ideal assumption that the label noise is class-conditional, i.e., instances of the same class share the same noise model, and are independent of features. While in practice, the real-world noise patterns are usually more fine-grained as instance-dependent ones, which poses a big challenge, especially in the presence of inter-class imbalance. In this paper, we propose a two-stage clean samples identification method to address the aforementioned challenge. First, we employ a class-level feature clustering procedure for the early identification of clean samples that are near the class-wise prediction centers. Notably, we address the class imbalance problem by aggregating rare classes according to their prediction entropy. Second, for the remaining clean samples that are close to the ground truth class boundary (usually mixed with the samples with instance-dependent noises), we propose a novel consistency-based classification method that identifies them using the consistency of two classifier heads: the higher the consistency, the larger the probability that a sample is clean. Extensive experiments on several challenging benchmarks demonstrate the superior performance of our method against the state-of-the-art.



Paperid:1019
Authors:Bo Ke; Yunquan Zhu; Mengtian Li; Xiujun Shu; Ruizhi Qiao; Bo Ren
Title: Hyperspherical Learning in Multi-Label Classification
Abstract:
Learning from online data with noisy web labels is gaining more attention due to the increasing cost of fully annotated datasets in large-scale multi-label classification tasks. Partial (positive) annotated data, as a particular case of data with noisy labels, are economically accessible. And they serve as benchmarks to evaluate the learning capacity of state-of-the-art methods in real scenarios, though they contain a large number of samples with false negative labels. Existing (partial) multi-label methods are usually studied in the Euclidean space, where the relationship between the label embeddings and image features is not symmetrical and thus can be challenging to learn. To alleviate this problem, we propose reformulating the task into a hyperspherical space, where an angular margin can be incorporated into a hyperspherical multi-label loss function. This margin allows us to effectively balance the impact of false negative and true positive labels. We further design a mechanism to tune the angular margin and scale adaptively. We investigate the effectiveness of our method under three multi-label scenarios (single positive labels, partial positive labels and full labels) on four datasets (VOC12, COCO, CUB-200 and NUS-WIDE). In the single and partial positive labels scenarios, our method achieves state-of-the-art performance. The robustness of our method is verified by comparing the performances at different proportions of partial positive labels in the datasets. Our method also obtains more than 1\% improvement over the BCE loss even on the fully annotated scenario. Analysis shows that the learned label embeddings potentially correspond to actual label correlation, since in hyperspherical space label embeddings and image features are symmetrical and interchangeable. This further indicates the geometric interpretability of our method. Code is available at https://github.com/TencentYoutuResearch/MultiLabel-HML.



Paperid:1020
Authors:Zhuangzhuang Chen; Jin Zhang; Pan Wang; Jie Chen; Jianqiang Li
Title: When Active Learning Meets Implicit Semantic Data Augmentation
Abstract:
Active learning (AL) is a label-efficient technique for training deep models when only a limited labeled set is available and the manual annotation is expensive. Implicit semantic data augmentation (ISDA) effectively extends the limited amount of labeled samples and increases the diversity of labeled sets without introducing a noticeable extra computational cost. The scarcity of labeled instances and the huge annotation cost of unlabelled samples encourage us to ponder on the combination of AL and ISDA. A nature direction is a pipelined integration, which selects the unlabeled samples via acquisition function in AL for labeling and generates virtual samples by changing the selected samples to semantic transformation directions within ISDA. However, this pipelined combination would not guarantee the diversity of virtual samples. This paper proposes diversity-aware semantic transformation active learning, or DAST-AL framework, that looks ahead the effect of ISDA in the selection of unlabeled samples. Specifically, DAST-AL exploits expected partial model change maximization (EPMCM) to consider selected samples’ potential contribution of the diversity to the labeled set by leveraging the semantic transformation within ISDA when selecting the unlabeled samples. After that, DAST-AL can confidently and efficiently augment the labeled set by implicitly generating more diverse samples. The empirical results on both image classification and semantic segmentation tasks show that the proposed DAST-AL can slightly outperform the state-of-the-art AL approaches. Under the same condition, the proposed method takes less than 3 minutes for the first cycle of active labeling while the existing agreement discrepancy selection incurs more than 40 minutes.



Paperid:1021
Authors:Changyao Tian; Wenhai Wang; Xizhou Zhu; Jifeng Dai; Yu Qiao
Title: VL-LTR: Learning Class-Wise Visual-Linguistic Representation for Long-Tailed Visual Recognition
Abstract:
Recently, computer vision foundation models such as CLIP and ALI-GN, have shown impressive generalization capabilities on various downstream tasks. But their abilities to deal with the long-tailed data still remain to be proved. In this work, we present a novel framework based on pre-trained visual-linguistic models for long-tailed recognition (LTR), termed VL-LTR, and conduct empirical studies on the benefits of introducing text modality for long-tailed recognition tasks. Compared to existing approaches, the proposed VL-LTR has the following merits. (1) Our method can not only learn visual representation from images but also learn corresponding linguistic representation from noisy class-level text descriptions collected from the Internet; (2) Our method can effectively use the learned visual-linguistic representation to improve the visual recognition performance, especially for classes with fewer image samples. We also conduct extensive experiments and set the new state-of-the-art performance on widely-used LTR benchmarks. Notably, our method achieves 77.2\% overall accuracy on ImageNet-LT, which significantly outperforms the previous best method by over 17 points, and is close to the prevailing performance training on the full ImageNet. Code shall be released.



Paperid:1022
Authors:Jiaxin Qi; Kaihua Tang; Qianru Sun; Xian-Sheng Hua; Hanwang Zhang
Title: Class Is Invariant to Context and Vice Versa: On Learning Invariance for Out-of-Distribution Generalization
Abstract:
Out-Of-Distribution generalization (OOD) is all about learning invariance against environmental changes. If the context in every class is evenly distributed, OOD would be trivial because the context can be easily removed due to an underlying principle: class is invariant to context. However, collecting such a balanced dataset is impractical. Learning on imbalanced data makes the model bias to context and thus hurts OOD. Therefore, the key to OOD is context balance.We argue that the widely adopted assumption in prior work--the context bias can be directly annotated or estimated from biased class prediction--renders the context incomplete or even incorrect. In contrast, we point out the everoverlooked other side of the above principle: context is also invariant to class, which motivates us to consider the classes (which are already labeled) as the varying environments to resolve context bias (without context labels). We implement this idea by minimizing the contrastive loss of intra-class sample similarity while assuring this similarity to be invariant across all classes. On benchmarks with various context biases and domain gaps, we show that a simple re-weighting based classifier equipped with our context estimation achieves state-of-the-art performance. We provide theoretical justifications and source code in Appendix.



Paperid:1023
Authors:Gaoang Wang; Yibing Zhan; Xinchao Wang; Mingli Song; Klara Nahrstedt
Title: Hierarchical Semi-Supervised Contrastive Learning for Contamination-Resistant Anomaly Detection
Abstract:
Anomaly detection aims at identifying deviant samples from the normal data distribution. Contrastive learning has provided a successful way to sample representation that enables effective discrimination on anomalies. However, when contaminated with unlabeled abnormal samples in training set under semi-supervised settings, current contrastive-based methods generally 1) ignore the comprehensive relation between training data, leading to suboptimal performance, and 2) require fine-tuning, resulting in low efficiency. To address the above two issues, in this paper, we propose a novel hierarchical semi-supervised contrastive learning (HSCL) framework, for contamination-resistant anomaly detection. Specifically, HSCL hierarchically regulates three complementary relations: sample-to-sample, sample-to-prototype, and normal-to-abnormal relations, enlarging the discrimination between normal and abnormal samples with a comprehensive exploration of the contaminated data. Besides, HSCL is an end-to-end learning approach that can efficiently learn discriminative representations without fine-tuning. HSCL achieves state-of-the-art performance in multiple scenarios, such as one-class classification and cross-dataset detection. Extensive ablation studies further verify the effectiveness of each considered relation. The code is available at https://github.com/GaoangW/HSCL.



Paperid:1024
Authors:Sanghyun Woo; Kwanyong Park; Seoung Wug Oh; In So Kweon; Joon-Young Lee
Title: Tracking by Associating Clips
Abstract:
The tracking-by-detection paradigm today has become the dominant method for multi-object tracking and works by detecting objects in each frame and then performing data association across frames. However, its sequential frame-wise matching property fundamentally suffers from the intermediate interruptions in a video, such as object occlusions, fast camera movements, and abrupt light changes. Moreover, it typically overlooks temporal information beyond the two frames for matching. In this paper, we investigate an alternative by treating object association as clip-wise matching. Our new perspective views a single long video sequence as multiple short clips, and then the tracking is performed both within and between the clips. The benefits of this new approach are two folds. First, our method is robust to tracking error accumulation or propagation, as the video chunking allows bypassing the interrupted frames, and the short clip tracking avoids the conventional error-prone long-term track memory management. Second, the multiple frame information is aggregated during the clip-wise matching, resulting in a more accurate long-range track association than the current frame-wise matching. Given the state-of-the-art tracking-by-detection tracker, QDTrack, we showcase how the tracking performance improves with our new tracking formulation. In addition, we designed a novel Transformer-based clip tracker tailored for the effective inter-clip association. We evaluate our proposals on two tracking benchmarks, TAO and MOT17 that have complementary characteristics and challenges each other. Our new tracking formulation greatly improves association quality, leading to a new state-of-the-art in TAO.



Paperid:1025
Authors:Sara Romiti; Christopher Inskip; Viktoriia Sharmanska; Novi Quadrianto
Title: RealPatch: A Statistical Matching Framework for Model Patching with Real Samples
Abstract:
Machine learning classifiers are typically trained to minimise the average error across a dataset. Unfortunately, in practice, this process often exploits spurious correlations caused by subgroup imbalance within the training data, resulting in high average performance but highly variable performance across subgroups. Recent work to address this problem proposes model patching with CAMEL. This previous approach uses generative adversarial networks to perform intra-class inter-subgroup data augmentations, requiring (a) the training of a number of computationally expensive models and (b) sufficient quality of model’s synthetic outputs for the given domain. In this work, we propose RealPatch, a framework for simpler, faster, and more data-efficient data augmentation based on statistical matching. Our framework performs model patching by augmenting a dataset with real samples, mitigating the need to train generative models for the target task. We demonstrate the effectiveness of RealPatch on three benchmark datasets, CelebA, Waterbirds and a subset of iWildCam, showing improvements in worst-case subgroup performance and in subgroup performance gap in binary classification. Furthermore, we conduct experiments with the imSitu dataset with 211 classes, a setting where generative model-based patching such as CAMEL is impractical. We show that RealPatch can successfully eliminate dataset leakage while reducing model leakage and maintaining high utility. The code for RealPatch can be found at https://github.com/wearepal/RealPatch.



Paperid:1026
Authors:Liang Zhao; Zhenyao Wu; Xinyi Wu; Greg Wilsbacher; Song Wang
Title: Background-Insensitive Scene Text Recognition with Text Semantic Segmentation
Abstract:
Scene Text Recognition (STR) has many important applications in computer vision. Complex backgrounds continue to be a big challenge for STR because they interfere with text feature extraction. Many existing methods use attentional regions, bounding boxes or polygons to reduce such interference. However, the text regions located by these methods still contain much undesirable background interference. In this paper, we propose a Background-Insensitive approach BINet by explicitly leveraging the text Semantic Segmentation (SSN) to extract texts more accurately. SSN is trained on a set of existing segmentation data, whose volume is only 0.03% of STR training data. This prevents the large-scale pixel-level annotations of the STR training data. To effectively utilize the segmentation cues, we design new segmentation refinement and embedding blocks for refining text-masks and reinforcing visual features. Additionally, we propose an efficient pipeline that utilizes Synthetic Initialization (SI) for STR models trained only on real data (1.7% of STR training data), instead of on both synthetic and real data from scratch. Experiments show that the proposed method can recognize text from complex backgrounds more effectively, achieving state-of-the-art performance on several public datasets.



Paperid:1027
Authors:Francesco Cappio Borlino; Silvia Bucci; Tatiana Tommasi
Title: Semantic Novelty Detection via Relational Reasoning
Abstract:
Semantic novelty detection aims at discovering unknown categories in the test data. This task is particularly relevant in safety-critical applications, such as autonomous driving or healthcare, where it is crucial to recognize unknown objects at deployment time and issue a warning to the user accordingly. Despite the impressive advancements of deep learning research, existing models still need a finetuning stage on the known categories in order to recognize the unknown ones. This could be prohibitive when privacy rules limit data access, or in case of strict memory and computational constraints (e.g. edge computing). We claim that a tailored representation learning strategy may be the right solution for effective and efficient semantic novelty detection. Besides extensively testing state-of-the-art approaches for this task, we propose a novel representation learning paradigm based on relational reasoning. It focuses on learning how to measure semantic similarity rather than recognizing known categories. Our experiments show that this knowledge is directly transferable to a wide range of scenarios, and it can be exploited as a plug-and-play module to convert closed-set recognition models into reliable open-set ones.



Paperid:1028
Authors:Khoi Pham; Kushal Kafle; Zhe Lin; Zhihong Ding; Scott Cohen; Quan Tran; Abhinav Shrivastava
Title: Improving Closed and Open-Vocabulary Attribute Prediction Using Transformers
Abstract:
We study recognizing attributes for objects in visual scenes. We consider attributes to be any phrases that describe an object’s physical and semantic properties, and its relationships with other objects. Existing work studies attribute prediction in a closed setting with a fixed set of attributes, and implements a model that uses limited context. We propose TAP, a new Transformer-based model that can utilize context and predict attributes for multiple objects in a scene in a single forward pass, and a training scheme that allows this model to learn attribute prediction from image-text datasets. Experiments on the large closed attribute benchmark VAW show that TAP outperforms the SOTA by 5.1% mAP. In addition, by utilizing pretrained text embeddings, we extend our model to OpenTAP which can recognize novel attributes not seen during training. In a large-scale setting, we further show that OpenTAP can predict a large number of seen and unseen attributes that outperforms large-scale vision-text model CLIP by a decisive margin.



Paperid:1029
Authors:Yun-Hao Cao; Hao Yu; Jianxin Wu
Title: Training Vision Transformers with Only 2040 Images
Abstract:
Vision Transformers (ViTs) is emerging as an alternative to convolutional neural networks (CNNs) for visual recognition. They achieve competitive results with CNNs but the lack of the typical convolutional inductive bias makes them more data-hungry than common CNNs. They are often pretrained on JFT-300M or at least ImageNet and few works study training ViTs with limited data. In this paper, we investigate how to train ViTs with limited data (e.g., 2040 images). We give theoretical analyses that our method (based on parametric instance discrimination) is superior to other methods in that it can capture both feature alignment and instance similarities. We achieve state-of-the-art results when training from scratch on 7 small datasets under various ViT backbones. We also investigate the transferring ability of small datasets and find that representations learned from small datasets can even improve large-scale ImageNet training.



Paperid:1030
Authors:Sanghyun Woo; Kwanyong Park; Seoung Wug Oh; In So Kweon; Joon-Young Lee
Title: Bridging Images and Videos: A Simple Learning Framework for Large Vocabulary Video Object Detection
Abstract:
Scaling object taxonomies is one of the important steps toward a robust real-world deployment of recognition systems. We have faced remarkable progress in images since the introduction of the LVIS benchmark. To continue this success in videos, a new video benchmark, TAO, was recently presented. Given the recent encouraging results from both large-vocabulary detection and tracking communities, we are interested in marrying those two advances and building a strong large vocabulary video tracker. However, supervisions in LVIS and TAO are inherently sparse or even missing, posing two new challenges for training the trackers. First, no tracking supervisions are in LVIS, which leads to inconsistent learning of detection (with LVIS and TAO) and tracking (only with TAO). Second, the detection supervisions in TAO are partial, which results in catastrophic forgetting of absent LVIS categories. To resolve these challenges, we present an effective unified learning framework that takes full advantage of all available training data to learn detection and tracking while not losing any LVIS categories to recognize. With this new learning scheme, we show that consistent improvements of various large vocabulary trackers are capable, setting strong state-of-the-art results on TAO benchmarks.



Paperid:1031
Authors:Shantanu Jaiswal; Basura Fernando; Cheston Tan
Title: TDAM: Top-Down Attention Module for Contextually Guided Feature Selection in CNNs
Abstract:
Attention modules for Convolutional Neural Networks (CNNs) are an effective method to enhance performance on multiple computer-vision tasks. While existing methods appropriately model channel-, spatial- and self-attention, they primarily operate in a feedforward bottom-up manner. Consequently, the attention mechanism strongly depends on the local information of a single input feature map and does not incorporate relatively semantically-richer contextual information available at higher layers that can specify “what and where to look” in lower-level feature maps through top-down information flow. Accordingly, in this work, we propose a lightweight top-down attention module (TDAM) that iteratively generates a “visual searchlight” to perform channel and spatial modulation of its inputs and outputs more contextually-relevant feature maps at each computation step. Our experiments indicate that TDAM enhances the performance of CNNs across multiple object-recognition benchmarks and outperforms prominent attention modules while being more parameter and memory efficient. Further, TDAM-based models learn to “shift attention” by localizing individual objects or features at each computation step without any explicit supervision resulting in a 5% improvement for ResNet50 on weakly-supervised object localization. Source code and models are publicly available at: https://github.com/shantanuj/TDAM_Top_down_attention_module"



Paperid:1032
Authors:Hao Chen; Xiu-Shen Wei; Faen Zhang; Yang Shen; Hui Xu; Liang Xiao
Title: Automatic Check-Out via Prototype-Based Classifier Learning from Single-Product Exemplars
Abstract:
Automatic Check-Out (ACO) aims to accurately predict the presence and count of each category of products in check-out images, where a major challenge is the significant domain gap between training data (single-product exemplars) and test data (check-out images). To mitigate the gap, we propose a method, termed as PSP, to perform Prototype-based classifier learning from Single-Product exemplars. In PSP, by revealing the advantages of representing category semantics, the prototype representation of each product category is firstly obtained from single-product exemplars. Based on the prototypes, it then generates categorical classifiers with a background classifier to not only recognize fine-grained product categories but also distinguish background upon product proposals derived from check-out images. To further improve ACO accuracy, we develop discriminative re-ranking to both adjust the predicted scores of product proposals for bringing more discriminative ability in classifier learning and provide a reasonable sorting possibility by considering the fine-grained nature. Moreover, a multi-label recognition loss is also equipped for modeling co-occurrence of products in check-out images. Experiments are conducted on the large-scale RPC dataset for evaluations. Our ACO result achieves 86.69%, by 6.18% improvements over state-of-the-arts, which demonstrates the superiority of PSP. Our codes are available at https://github.com/Hao-Chen-NJUST/PSP.



Paperid:1033
Authors:Piyapat Saranrittichai; Chaithanya Kumar Mummadi; Claudia Blaiotta; Mauricio Munoz; Volker Fischer
Title: Overcoming Shortcut Learning in a Target Domain by Generalizing Basic Visual Factors from a Source Domain
Abstract:
Shortcut learning occurs when a deep neural network overly relies on spurious correlations in the training dataset in order to solve downstream tasks. Prior works have shown how this impairs the compositional generalization capability of deep learning models. To address this problem, we propose a novel approach to mitigate shortcut learning in uncontrolled target domains. Our approach extends the training set with an additional dataset (the source domain), which is specifically designed to facilitate learning independent representations of basic visual factors. We benchmark our idea on synthetic target domains where we explicitly control shortcut opportunities as well as real-world target domains. Furthermore, we analyze the effect of different specifications of the source domain and the network architecture on compositional generalization. Our main finding is that leveraging data from a source domain is an effective way to mitigate shortcut learning. By promoting independence across different factors of variation in the learned representations, networks can learn to consider only predictive factors and ignore potential shortcut factors during inference.



Paperid:1034
Authors:Sergey Zakharov; Rareș Ambruș; Vitor Guizilini; Wadim Kehl; Adrien Gaidon
Title: Photo-Realistic Neural Domain Randomization
Abstract:
Synthetic data is a scalable alternative to manual supervision, but it requires overcoming the sim-to-real domain gap. This discrepancy between virtual and real worlds is addressed by two seemingly opposed approaches: improving the realism of simulation or foregoing realism entirely via domain randomization. In this paper, we show that the recent progress in neural rendering enables a new unified approach we call Photo-realistic Neural Domain Randomization (PNDR). We propose to learn a composition of neural networks that acts as a physics-based ray tracer generating high-quality renderings from scene geometry alone. Our approach is modular, composed of different neural networks for materials, lighting, and rendering, thus enabling randomization of different key image generation components in a differentiable pipeline. Once trained, our method can be combined with other methods and used to generate photo-realistic image augmentations online and significantly more efficiently than via traditional ray-tracing. We demonstrate the usefulness of PNDR through two downstream tasks: 6D object detection and monocular depth estimation. Our experiments show that training with PNDR enables generalization to novel scenes and significantly outperforms the state of the art in terms of real-world transfer.



Paperid:1035
Authors:Ting Yao; Yingwei Pan; Yehao Li; Chong-Wah Ngo; Tao Mei
Title: Wave-ViT: Unifying Wavelet and Transformers for Visual Representation Learning
Abstract:
Multi-scale Vision Transformer (ViT) has emerged as a powerful backbone for computer vision tasks, while the self-attention computation in Transformer scales quadratically w.r.t. the input patch number. Thus, existing solutions commonly employ down-sampling operations (e.g., average pooling) over keys/values to dramatically reduce the computational cost. In this work, we argue that such over-aggressive down-sampling design is not invertible and inevitably causes information dropping especially for high-frequency components in objects (e.g., texture details). Motivated by the wavelet theory, we construct a new Wavelet Vision Transformer (Wave-ViT) that formulates the invertible down-sampling with wavelet transforms and self-attention learning in a unified way. This proposal enables self-attention learning with lossless down-sampling over keys/values, facilitating the pursuing of a better efficiency-vs-accuracy trade-off. Furthermore, inverse wavelet transforms are leveraged to strengthen self-attention outputs by aggregating local contexts with enlarged receptive field. We validate the superiority of Wave-ViT through extensive experiments over multiple vision tasks (e.g., image recognition, object detection and instance segmentation). Its performances surpass state-of-the-art ViT backbones with comparable FLOPs. Source code is available at \url{https://github.com/YehLi/ImageNetModel}.



Paperid:1036
Authors:WonJun Moon; Ji-Hwan Kim; Jae-Pil Heo
Title: Tailoring Self-Supervision for Supervised Learning
Abstract:
Recently, it is shown that deploying a proper self-supervision is a prospective way to enhance the performance of supervised learning. Yet, the benefits of self-supervision are not fully exploited as previous pretext tasks are specialized for unsupervised representation learning. To this end, we begin by presenting three desirable properties for such auxiliary tasks to assist the supervised objective. First, the tasks need to guide the model to learn rich features. Second, the transformations involved in the self-supervision should not significantly alter the training distribution. Third, the tasks are preferred to be light and generic for high applicability to prior arts. Subsequently, to show how existing pretext tasks can fulfill these and be tailored for supervised learning, we propose a simple auxiliary self-supervision task, predicting localizable rotation (\textbf{LoRot}). Our exhaustive experiments validate the merits of LoRot as a pretext task tailored for supervised learning in terms of robustness and generalization capability.



Paperid:1037
Authors:WonJun Moon; Junho Park; Hyun Seok Seong; Cheol-Ho Cho; Jae-Pil Heo
Title: Difficulty-Aware Simulator for Open Set Recognition
Abstract:
Open set recognition (OSR) assumes unknown instances appear out of the blue at the inference time. The main challenge of OSR is that the response of models for unknowns is totally unpredictable. Furthermore, the diversity of open set makes it harder since instances have different difficulty levels. Therefore, we present a novel framework, DIfficulty-Aware Simulator (DIAS), that generates fakes with diverse difficulty levels to simulate the real world. We first investigate fakes from generative adversarial network (GAN) in the classifier’s viewpoint and observe that these are not severely challenging. This leads us to define the criteria for difficulty by regarding samples generated with GANs having moderate-difficulty. To produce hard-difficulty examples, we introduce Copycat, imitating the behavior of the classifier. Furthermore, moderate- and easy-difficulty samples are also yielded by our modified GAN and Copycat, respectively. As a result, DIAS outperforms state-of-the-art methods with both metrics of AUROC and F-score.



Paperid:1038
Authors:Can Peng; Kun Zhao; Tianren Wang; Meng Li; Brian C. Lovell
Title: Few-Shot Class-Incremental Learning from an Open-Set Perspective
Abstract:
The continual appearance of new objects in the visual world poses considerable challenges for current deep learning methods in real-world deployments. The challenge of new task learning is often exacerbated by the scarcity of data for the new categories due to rarity or cost. Here we explore the important task of Few-Shot Class-Incremental Learning (FSCIL) and its extreme data scarcity condition of one-shot. An ideal FSCIL model needs to perform well on all classes, regardless of their presentation order or paucity of data. It also needs to be robust to open-set real-world conditions and be easily adapted to the new tasks that always arise in the field. In this paper, we first reevaluate the current task setting and propose a more comprehensive and practical setting for the FSCIL task. Then, inspired by the similarity of the goals for FSCIL and modern face recognition systems, we propose our method --- Augmented Angular Loss Incremental Classification or ALICE. In ALICE, instead of the commonly used cross-entropy loss, we propose to use the angular penalty loss to obtain well-clustered features. As the obtained features not only need to be compactly clustered but also diverse enough to maintain generalization for future incremental classes, we further discuss how class augmentation, data augmentation, and data balancing affect classification performance. Experiments on benchmark datasets, including CIFAR100, miniImageNet, and CUB200, demonstrate the improved performance of ALICE over the state-of-the-art FSCIL methods.



Paperid:1039
Authors:Fu-Yun Wang; Da-Wei Zhou; Han-Jia Ye; De-Chuan Zhan
Title: FOSTER: Feature Boosting and Compression for Class-Incremental Learning
Abstract:
The ability to learn new concepts continually is necessary in this ever-changing world. However, deep neural networks suffer from catastrophic forgetting when learning new categories. Many works have been proposed to alleviate this phenomenon, whereas most of them either fall into the stability-plasticity dilemma or take too much computation or storage overhead. Inspired by the gradient boosting algorithm to gradually fit the residuals between the target model and the previous ensemble model, we propose a novel two-stage learning paradigm FOSTER, empowering the model to learn new categories adaptively. Specifically, we first dynamically expand new modules to fit the residuals between the target and the output of the original model. Next, we remove redundant parameters and feature dimensions through an effective distillation strategy to maintain the single backbone model. We validate our method FOSTER on CIFAR-100 and ImageNet-100/1000 under different settings. Experimental results show that our method achieves state-of-the-art performance. Code is available at https://github.com/G-U-N/ECCV22-FOSTER.



Paperid:1040
Authors:Neehar Kondapaneni; Pietro Perona; Oisin Mac Aodha
Title: Visual Knowledge Tracing
Abstract:
Each year, thousands of people learn new visual categorization tasks - radiologists learn to recognize tumors, birdwatchers learn to distinguish similar species, and crowd workers learn how to annotate valuable data for applications like autonomous driving. As humans learn, their brain updates the visual features it extracts and attend to, which ultimately informs their final classification decisions. In this work, we propose a novel task of tracing the evolving classification behavior of human learners as they engage in challenging visual classification tasks. We propose models that jointly extract the visual features used by learners as well as predicting the classification functions they utilize. We collect three challenging new datasets from real human learners in order to evaluate the performance of different visual knowledge tracing methods. Our results show that our recurrent models are able to predict the classification behavior of human learners on three challenging medical image and species identification tasks.



Paperid:1041
Authors:Jayateja Kalla; Soma Biswas
Title: S3C: Self-Supervised Stochastic Classifiers for Few-Shot Class-Incremental Learning
Abstract:
Few-shot class-incremental learning (FSCIL) aims to learn progressively about new classes with very few labeled samples, without forgetting the knowledge of already learnt classes. FSCIL suffers from two major challenges: (i) over-fitting on the new classes due to limited amount of data, (ii) catastrophically forgetting about the old classes due to unavailability of data from these classes in the incremental stages. In this work, we propose a self-supervised stochastic classifier (S3C) to counter both these challenges in FSCIL. The stochasticity of the classifier weights (or class prototypes) not only mitigates the adverse effect of absence of large number of samples of the new classes, but also the absence of samples from previously learnt classes during the incremental steps. This is complemented by the self-supervision component, which helps to learn features from the base classes which generalize well to unseen classes that are encountered in future, thus reducing catastrophic forgetting. Extensive evaluation on three benchmark datasets using multiple evaluation metrics show the effectiveness of the proposed framework. We also experiment on two additional realistic scenarios of FSCIL, namely where the number of annotated data available for each of the new classes can be different, and also where the number of base classes is much lesser, and show that the proposed S3C performs significantly better than the state-of-the-art for all these challenging scenarios.



Paperid:1042
Authors:Yangyang Shu; Baosheng Yu; Haiming Xu; Lingqiao Liu
Title: Improving Fine-Grained Visual Recognition in Low Data Regimes via Self-Boosting Attention Mechanism
Abstract:
The challenge of fine-grained visual recognition often lies in discovering the key discriminative regions. While such regions can be automatically identified from a large-scale labeled dataset, a similar method might become less effective when only a few annotations are available. In low data regimes, a network often struggles to choose the correct regions for recognition and tends to overfit spurious correlated patterns from the training data. To tackle this issue, this paper proposes the self-boosting attention mechanism, a novel method for regularizing the network to focus on the key regions shared across samples and classes. Specifically, the proposed method first generates an attention map for each training image, highlighting the discriminative part for identifying the ground-truth object category. Then the generated attention maps are used as pseudo-annotations. The network is enforced to fit them as an auxiliary task. We call this approach the self-boosting attention mechanism (SAM). We also develop a variant by using SAM to create multiple attention maps to pool convolutional maps in a style of bilinear pooling, dubbed SAM-Bilinear. Through extensive experimental studies, we show that both methods can significantly improve fine-grained visual recognition performance on low data regimes and can be incorporated into existing network architectures.



Paperid:1043
Authors:Qiming Zhang; Yufei Xu; Jing Zhang; Dacheng Tao
Title: VSA: Learning Varied-Size Window Attention in Vision Transformers
Abstract:
Attention within windows has been widely explored in vision transformers to balance the performance, computation complexity, and memory footprint. However, current models adopt a hand-crafted fixed-size window design, which restricts their capacity of modeling long-term dependencies and adapting to objects of different sizes. To address this drawback, we propose Varied-Size Window Attention (VSA) to learn adaptive window configurations from data. Specifically, based on the tokens within each default window, VSA employs a window regression module to predict the size and location of the target window, \ie, the attention area where the key and value tokens are sampled. By adopting VSA independently for each attention head, it can model long-term dependencies, capture rich context from diverse windows, and promote information exchange among overlapped windows. VSA is an easy-to-implement module that can replace the window attention in state-of-the-art representative models with minor modifications and negligible extra computational cost while improving their performance by a large margin, e.g., 1.1 % for Swin-T on ImageNet classification. In addition, the performance gain increases when using larger images for training and test. Experimental results on more downstream tasks, including object detection, instance segmentation, and semantic segmentation, further demonstrate the superiority of VSA over the vanilla window attention in dealing with objects of different sizes. The code is available at https://github.com/ViTAE-Transformer/ViTAE-VSA.



Paperid:1044
Authors:Baoming Yan; Ke Gao; Bo Gao; Lin Wang; Jiang Yang; Xiaobo Li
Title: Unbiased Manifold Augmentation for Coarse Class Subdivision
Abstract:
Class Subdivision (CCS) is important for many practical applications, where the training set originally annotated for a coarse class (e.g. bird) needs to further support its sub-classes recognition (e.g. swan, crow) with only very few fine-grained labeled samples. From the perspective of causal representation learning, these sub-classes inherit the same determinative factors of the coarse class, and their difference lies only in values. Therefore, to support the challenging CCS task with minimum fine-grained labeling cost, an ideal data augmentation method should generate abundant variants by manipulating these sub-class samples at the granularity of generating factors. For this goal, traditional data augmentation methods are far from sufficient. They often perform in highly-coupled image or feature space, thus can only simulate global geometric or photometric transformations. Leveraging the recent progress of factor-disentangled generators, Unbiased Manifold Augmentation (UMA) is proposed for CCS. With a controllable StyleGAN pre-trained for a coarse class, an approximate unbiased augmentation is conducted on the factor-disentangled manifolds for each sub-class, revealing the unbiased mutual information between the target sub-class and its determinative factors. Extensive experiments have shown that in the case of small data learning (less than 1\% fine-grained samples of commonly used), our UMA can achieve 10.37\% average improvement compared with existing data augmentation methods. On challenging tasks with severe bias, the accuracy is improved by up to 16.79\%.



Paperid:1045
Authors:Matej Grcić; Petra Bevandić; Siniša Šegvić
Title: DenseHybrid: Hybrid Anomaly Detection for Dense Open-Set Recognition
Abstract:
Anomaly detection can be conceived either through generative modelling of regular training data or by discriminating with respect to negative training data. These two approaches exhibit different failure modes. Consequently, hybrid algorithms present an attractive research goal. Unfortunately, dense anomaly detection requires translational equivariance and very large input resolutions. These requirements disqualify all previous hybrid approaches to the best of our knowledge. We therefore design a novel hybrid algorithm based on reinterpreting discriminative logits as a logarithm of the unnormalized joint distribution $\hat{p}(\mathbf{x},\mathbf{y})$. Our model builds on a shared convolutional representation from which we recover three dense predictions: i) the closed-set class posterior $P(\mathbf{y}|\mathbf{x})$, ii) the dataset posterior $P(\mathbf{d}_{in}|\mathbf{x})$, iii) unnormalized data likelihood $\hat{p}(\mathbf{x})$. The latter two predictions are trained both on the standard training data and on a generic negative dataset. We blend these two predictions into a hybrid anomaly score which allows dense open-set recognition on large natural images. We carefully design a custom loss for the data likelihood in order to avoid backpropagation through the untractable normalizing constant $Z(\theta)$. Experiments evaluate our contributions on standard dense anomaly detection benchmarks as well as in terms of open-mIoU - a novel metric for dense open-set performance. Our submissions achieve state-of-the-art performance despite neglectable computational overhead over the standard semantic segmentation baseline.



Paperid:1046
Authors:Fei Zhu; Zhen Cheng; Xu-Yao Zhang; Cheng-Lin Liu
Title: Rethinking Confidence Calibration for Failure Prediction
Abstract:
Reliable confidence estimation for the predictions is important in many safety-critical applications. However, modern deep neural networks are often overconfident for their incorrect predictions. Recently, many calibration methods have been proposed to alleviate the overconfidence problem. With calibrated confidence, a primary and practical purpose is to detect misclassification errors by filtering out low-confidence predictions (known as failure prediction). In this paper, we find a general, widely-existed but actually-neglected phenomenon that most confidence calibration methods are useless or harmful for failure prediction. We investigate this problem and reveal that popular confidence calibration methods often lead to worse confidence separation between correct and incorrect samples, making it more difficult to decide whether to trust a prediction or not. Finally, inspired by the natural connection between flat minima and confidence separation, we propose a simple hypothesis: flat minima is beneficial for failure prediction. We verify this hypothesis via extensive experiments and further boost the performance by combining two different flat minima techniques.



Paperid:1047
Authors:Subhankar Roy; Martin Trapp; Andrea Pilzer; Juho Kannala; Nicu Sebe; Elisa Ricci; Arno Solin
Title: Uncertainty-Guided Source-Free Domain Adaptation
Abstract:
Source-free domain adaptation (SFDA) aims to adapt a classifier to an unlabelled target data set by only using a pre-trained source model. However, the absence of the source data and the domain shift makes the predictions on the target data unreliable. We propose quantifying the uncertainty in the source model predictions and utilizing it to guide the target adaptation. For this, we construct a probabilistic source model by incorporating priors on the network parameters inducing a distribution over the model predictions. Uncertainties are estimated by employing a Laplace approximation and incorporated to identify target data points that do not lie in the source manifold and to down-weight them when maximizing the mutual information on the target data. Unlike recent works, our probabilistic treatment is computationally lightweight, decouples source training and target adaptation, and requires no specialized source training or changes of the model architecture. We show the advantages of uncertainty-guided SFDA over traditional SFDA in the closed-set and open-set settings and provide empirical evidence that our approach is more robust to strong domain shifts even without tuning.



Paperid:1048
Authors:Yunsheng Li; Yinpeng Chen; Xiyang Dai; Dongdong Chen; Mengchen Liu; Pei Yu; Ying Jin; Lu Yuan; Zicheng Liu; Nuno Vasconcelos
Title: Should All Proposals Be Treated Equally in Object Detection?
Abstract:
The complexity-precision trade-off of an object detector is a critical problem for resource constrained vision tasks. Previous works have emphasized detectors implemented with efficient backbones. The impact on this trade-off of proposal processing by the detection head is investigated in this work. It is hypothesized that improved detection efficiency requires a paradigm shift, towards the unequal processing of proposals, assigning more computation to good proposals than poor ones. This results in better utilization of available computational budget, enabling higher accuracy for the same FLOPS. We formulate this as a learning problem where the goal is to assign operators to proposals, in the detection head, so that the total computational cost is constrained and the precision is maximized. The key finding is that such matching can be learned as a function that maps each proposal embedding into a one-hot code over operators. While this function induces a complex dynamic network routing mechanism, it can be implemented by a simple MLP and learned end-to-end with off-the-shelf object detectors. This it dynamic proposal processing (DPP) is shown to outperform state-of-the-art end-to-end object detectors (DETR, Sparse R-CNN) by a clear margin for a given computational complexity.



Paperid:1049
Authors:Junbo Li; Huan Zhang; Cihang Xie
Title: VIP: Unified Certified Detection and Recovery for Patch Attack with Vision Transformers
Abstract:
Patch attack, which introduces a perceptible but localized change to the input image, has gained significant momentum in recent years. In this paper, we propose a unified framework to analyze certified patch defense tasks (including both certified detection and certified recovery) using the recently emerged vision transformer. In addition to the existing patch defense setting where only one patch is considered, we provide the very first study on developing certified detection against the \emph{dual patch attack}, in which the attacker is allowed to adversarially manipulate pixels in two different regions. Benefiting from the recent progress in self-supervised vision transformers (\ie, masked autoencoder), our method achieves state-of-the-art performance in both certified detection and certified recovery of adversarial patches. For certified detection, we improve the performance by up to $\app16\%$ on ImageNet without additional training for a single adversarial patch, and for the first time, can also tackle the more challenging dual patch setting. Our method largely \emph{closes the gap} between detection-based certified robustness and clean image accuracy. For certified recovery, our approach improves certified accuracy by $\app2\%$ on ImageNet across all attack sizes, attaining the new state-of-the-art performance.



Paperid:1050
Authors:Amanda Rios; Nilesh Ahuja; Ibrahima Ndiour; Utku Genc; Laurent Itti; Omesh Tickoo
Title: incDFM: Incremental Deep Feature Modeling for Continual Novelty Detection
Abstract:
Novelty detection is a key capability for practical machine learning in the real world, where models operate in non-stationary conditions and are repeatedly exposed to new, unseen data. Yet, most current novelty detection approaches have been developed exclusively for static, offline use. They scale poorly under more realistic, continual learning regimes in which data distribution shifts occur. To address this critical gap, this paper proposes incDFM (incremental Deep Feature Modeling), a self-supervised continual novelty detector. The method builds a statistical model over the space of intermediate features produced by a deep network, and utilizes feature reconstruction errors as uncertainty scores to guide the detection of novel samples. Most importantly, incDFM estimates the statistical model incrementally (via several iterations within a task), instead of a single-shot. Each time it selects only the most confident novel samples which will then guide subsequent recruitment incrementally. For a certain task where the ML model encounters a mixture of old and novel data, the detector flags novel samples to incorporate them to old knowledge. Then the detector is updated with the flagged novel samples, in preparation for a next task. To quantify and benchmark performance, we adapted multiple datasets for continual learning: CIFAR-10, CIFAR-100, SVHN, iNaturalist, and the 8-dataset. Our experiments show that incDFM achieves state of the art continual novelty detection performance. Furthermore, when examined in the greater context of continual learning for classification, our method is successful in minimizing catastrophic forgetting and error propagation.



Paperid:1051
Authors:Yunsheng Pang; Qiuhong Ke; Hossein Rahmani; James Bailey; Jun Liu
Title: IGFormer: Interaction Graph Transformer for Skeleton-Based Human Interaction Recognition
Abstract:
Human interaction recognition is very important in many applications. One crucial cue in recognizing an interaction is the interactive body parts. In this work, we propose a novel Interaction Graph Transformer (IGFormer) network for skeleton-based interaction recognition via modeling the interactive body parts as graphs. More specifically, the proposed IGFormer constructs interaction graphs according to the semantic and distance correlations between the interactive body parts, and enhances the representation of each person by aggregating the information of the interactive body parts based on the learned graphs. Furthermore, we propose a Semantic Partition Module to transform each human skeleton sequence into a Body-Part-Time sequence to better capture the spatial and temporal information of the skeleton sequence for learning the graphs. Extensive experiments on three benchmark datasets demonstrate that our model outperforms the state-of-the-art with a significant margin.



Paperid:1052
Authors:Apostolos Modas; Rahul Rade; Guillermo Ortiz-Jiménez; Seyed-Mohsen Moosavi-Dezfooli; Pascal Frossard
Title: PRIME: A Few Primitives Can Boost Robustness to Common Corruptions
Abstract:
Despite their impressive performance on image classification tasks, deep networks have a hard time generalizing to unforeseen corruptions of their data. To fix this vulnerability, prior works have built complex data augmentation strategies, combining multiple methods to enrich the training data. However, introducing intricate design choices or heuristics makes it hard to understand which elements of these methods are indeed crucial for improving robustness. In this work, we take a step back and follow a principled approach to achieve robustness to common corruptions. We propose PRIME, a general data augmentation scheme that relies on simple yet rich families of max-entropy image transformations. PRIME outperforms the prior art in terms of corruption robustness, while its simplicity and plug-and-play nature enable combination with other methods to further boost their robustness. We analyze PRIME to shed light on the importance of the mixing strategy on synthesizing corrupted images, and to reveal the robustness-accuracy trade-offs arising in the context of common corruptions. Finally, we show that the computational efficiency of our method allows it to be easily used in both on-line and off-line data augmentation schemes.



Paperid:1053
Authors:Takumi Kobayashi
Title: Rotation Regularization without Rotation
Abstract:
In various visual classification tasks, we enjoy significant performance improvement by deep convolutional neural networks (CNNs). To further boost performance, it is effective to regularize feature representation learning of CNNs such as by considering margin to improve feature distribution across classes. In this paper, we propose a regularization method based on random rotation of feature vectors. Random rotation is derived from cone representation to describe angular margin of a sample. While it induces geometric regularization to randomly rotate vectors by means of rotation matrices, we theoretically formulate the regularization in a statistical form which excludes costly geometric rotation as well as effectively imposes rotation-based regularization on classification in training CNNs. In the experiments on classification tasks, the method is thoroughly evaluated from various aspects, while producing favorable performance compared to the other regularization methods. Codes are available at https://github.com/tk1980/StatRot.



Paperid:1054
Authors:Wonwoo Cho; Jaegul Choo
Title: Towards Accurate Open-Set Recognition via Background-Class Regularization
Abstract:
In open-set recognition (OSR), classifiers should be able to reject unknown-class samples while maintaining high closed-set classification accuracy. To effectively solve the OSR problem, previous studies attempted to limit latent feature space and reject data located outside the limited space via offline analyses, e.g., distance-based feature analyses, or complicated network architectures. To conduct OSR via a simple inference process (without offline analyses) in standard classifier architectures, we use distance-based classifiers instead of conventional Softmax classifiers. Afterwards, we design a background-class regularization strategy, which uses background-class data as surrogates of unknown-class ones during training phase. Specifically, we formulate a novel regularization loss suitable for distance-based classifiers, which reserves sufficiently large class-wise latent feature spaces for known classes and forces background-class samples to be located far away from the limited spaces. Through our extensive experiments, we show that the proposed method provides robust OSR results, while maintaining high closed-set classification accuracy.



Paperid:1055
Authors:Xianhang Li; Huiyu Wang; Chen Wei; Jieru Mei; Alan Yuille; Yuyin Zhou; Cihang Xie
Title: In Defense of Image Pre-training for Spatiotemporal Recognition
Abstract:
Image pre-training, the current de-facto paradigm for a wide range of visual tasks, is generally less favored in the field of video recognition. By contrast, a common strategy is to directly train with spatiotemporal convolutional neural networks (CNNs) from scratch. Nonetheless, interestingly, by taking a closer look at these from-scratch learned CNNs, we note there exist certain 3D kernels that exhibit much stronger appearance modeling ability than others, arguably suggesting appearance information is already well disentangled in learning. Inspired by this observation, we hypothesize that the key to effectively leveraging image pre-training lies in the decomposition of learning spatial and temporal features, and revisiting image pre-training as the appearance prior to initializing 3D kernels. In addition, we propose Spatial-Temporal Separable (STS) convolution, which explicitly splits the feature channels into spatial and temporal groups, to further enable a more thorough decomposition of spatiotemporal features for fine-tuning 3D CNNs. Our experiments show that simply replacing 3D convolution with STS notably improves a wide range of 3D CNNs without increasing parameters and computation on both Kinetics-400 and Something-Something V2. Moreover, this new training pipeline consistently achieves better results on video recognition with significant speedup. For instance, we achieve +0.6% top-1 of Slowfast on Kinetics-400 over the strong 256-epoch 128-GPU baseline while fine-tuning for only 50 epochs with 4 GPUs. The code and models are available at https://github.com/UCSC-VLAA/Image-Pretraining-for-Video"



Paperid:1056
Authors:Grigorios G. Chrysos; Markos Georgopoulos; Jiankang Deng; Jean Kossaifi; Yannis Panagakis; Anima Anandkumar
Title: Augmenting Deep Classifiers with Polynomial Neural Networks
Abstract:
Deep neural networks have been the driving force behind the success in classification tasks, e.g., object and audio recognition. Impressive results and generalization have been achieved by a variety of recently proposed architectures, the majority of which are seemingly disconnected. In this work, we cast the study of deep classifiers under a unifying framework. In particular, we express state-of-the-art architectures (e.g., residual and non-local networks) in the form of different degree polynomials of the input. Our framework provides insights on the inductive biases of each model and enables natural extensions building upon their polynomial nature. The efficacy of the proposed models is evaluated on standard image and audio classification benchmarks. The expressivity of the proposed models is highlighted both in terms of increased model performance as well as model compression. Lastly, the extensions allowed by this taxonomy showcase benefits in the presence of limited data and long-tailed data distributions. We expect this taxonomy to provide links between existing domain-specific architectures.



Paperid:1057
Authors:Seong Min Kye; Kwanghee Choi; Joonyoung Yi; Buru Chang
Title: Learning with Noisy Labels by Efficient Transition Matrix Estimation to Combat Label Miscorrection
Abstract:
Recent studies on learning with noisy labels have shown remarkable performance by exploiting a small clean dataset. In particular, model agnostic meta-learning-based label correction methods further improve performance by correcting noisy labels on the fly. However, there is no safeguard on the label miscorrection, resulting in unavoidable performance degradation. Moreover, every training step requires at least three back-propagations, significantly slowing down the training speed. To mitigate these issues, we propose a robust and efficient method, FasTEN, which learns a label transition matrix on the fly. Employing the transition matrix makes the classifier skeptical about all the corrected samples, which alleviates the miscorrection issue. We also introduce a two-head architecture to efficiently estimate the label transition matrix every iteration within a single back-propagation, so that the estimated matrix closely follows the shifting noise distribution induced by label correction. Extensive experiments demonstrate that our FasTEN shows the best performance in training efficiency while having comparable or better accuracy than existing methods, especially achieving state-of-the-art performance in a real-world noisy dataset, Clothing1M.



Paperid:1058
Authors:Julien Pourcel; Ngoc-Son Vu; Robert M. French
Title: Online Task-Free Continual Learning with Dynamic Sparse Distributed Memory
Abstract:
This paper addresses the very challenging problem of online task-free continual learning in which a sequence of new tasks is learned from non-stationary data using each sample only once for training and without knowledge of task boundaries. We propose in this paper an efficient semi-distributed associative memory algorithm called Dynamic Sparse Distributed Memory (DSDM) where learning and evaluating can be carried out at any point of time. DSDM evolves dynamically and continually modeling the distribution of any non-stationary data streams. DSDM relies on locally distributed, but only partially overlapping clusters of representations to effectively eliminate catastrophic forgetting, while at the same time, maintaining the generalization capacities of distributed networks. In addition, a local density-based pruning technique is used to control the network’s memory footprint. DSDM significantly outperforms state-of-the-art continual learning methods on different image classification baselines, even in a low data regime. Code is publicly available: https://github.com/Julien-pour/Dynamic-Sparse-Distributed-Memory "



Paperid:1059
Authors:Linfeng Zhang; Xin Chen; Junbo Zhang; Runpei Dong; Kaisheng Ma
Title: Contrastive Deep Supervision
Abstract:
The success of deep learning is usually accompanied by the growth in neural network depth. However, the traditional training method only supervises the neural network at its last layer and propagates the supervision layer-by-layer, which leads to hardship in optimizing the intermediate layers. Recently, deep supervision has been proposed to add auxiliary classifiers to the intermediate layers of deep neural networks. By optimizing these auxiliary classifiers with the supervised task loss, the supervision can be applied to the shallow layers directly. However, deep supervision conflicts with the well-known observation that the shallow layers learn low-level features instead of task-biased high-level semantic features. To address this issue, this paper proposes a novel training framework named Contrastive Deep Supervision, which supervises the intermediate layers with augmentation-based contrastive learning. Experimental results on nine popular datasets with eleven models demonstrate its effects on general image classification, fine-grained image classification and object detection in supervised learning, semi-supervised learning and knowledge distillation. Codes have been released in Github.



Paperid:1060
Authors:Quan Cui; Bingchen Zhao; Zhao-Min Chen; Borui Zhao; Renjie Song; Boyan Zhou; Jiajun Liang; Osamu Yoshie
Title: Discriminability-Transferability Trade-Off: An Information-Theoretic Perspective
Abstract:
This work simultaneously considers the discriminability and transferability properties of deep representations in the typical supervised learning task, i.e., image classification. By a comprehensive temporal analysis, we observe a trade-off between these two properties. The discriminability keeps increasing with the training progressing while the transferability intensely diminishes in the later training period. From the perspective of information-bottleneck theory, we reveal that the incompatibility between discriminability and transferability is attributed to the over-compression of input information. More importantly, we investigate why and how the InfoNCE loss can alleviate the over-compression, and further present a learning framework, named contrastive temporal coding (CTC), to counteract the over-compression and alleviate the incompatibility. Extensive experiments validate that CTC successfully mitigates the incompatibility, yielding discriminative and transferable representations. Noticeable improvements are achieved on the image classification task and challenging transfer learning tasks. We hope that this work will raise the significance of the transferability property in the conventional supervised learning setting.



Paperid:1061
Authors:Meng Cao; Tianyu Yang; Junwu Weng; Can Zhang; Jue Wang; Yuexian Zou
Title: LocVTP: Video-Text Pre-training for Temporal Localization
Abstract:
Video-Text Pre-training (VTP) aims to learn transferable representations for various downstream tasks from large-scale web videos. To date, almost all existing VTP methods are limited to retrieval-based downstream tasks, e.g., video retrieval, whereas their transfer potentials on localization-based tasks, e.g., temporal grounding, are underexplored. In this paper, we experimentally analyze and demonstrate the incompatibility of current VTP methods with localization tasks, and propose a novel Localization-oriented Video-Ttext Pre-training framework, dubbed as LocVTP. Specifically, we perform the fine-grained contrastive alignment as a complement to the coarse-grained one by a clip-word correspondence discovery scheme. To further enhance the temporal reasoning ability of the learned feature, we propose a context projection head and a temporal aware contrastive loss to perceive the contextual relationships. Extensive experiments on four downstream tasks across six datasets demonstrate that our LocVTP achieves state-of-the-art performance on both retrieval-based and localization-based tasks. Furthermore, we conduct comprehensive ablation studies and thorough analyses to explore the optimum model designs and training strategies.



Paperid:1062
Authors:Jiawei Ma; Guangxing Han; Shiyuan Huang; Yuncong Yang; Shih-Fu Chang
Title: Few-Shot End-to-End Object Detection via Constantly Concentrated Encoding across Heads
Abstract:
Few-shot object detection (FSOD) aims to detect objects of new classes and learn effective models without exhaustive annotation. The end-to-end detection framework has been proposed to generate sparse proposals and set a stack of detection heads to improve the performance. For each proposal, the predictions at lower heads are fed into deeper heads. However, the deeper head may not concentrate on the detected objects and then degrades, resulting in inefficient training and further limiting the performance gain in few-shot scenario. In this paper, we propose a few-shot adaptation strategy, Constantly Concentrated Encoding across heads (CoCo-RCNN), for the end-to-end detectors. For each class, we gather the encodings which detect on its object instances and then train them to be discriminative to avoid degraded prediction. In addition, we embed the class-relevant encodings to the learnable proposals to facilitate the adaptation at lower heads. Extensive experimental results show that our model brought clear gain on benchmarks. Detailed ablation studies are provided to justify the selection of each component.



Paperid:1063
Authors:Yannick Strümpler; Janis Postels; Ren Yang; Luc Van Gool; Federico Tombari
Title: Implicit Neural Representations for Image Compression
Abstract:
Implicit Neural Representations (INRs) gained attention as a novel and effective representation for various data types. Recently, prior work applied INRs to image compressing. Such compression algorithms are promising candidates as a general purpose approach for any coordinate-based data modality. However, in order to live up to this promise current INR-based compression algorithms need to improve their rate-distortion performance by a large margin. This work progresses on this problem. First, we propose meta learned initializations for INR-based compression which improves rate-distortion performance. As a side effect it also leads to faster convergence speed. Secondly, we introduce a simple yet highly effective change to the network architecture compared to prior work on INR-based compression. Namely, we combine SIREN networks with positional encodings which improves rate distortion performance. Our contributions to source compression with INRs vastly outperform prior work. We show that our INR-based compression algorithm, meta-learning combined with SIREN and positional encodings, outperforms JPEG2000 and Rate-Distortion Autoencoders on Kodak with 2x reduced dimensionality for the first time and closes the gap on full resolution images. To underline the generality of INR-based source compression, we further perform experiments on 3D shape compression where our method greatly outperforms Draco - a traditional compression algorithm.



Paperid:1064
Authors:Emre Aksan; Shugao Ma; Akin Caliskan; Stanislav Pidhorskyi; Alexander Richard; Shih-En Wei; Jason Saragih; Otmar Hilliges
Title: LiP-Flow: Learning Inference-Time Priors for Codec Avatars via Normalizing Flows in Latent Space
Abstract:
Neural face avatars that are trained from multi-view data captured in camera domes can produce photo-realistic 3D reconstructions. However, at inference time, they must be driven by limited inputs such as partial views recorded by headset-mounted cameras or a front-facing camera, and sparse facial landmarks. To mitigate this asymmetry, we introduce a prior model that is conditioned on the runtime inputs and tie this prior space to the 3D face model via a normalizing flow in the latent space. Our proposed model, LiP-Flow, consists of two encoders that learn representations from the rich training-time and impoverished inference-time observations. A normalizing flow bridges the two representation spaces and transforms latent samples from one domain to another, allowing us to define a latent likelihood objective. We trained our model end-to-end to maximize the similarity of both representation spaces and the reconstruction quality, making the 3D face model aware of the limited driving signals. We conduct extensive evaluations where the latent codes are optimized to reconstruct 3D avatars from partial or sparse observations. We show that our approach leads to an expressive and effective prior, capturing facial dynamics and subtle expressions better.



Paperid:1065
Authors:Qihang Zhang; Zhenghao Peng; Bolei Zhou
Title: Learning to Drive by Watching YouTube Videos: Action-Conditioned Contrastive Policy Pretraining
Abstract:
Deep visuomotor policy learning, which aims to map raw visual observation to action, achieves promising results in control tasks such as robotic manipulation and autonomous driving. However, it requires a huge number of online interactions with the training environment, which limits its real-world application. Compared to the popular unsupervised feature learning for visual recognition, feature pretraining for visuomotor control tasks is much less explored. In this work, we aim to pretrain policy representations for driving tasks by watching hours-long uncurated YouTube videos. Specifically, we train an inverse dynamic model with a small amount of labeled data and use it to predict action labels for all the YouTube video frames. A new contrastive policy pretraining method is then developed to learn action-conditioned features from the video frames with pseudo action labels. Experiments show that the resulting action-conditioned features obtain substantial improvements for the downstream reinforcement learning and imitation learning tasks, outperforming the weights pretrained from previous unsupervised learning methods and ImageNet pretrained weight. Code, model weights, and data are available at: https://metadriverse.github.io/ACO.



Paperid:1066
Authors:Jiachen Lu; Zheyuan Zhou; Xiatian Zhu; Hang Xu; Li Zhang
Title: Learning Ego 3D Representation As Ray Tracing
Abstract:
A self-driving perception model aims to extract 3D semantic representations from multiple cameras collectively into the bird’s-eye-view (BEV) coordinate frame of the ego car in order to ground downstream planner. Existing perception methods often rely on error-prone depth estimation of the whole scene or learning sparse virtual 3D representations without the target geometry structure, both of which remain limited in performance and/or capability. In this paper, we present a novel end-to-end architecture for ego 3D representation learning from an arbitrary number of unconstrained camera views. Inspired by the ray tracing principle, we design a polarized grid of “imaginary eyes"" as the learnable ego 3D representation and formulate the learning process with the adaptive attention mechanism in conjunction with the 3D-to-2D projection. Critically, this formulation allows extracting rich 3D representation from 2D images without any depth supervision, and with the built-in geometry structure consistent w.r.t. BEV. Despite its simplicity and versatility, extensive experiments on standard BEV visual tasks (e.g., camera-based 3D object detection and BEV segmentation) show that our model outperforms all state-of-the-art alternatives significantly, with an extra advantage in computational efficiency from multi-task learning.



Paperid:1067
Authors:Rui Qian; Shuangrui Ding; Xian Liu; Dahua Lin
Title: Static and Dynamic Concepts for Self-Supervised Video Representation Learning
Abstract:
In this paper, we propose a novel learning scheme for self-supervised video representation learning. Motivated by how humans understand videos, we propose to first learn general visual concepts then attend to discriminative local areas for video understanding. Specifically, we utilize static frame and frame difference to help decouple static and dynamic concepts, and respectively align the concept distributions in latent space. We add diversity and fidelity regularizations to guarantee that we learn a compact set of meaningful concepts. Then we employ a cross-attention mechanism to aggregate detailed local features of different concepts, and filter out redundant concepts with low activations to perform local concept contrast. Extensive experiments demonstrate that our method distills meaningful static and dynamic concepts to guide video understanding, and obtains state-of-the-art results on UCF-101, HMDB-51, and Diving-48.



Paperid:1068
Authors:Xin Dong; Sai Qian Zhang; Ang Li; H.T. Kung
Title: SphereFed: Hyperspherical Federated Learning
Abstract:
Federated Learning aims at training a global model from multiple decentralized devices (i.e. clients) without exchanging their private local data. A key challenge is the handling of non-i.i.d. (independent identically distributed) data across multiple clients that may induce disparities of their local features. We introduce the Hyperspherical Federated Learning (SphereFed) framework to address the non-i.i.d. issue by constraining learned representations of data points to be on a unit hypersphere shared by clients. Specifically, all clients learn their local representations by minimizing the loss with respect to a fixed classifier whose weights span the unit hypersphere. After federated training in improving the global model, this classifier is further calibrated with a closed-form solution by minimizing a mean squared loss. We show that the calibration solution can be computed efficiently and distributedly without direct access of local data. Extensive experiments indicate that our SphereFed approach is able to improve the accuracy of multiple existing federated learning algorithms by a considerable margin (up to 6% on challenging datasets) with enhanced computation and communication efficiency across datasets and model architectures.



Paperid:1069
Authors:Yuxiao Chen; Long Zhao; Jianbo Yuan; Yu Tian; Zhaoyang Xia; Shijie Geng; Ligong Han; Dimitris N. Metaxas
Title: Hierarchically Self-Supervised Transformer for Human Skeleton Representation Learning
Abstract:
Despite the success of fully-supervised human skeleton sequence modeling, utilizing self-supervised pre-training for skeleton sequence representation learning has been an active field because acquiring task-specific skeleton annotations at large scales is difficult. Recent studies focus on learning video-level temporal and discriminative information using contrastive learning, but overlook the hierarchical spatial-temporal nature of human skeletons. Different from such superficial supervision at the video level, we propose a self-supervised hierarchical pre-training scheme incorporated into a hierarchical Transformer-based skeleton sequence encoder (Hi-TRS), to explicitly capture spatial, short-term, and long-term temporal dependencies at frame, clip, and video levels, respectively. To evaluate the proposed self-supervised pre-training scheme with Hi-TRS, we conduct extensive experiments covering three skeleton-based downstream tasks including action recognition, action detection, and motion prediction. Under both supervised and semi-supervised evaluation protocols, our method achieves the state-of-the-art performance. Additionally, we demonstrate that the prior knowledge learned by our model in the pre-training stage has strong transfer capability for different downstream tasks.



Paperid:1070
Authors:Mingda Wang; Canqian Yang; Yi Xu
Title: Posterior Refinement on Metric Matrix Improves Generalization Bound in Metric Learning
Abstract:
Deep metric learning (DML) attempts to learn a representation model as well as a metric function with a limited generalization gap, so that the model trained on finite known data can achieve similitude performance on infinite unseen data. While considerable efforts have been made to bound the generalization gap by enhancing the model architecture and training protocol a priori in the training phase, none of them notice that a lightweight posterior refinement operation on the trained metric matrix can significantly improve the generalization ability. In this paper, we attempt to fill up this research gap and theoretically analyze the impact of the refined metric matrix property on the generalization gap. Based on our theory, two principles, which suggest a smaller trace or a smaller Frobenius norm of the refined metric matrix, are proposed as guidance for the posterior refinement operation. Experiments on three benchmark datasets verify the correctness of our principles and demonstrate that a pluggable posterior refinement operation is potential to significantly improve the performance of existing models with negligible extra computation burden.



Paperid:1071
Authors:Yajing Kong; Liu Liu; Zhen Wang; Dacheng Tao
Title: Balancing Stability and Plasticity through Advanced Null Space in Continual Learning
Abstract:
Continual learning is a learning paradigm that learns tasks sequentially with resources constraints, in which the key challenge is stability-plasticity dilemma, i.e., it is uneasy to simultaneously have the stability to prevent catastrophic forgetting of old tasks and the plasticity to learn new tasks well. In this paper, we propose a new continual learning approach, Advanced Null Space (AdNS), to balance the stability and plasticity without storing any old data of previous tasks. Specifically, to obtain better stability, AdNS makes use of low-rank approximation to obtain a novel null space and projects the gradient onto the null space to prevent the interference on the past tasks. To control the generation of the null space, we introduce a non-uniform constraint strength to further reduce forgetting. Furthermore, we present a simple but effective method, intra-task distillation, to improve the performance of the current task. Finally, we theoretically find that null space plays a key role in plasticity and stability, respectively. Experimental results show that the proposed method can achieve better performance compared to state-of-the-art continual learning approaches.



Paperid:1072
Authors:Yuting Gao; Jia-Xin Zhuang; Shaohui Lin; Hao Cheng; Xing Sun; Ke Li; Chunhua Shen
Title: DisCo: Remedying Self-Supervised Learning on Lightweight Models with Distilled Contrastive Learning
Abstract:
While Self-Supervised Learning (SSL) has received widespread attention from the community, recent researches argue that its performance often suffers a cliff fall when the model size decreases. Since current SSL methods mainly rely on contrastive learning to train the network, we propose a simple yet effective method termed Distilled Contrastive Learning (DisCo) to ease this issue. Specifically, we find that the final inherent embedding of the mainstream SSL methods contains the most important information and propose to distill the final embedding to maximally transmit a teacher’s knowledge to a lightweight model by constraining the last embedding of the student to be consistent with that of the teacher. In addition, we find that there exists a phenomenon termed Distilling BottleNeck and propose to enlarge the embedding dimension to alleviate this problem. Since the MLP only exists during the SSL phase, our method does not introduce any extra parameters to lightweight models for the downstream task deployment. Experimental results demonstrate that our method surpasses the state-of-the-art on many lightweight models by a large margin. Particularly, when ResNet-101/ResNet-50 is used respectively as a teacher to teach EfficientNet-B0, the linear result of EfficientNet-B0 on ImageNet is improved by 22.1% and 19.7%, respectively, which is very close to ResNet-101/ResNet-50 with much fewer parameters. Code is available at https://github.com/Yuting-Gao/DisCo-pytorch.



Paperid:1073
Authors:Liyuan Wang; Xingxing Zhang; Qian Li; Jun Zhu; Yi Zhong
Title: CoSCL: Cooperation of Small Continual Learners Is Stronger than a Big One
Abstract:
Continual learning requires incremental compatibility with a sequence of tasks. However, the design of model architecture remains an open question: In general, learning all tasks with a shared set of parameters suffers from severe interference between tasks; while learning each task with a dedicated parameter subspace is limited by scalability. In this work, we theoretically analyze the generalization errors for learning plasticity and memory stability in continual learning, which can be uniformly upper-bounded by (1) discrepancy between task distributions, (2) flatness of loss landscape and (3) cover of parameter space. Then, inspired by the robust biological learning system that processes sequential experiences with multiple parallel compartments, we propose Cooperation of Small Continual Learners (CoSCL) as a general strategy for continual learning. Specifically, we present an architecture with a fixed number of narrower sub-networks to learn all incremental tasks in parallel, which can naturally reduce the two errors through improving the three components of the upper bound. To strengthen this advantage, we encourage to cooperate these sub-networks by penalizing the difference of predictions made by their feature representations. With a fixed parameter budget, CoSCL can improve a variety of representative continual learning approaches by a large margin (e.g., up to 10.64% on CIFAR-100-SC, 9.33% on CIFAR-100-RS, 11.45% on CUB-200-2011 and 6.72% on Tiny-ImageNet) and achieve the new state-of-the-art performance. Our code is available at https://github.com/lywang3081/CoSCL.



Paperid:1074
Authors:Hao Huang; Cheng Chen; Yi Fang
Title: Manifold Adversarial Learning for Cross-Domain 3D Shape Representation
Abstract:
Deep neural networks (DNNs) for point clouds have achieved superior performance on a range of 3D vision tasks. However, generalization to out-of-distribution 3D point clouds remains challenging for DNNs. Due to the expensive cost or even infeasibility of annotating large-scale point clouds, it is critical but yet has not been fully explored to design methods to generalize DNN models to unseen domains of point clouds without any access to them during the training process. In this paper, we propose to learn 3D point cloud representation on a seen source domain and generalize to an unseen target domain via adversarial learning. Specifically, we unify several geometric transformations in a manifold-based framework under which distance between transformations is well-defined. Measured by the distance, adversarial samples are mined to form intermediate domains and retained in an adaptive replay-based memory. We further provide theoretical justification for the intermediate domains to reduce the generalization error of the DNN models. Experimental results on synthetic-to-real datasets illustrate that our method outperforms existing 3D deep learning models for domain generalization.



Paperid:1075
Authors:Yuanzheng Ci; Chen Lin; Lei Bai; Wanli Ouyang
Title: Fast-MoCo: Boost Momentum-Based Contrastive Learning with Combinatorial Patches
Abstract:
Contrastive-based self-supervised learning methods achieved great success in recent years. However, self-supervision requires extremely long training epochs (e.g., 800 epochs for MoCo v3) to achieve promising results, which is unacceptable for the general academic community and hinders the development of this topic. This work revisits the momentum-based contrastive learning frameworks and identifies the inefficiency in which two augmented views generate only one positive pair. We propose Fast-MoCo - a novel framework that utilizes combinatorial patches to construct multiple positive pairs from two augmented views, which provides abundant supervision signals that bring significant acceleration with neglectable extra computational cost. Fast-MoCo trained with 100 epochs achieves 73.5% linear evaluation accuracy, similar to MoCo v3 (ResNet-50 backbone) trained with 800 epochs. Extra training (200 epochs) further improves the result to 75.1%, which is on par with state-of-the-art methods. Experiments on several downstream tasks also confirm the effectiveness of Fast-MoCo.



Paperid:1076
Authors:Boyan Jiang; Xinlin Ren; Mingsong Dou; Xiangyang Xue; Yanwei Fu; Yinda Zhang
Title: LoRD: Local 4D Implicit Representation for High-Fidelity Dynamic Human Modeling
Abstract:
Recent progress in 4D implicit representation focuses on globally controlling the shape and motion with low dimensional latent vectors, which is prone to missing surface details and accumulating tracking error. While many deep local representations have shown promising results for 3D shape modeling, their 4D counterpart does not exist yet. In this paper, we fill this blank by proposing a novel Local 4D implicit Representation for Dynamic clothed human, named LoRD, which has the merits of both 4D human modeling and local representation, and enables high-fidelity reconstruction with detailed surface deformations, such as clothing wrinkles. Particularly, our key insight is to encourage the network to learn the latent codes of local part-level representation, capable of explaining the local geometry and temporal deformations. To make the inference at test-time, we first estimate the inner body skeleton motion to track local parts at each time step, and then optimize the latent codes for each part via auto-decoding based on different types of observed data. Extensive experiments demonstrate that the proposed method has strong capability for representing 4D human, and outperforms state-of-the-art methods on practical applications, including 4D reconstruction from sparse points, non-rigid depth fusion, both qualitatively and quantitatively.



Paperid:1077
Authors:Xingjian Zhen; Zihang Meng; Rudrasis Chakraborty; Vikas Singh
Title: On the Versatile Uses of Partial Distance Correlation in Deep Learning
Abstract:
Comparing the functional behavior of neural network models, whether it is a single network over time or two (or more networks) during or post-training, is an essential step in understanding what they are learning (and what they are not), and for identifying strategies for regularization or efficiency improvements. Despite recent progress, e.g., comparing vision transformers to CNNs, systematic comparison of function, especially across different networks, remains difficult and is often carried out layer by layer. Approaches such as canonical correlation analysis (CCA) are applicable in principle, but have been sparingly used so far. In this paper, we revisit a (less widely known) from statistics, called distance correlation (and its partial variant), designed to evaluate correlation between feature spaces of different dimensions. We describe the steps necessary to carry out its deployment for large scale models -- this opens the door to a surprising array of applications ranging from conditioning one deep model w.r.t. another, learning disentangled representations as well as optimizing diverse models that would directly be more robust to adversarial attacks. Our experiments suggest a versatile regularizer (or constraint) with many advantages, which avoids some of the common difficulties one faces in such analyses.



Paperid:1078
Authors:Lujun Li
Title: Self-Regulated Feature Learning via Teacher-Free Feature Distillation
Abstract:
Knowledge distillation conditioned on intermediate feature representations always leads to significant performance improvements. Conventional feature distillation framework demands extra selecting/training budgets of teachers and complex transformations to align the features between teacher-student models. To address the problem, we analyze teacher roles in feature distillation and have an intriguing observation: additional teacher architectures are not always necessary. Then we propose Tf-FD, a simple yet effective Teacher-free Feature Distillation framework, reusing channel-wise and layer-wise meaningful features within the student to provide teacher-like knowledge without an additional model. In particular, our framework is subdivided into intra-layer and inter-layer distillation. The intra-layer Tf-FD performs feature salience ranking and transfers the knowledge from salient feature to redundant feature within the same layer. For inter-layer Tf-FD, we deal with distilling high-level semantic knowledge embedded in the deeper layer representations to guide the training of shallow layers. Benefiting from the small gap between these self-features, Tf-FD simply needs to optimize extra feature mimicking losses without complex transformations. Furthermore, we provide insightful discussions to shed light on Tf-FD from feature regularization perspectives. Our experiments conducted on classification and object detection tasks demonstrate that our technique achieves state-of-the-art results on different models with fast training speeds. Code is available at https://lilujunai.github.io/Teacher-free-Distillation/.



Paperid:1079
Authors:Mingfu Liang; Jiahuan Zhou; Wei Wei; Ying Wu
Title: Balancing between Forgetting and Acquisition in Incremental Subpopulation Learning
Abstract:
The subpopulation shifting challenge, known as some subpopulations of a category that are not seen during training, severely limits the classification performance of the state-of-the-art convolutional neural networks. Thus, to mitigate this practical issue, we explore incremental subpopulation learning (ISL) to adapt the original model via incrementally learning the unseen subpopulations without retaining the seen population data. However, striking a great balance between subpopulation learning and seen population forgetting is the main challenge in ISL but is not well studied by existing approaches. These incremental learners simply use a pre-defined and fixed hyperparameter to balance the learning objective and forgetting regularization, but their learning is usually biased towards either side in the long run. In this paper, we propose a novel two-stage learning scheme to explicitly disentangle the acquisition and forgetting for achieving a better balance between subpopulation learning and seen population forgetting: in the first “gain-acquisition” stage, we progressively learn a new classifier based on the margin-enforce loss, which enforces the hard samples and population to have a larger weight for classifier updating and avoid uniformly updating all the population; in the second “counter-forgetting” stage, we search for the proper combination of the new and old classifiers by optimizing a novel objective based on proxies of forgetting and acquisition. We benchmark the representative and state-of-the-art non-exemplar-based incremental learning methods on a large-scale subpopulation shifting dataset for the first time. Under almost all the challenging ISL protocols, we significantly outperform other methods by a large margin, demonstrating our superiority to alleviate the subpopulation shifting problem.



Paperid:1080
Authors:Xulin Li; Yan Lu; Bin Liu; Yating Liu; Guojun Yin; Qi Chu; Jinyang Huang; Feng Zhu; Rui Zhao; Nenghai Yu
Title: Counterfactual Intervention Feature Transfer for Visible-Infrared Person Re-identification
Abstract:
Graph-based models have achieved great success in person re-identification tasks recently, which compute the graph topology structure (affinities) among different people first and then pass the information across them to achieve stronger features. But we find existing graph-based methods in the visible-infrared person re-identification task (VI-ReID) suffer from bad generalization because of two issues: 1) train-test modality balance gap, which is a property of VI-ReID task. The number of two modalities data are balanced in the training stage, but extremely unbalanced in inference, causing the low generalization of graph-based VI-ReID methods. 2) sub-optimal topology structure caused by the end-to-end learning manner to the graph module. We analyze that the well-trained input features weaken the learning of graph topology, making it not generalized enough during the inference process. In this paper, we propose a Counterfactual Intervention Feature Transfer (CIFT) method to tackle these problems. Specifically, a Homogeneous and Heterogeneous Feature Transfer (H2FT) is designed to reduce the train-test modality balance gap by two independent types of well-designed graph modules and an unbalanced scenario simulation. Besides, a Counterfactual Relation Intervention (CRI) is proposed to utilize the counterfactual intervention and causal effect tools to highlight the role of topology structure in the whole training process, which makes the graph topology structure more reliable. Extensive experiments on standard VI-ReID benchmarks demonstrate that CIFT outperforms the state-of-the-art methods under various settings.



Paperid:1081
Authors:Lizhao Liu; Shangxin Huang; Zhuangwei Zhuang; Ran Yang; Mingkui Tan; Yaowei Wang
Title: DAS: Densely-Anchored Sampling for Deep Metric Learning
Abstract:
Deep Metric Learning (DML) serves to learn an embedding function to project semantically similar data into nearby embedding space and plays a vital role in many applications, such as image retrieval and face recognition. However, the performance of DML methods often highly depends on sampling methods to choose effective data from the embedding space in the training. In practice, the embeddings in the embedding space are obtained by some deep models, where the embedding space is often with barren area due to the absence of training points, resulting in so called “missing embedding” issue. This issue may impair the sample quality, which leads to degenerated DML performance. In this work, we investigate how to alleviate the “missing embedding” issue to improve the sampling quality and achieve effective DML. To this end, we propose a Densely-Anchored Sampling (DAS) scheme that considers the embedding with corresponding data point as “anchor” and exploits the anchor’s nearby embedding space to densely produce embeddings without data points. Specifically, we propose to exploit the embedding space around single anchor with Discriminative Feature Scaling (DFS) and multiple anchors with Memorized Transformation Shifting (MTS). In this way, by combing the embeddings with and without data points, we are able to provide more embeddings to facilitate the sampling process thus boosting the performance of DML. Our method is effortlessly integrated into existing DML frameworks and improves them without bells and whistles. Extensive experiments on three benchmark datasets demonstrate the superiority of our method.



Paperid:1082
Authors:Yuhang Zhang; Chengrui Wang; Xu Ling; Weihong Deng
Title: Learn from All: Erasing Attention Consistency for Noisy Label Facial Expression Recognition
Abstract:
Noisy label Facial Expression Recognition (FER) is more challenging than traditional noisy label classification tasks due to the inter-class similarity and the annotation ambiguity. Recent works mainly tackle this problem by filtering out large-loss samples. In this paper, we explore dealing with noisy labels from a new feature-learning perspective. We find that FER models remember noisy samples by focusing on a part of the features that can be considered related to the noisy labels instead of learning from the whole features that lead to the latent truth. Inspired by that, we propose a novel Erasing Attention Consistency (EAC) method to suppress the noisy samples during the training process automatically. Specifically, we first utilize the flip semantic consistency of facial images to design an imbalanced framework. We then randomly erase input images and use flip attention consistency to prevent the model from focusing on a part of the features. EAC significantly outperforms state-of-the-art noisy label FER methods and generalizes well to other tasks with a large number of classes like CIFAR100 and Tiny-ImageNet. The code is available at https://github.com/zyh-uaiaaaa/Erasing-Attention-Consistency.



Paperid:1083
Authors:Michael Kirchhof; Karsten Roth; Zeynep Akata; Enkelejda Kasneci
Title: A Non-Isotropic Probabilistic Take On Proxy-Based Deep Metric Learning
Abstract:
Proxy-based Deep Metric Learning (DML) learns deep metric spaces by embedding images and class representatives (proxies) close to one another during training, as commonly measured by the angle between them. However, this disregards the embedding norm, which can carry additional beneficial context such as class- or image-intrinsic uncertainty. In addition, proxy-based DML struggles to learn class-internal structures. To address both issues at once, we introduce non-isotropic probabilistic proxy-based DML. We model images as directional von Mises-Fisher (vMF) distributions on the hypersphere and motivate a non-isotropic von Mises-Fisher (nivMF) model for class proxies. This allows proxies to better represent complex class-specific variances in the embedding space. To cast these probabilistic models into proxy-to-image metrics, we further develop and investigate multiple distribution-to-point and distribution-to-distribution metrics. Each framework choice is motivated through a set of ablational studies, which showcase beneficial properties of our probabilistic approach to proxy-based DML. This comprises uncertainty-awareness, better behaved gradients during training and overall improved generalization performance. The latter is especially reflected in the competitive performance on the standard DML benchmarks. We find our proposed approach to compare favourably, suggesting that existing proxy-based DML can significantly benefit from a more probabilistic treatment.



Paperid:1084
Authors:Jihao Liu; Boxiao Liu; Hang Zhou; Hongsheng Li; Yu Liu
Title: TokenMix: Rethinking Image Mixing for Data Augmentation in Vision Transformers
Abstract:
CutMix is a popular augmentation technique commonly used for training modern convolutional and transformer vision networks. It was originally designed to encourage Convolution Neural Networks (CNNs) to focus more on an image’s global context instead of local information, which greatly improves the performance of CNNs. However, we found it to have limited benefits for transformer-based architectures that naturally have a global receptive field. In this paper, we propose a novel data augmentation technique TokenMix to improve the performance of vision transformers. TokenMix mixes two images at token level via partitioning the mixing region into multiple separated parts. Besides, we show that the mixed learning target in CutMix, a linear combination of a pair of the ground truth labels, might be inaccurate and sometimes counter-intuitive. To obtain a more suitable target, we propose to assign the target score according to the content-based neural activation maps of the two images from a pre-trained teacher model, which does not need to have high performance. With plenty of experiments on various vision transformer architectures, we show that our proposed TokenMix helps vision transformers focus on the foreground area to infer the classes and enhances their robustness to occlusion, with consistent performance gains. Notably, we improve DeiT-T/S/B with +1% ImageNet top-1 accuracy. Besides, TokenMix enjoys longer training, which achieves 81.2% top-1 accuracy on ImageNet with DeiT-S trained for 400 epochs.



Paperid:1085
Authors:Teng Xi; Yifan Sun; Deli Yu; Bi Li; Nan Peng; Gang Zhang; Xinyu Zhang; Zhigang Wang; Jinwen Chen; Jian Wang; Lufei Liu; Haocheng Feng; Junyu Han; Jingtuo Liu; Errui Ding; Jingdong Wang
Title: UFO: Unified Feature Optimization
Abstract:
This paper proposes a novel Unified Feature Optimization (UFO) paradigm for training and deploying deep models under real-world and large-scale scenarios, which requires a collection of multiple AI functions. UFO aims to benefit each single task with a large-scale pretraining on all tasks. Compared with existing foundation models, UFO has two different points of emphasis, i.e., relatively smaller model size and NO adaptation cost: 1) UFO squeezes a wide range of tasks into a moderate-sized unified model in a multi-task learning manner and further trims the model size when transferred to down-stream tasks. 2) UFO does not emphasize transfer to novel tasks. Instead, it aims to make the trimmed model dedicated for one or more already-seen task. To this end, it directly selects partial modules in the unified model, and thus requires completely NO adaptation cost. With these two characteristics, UFO provides great convenience for flexible deployment, while maintaining the benefits of large-scale pretraining. A key merit of UFO is that the trimming process not only reduces the model size and inference consumption, but also even improves the accuracy on certain tasks. Specifically, UFO considers the multi-task training and brings two-fold impact on the unified model: some closely related tasks have mutual benefits, while some tasks have conflicts against each other. UFO manages to reduce the conflicts and to preserve the mutual benefits through a novel Network Architecture Search (NAS) method. Experiments on a wide range of deep representation learning tasks (i.e., face recognition, person re-identification, vehicle re-identification and product retrieval) show that the model trimmed from UFO achieves higher accuracy than its single-task-trained counterpart and yet has smaller model size, validating the concept of UFO. Besides, UFO also supported the release of 17 billion parameters computer vision (CV) foundation model which is the largest CV model in the industry.



Paperid:1086
Authors:Ziyang Chen; David F. Fouhey; Andrew Owens
Title: Sound Localization by Self-Supervised Time Delay Estimation
Abstract:
Sounds reach one microphone in a stereo pair sooner than the other, resulting in an interaural time delay that conveys their directions. Estimating a sound’s time delay requires finding correspondences between the signals recorded by each microphone. We propose to learn these correspondences through self-supervision, drawing on recent techniques from visual tracking. We adapt the contrastive random walk of Jabri et al. to learn a cycle-consistent representation from unlabeled stereo sounds, resulting in a model that performs on par with supervised methods on ""in the wild"" internet recordings. We also propose a multimodal contrastive learning model that solves a visually-guided localization task: estimating the time delay for a particular person in a multi-speaker mixture, given a visual representation of their face.



Paperid:1087
Authors:Yinan He; Gengshi Huang; Siyu Chen; Jianing Teng; Kun Wang; Zhenfei Yin; Lu Sheng; Ziwei Liu; Yu Qiao; Jing Shao
Title: X-Learner: Learning Cross Sources and Tasks for Universal Visual Representation
Abstract:
In computer vision, pre-training models based on large-scale supervised learning have been proven effective over the past few years. However, existing works mostly focus on learning from the individual tasks with the single data source e.g., ImageNet for classification or COCO for detection). This restricted form limits their generalizability and usability due to the lack of vast semantic information from various tasks and data sources. Here, we demonstrate that jointly learning from heterogeneous tasks and multiple data sources contributes to universal visual representation, leading to better transferring results of various downstream tasks. Thus, learning how to bridge the gaps among different tasks and data sources is the key, but it still remains an open question. In this work, we propose a representation learning framework called X-Learner, which learns the universal feature of multiple vision tasks supervised by various sources, with expansion and squeeze stage:1) Expansion Stage: X-Learner learns the task-specific feature to alleviate task interference and enrich the representation by reconciliation layer. 2) Squeeze Stage: X-Learner condenses the model to a reasonable size and learns the universal and generalizable representation for various tasks transferring. Extensive experiments demonstrate that X-Learner achieves strong performance on different tasks without extra annotations, modalities and computational costs compared to existing representation learning methods. Notably, a single X-Learner model shows remarkable gains of 3.0%, 3.3% and 1.8% over current pre-trained models on 12 downstream datasets for classification, object detection and semantic segmentation.



Paperid:1088
Authors:Norman Mu; Alexander Kirillov; David Wagner; Saining Xie
Title: SLIP: Self-Supervision Meets Language-Image Pre-training
Abstract:
Recent work has shown that self-supervised pre-training leads to improvements over supervised learning on challenging visual recognition tasks. CLIP, an exciting new approach to learning with language supervision, demonstrates promising performance on a wide variety of benchmarks. In this work, we explore whether self-supervised learning can aid in the use of language supervision for visual representation learning with Vision Transformers. We introduce SLIP, a multi-task learning framework for combining self-supervised learning and CLIP pre-training. After pre-training, we thoroughly evaluate representation quality and compare performance to both CLIP and self-supervised learning under three distinct settings: zero-shot transfer, linear classification, and end-to-end finetuning. Across ImageNet and a battery of additional datasets, we find that SLIP improves accuracy by a large margin. We validate our results further with experiments on different model sizes, training schedules, and pre-training datasets. Our findings show that SLIP enjoys the best of both worlds: better performance than self-supervision (+8.1% linear accuracy) and language supervision (+5.2% zero-shot accuracy). Our code is available at: github.com/facebookresearch/SLIP.



Paperid:1089
Authors:Jianing Qian; Anastasios Panagopoulos; Dinesh Jayaraman
Title: Discovering Deformable Keypoint Pyramids
Abstract:
The locations of objects and their associated landmark keypoints can serve as versatile and semantically meaningful image representations. In natural scenes, these keypoints are often hierarchically grouped into sets corresponding to coherently moving objects and their moveable and deformable parts. Motivated by this observation, we propose Keypoint Pyramids, an approach to exploit this property for discovering keypoints without explicit supervision. Keypoint Pyramids discovers multi-level keypoint hierarchies satisfying three desiderata: comprehensiveness of the overall keypoint representation, coarse-to-fine informativeness of individual hierarchy levels, and parent-child associations of keypoints across levels. On human pose and tabletop multi-object scenes, our experimental results show that Keypoint Pyramids jointly discovers object keypoints and their natural hierarchical groupings, with finer levels adding detail to coarser levels to more comprehensively represent the visual scene. Further, we show qualitatively and quantitatively that keypoints discovered by Keypoint Pyramids using its hierarchical prior bind more consistently, and are more predictive of manually annotated semantic keypoints, compared to prior flat keypoint discovery approaches.



Paperid:1090
Authors:Fabian Mentzer; Eirikur Agustsson; Johannes Ballé; David Minnen; Nick Johnston; George Toderici
Title: Neural Video Compression Using GANs for Detail Synthesis and Propagation
Abstract:
We present the first neural video compression method based on generative adversarial networks (GANs). Our approach significantly outperforms previous neural and non-neural video compression methods in a user study, setting a new state-of-the-art in visual quality for neural methods. We show that the GAN loss is crucial to obtain this high visual quality. Two components make the GAN loss effective: we i) synthesize detail by conditioning the generator on a latent extracted from the warped previous reconstruction to then ii) propagate this detail with high-quality flow. We find that user studies are required to compare methods, i.e., none of our quantitative metrics were able to predict all studies. We present the network design choices in detail, and ablate them with user studies.



Paperid:1091
Authors:Jonathan Kahana; Yedid Hoshen
Title: A Contrastive Objective for Learning Disentangled Representations
Abstract:
Learning representations of images that are invariant to sensitive or unwanted attributes is important for many tasks including bias removal and cross domain retrieval. Here, our objective is to learn representations that are invariant to the domain (sensitive attribute) for which labels are provided, while being informative over all other image attributes, which are unlabeled. We present a new approach, proposing a new domain-wise contrastive objective for ensuring invariant representations. This objective crucially restricts negative image pairs to be drawn from the same domain, which enforces domain invariance whereas the standard contrastive objective does not. This domain-wise objective is insufficient on its own as it suffers from shortcut solutions resulting in feature suppression. We overcome this issue by a combination of a reconstruction constraint, image augmentations and initialization with pre-trained weights. Our analysis shows that the choice of augmentations is important, and that a misguided choice of augmentations can harm the invariance and informativeness objectives. In an extensive evaluation, our method convincingly outperforms the state-of-the-art in terms of representation invariance, representation informativeness, and training speed. Furthermore, we find that in some cases our method can achieve excellent results even without the reconstruction constraint, leading to a much faster and resource efficient training.



Paperid:1092
Authors:John Seon Keun Yi; Minseok Seo; Jongchan Park; Dong-Geol Choi
Title: PT4AL: Using Self-Supervised Pretext Tasks for Active Learning
Abstract:
Labeling a large set of data is expensive. Active learning aims to tackle this problem by asking to annotate only the most informative data from the unlabeled set. We propose a novel active learning approach that utilizes self-supervised pretext tasks and a unique data sampler to select data that are both difficult and representative. We discover that the loss of a simple self-supervised pretext task, such as rotation prediction, is closely correlated to the downstream task loss. Before the active learning iterations, the pretext task learner is trained on the unlabeled set, and the unlabeled data are sorted and split into batches by their pretext task losses. In each active learning iteration, the main task model is used to sample the most uncertain data in a batch to be annotated. We evaluate our method on various image classification and segmentation benchmarks and achieve compelling performances on CIFAR10, Caltech-101, ImageNet, and Cityscapes. We further show that our method performs well on imbalanced datasets, and can be an effective solution to the cold-start problem where active learning performance is affected by the randomly sampled initial labeled set.



Paperid:1093
Authors:Haokui Zhang; Wenze Hu; Xiaoyu Wang
Title: ParC-Net: Position Aware Circular Convolution with Merits from ConvNets and Transformer
Abstract:
Recently, vision transformers started to show impressive results which outperform large convolution based models significantly. However, in the area of small models for mobile or resource constrained devices, ConvNet still has its own advantages in both performance and model complexity. We propose ParC-Net, a pure ConvNet based backbone model that further strengthens these advantages by fusing the merits of vision transformers into ConvNets. Specifically, we propose position aware circular convolution (ParC), a light-weight convolution op which boasts a global receptive field while producing location sensitive features as in local convolutions. We combine the ParCs and squeeze-exictation ops to form a meta-former like model block, which further has the attention mechanism like transformers. The aforementioned block can be used in plug-and-play manner to replace relevant blocks in ConvNets or transformers. Experiment results show that the proposed ParC-Net achieves better performance than popular light-weight ConvNets and vision transformer based models in common vision tasks and datasets, while having fewer parameters and faster inference speed. For classification on ImageNet-1k, ParC-Net achieves 78.6% top-1 accuracy with about 5.0 million parameters, saving 11% parameters and 13% computational cost but gaining 0.2% higher accuracy and 23% faster inference speed (on ARM based Rockchip RK3288) compared with MobileViT, and uses only 0.5× parameters but gaining 2.7% accuracy compared with DeIT. On MS-COCO object detection and PASCAL VOC segmentation tasks, ParC-Net also shows better performance. Source code is available at https://github.com/hkzhang91/ParC-Net"



Paperid:1094
Authors:Zifeng Wang; Zizhao Zhang; Sayna Ebrahimi; Ruoxi Sun; Han Zhang; Chen-Yu Lee; Xiaoqi Ren; Guolong Su; Vincent Perot; Jennifer Dy; Tomas Pfister
Title: DualPrompt: Complementary Prompting for Rehearsal-Free Continual Learning
Abstract:
Continual learning aims at enabling a single model to learn a sequence of tasks without catastrophic forgetting. Top-performing methods usually require a rehearsal buffer to store past pristine examples for experience replay, which, however, limits their practical values due to privacy and memory constraints. In this work, we present a simple yet effective framework, DualPrompt, which learns a tiny set of parameters, called prompt, to properly instruct a pre-trained model to learn tasks arriving sequentially, without buffering past examples. DualPrompt presents a novel approach to attach complementary prompts to the pre-trained backbone, and then formulates the objective as learning task-invariant and task-specific ""instructions"". With extensive experimental validation, DualPrompt consistently sets state-of-the-art performance under the challenging class-incremental setting. In particular, DualPrompt outperforms recent advanced continual learning methods with relatively large buffer size. We also introduce a more challenging benchmark, Split ImageNet-R, to help generalize rehearsal-free continual learning research. Source code is available at https://github.com/google-research/l2p.



Paperid:1095
Authors:Shixiang Tang; Feng Zhu; Lei Bai; Rui Zhao; Chenyu Wang; Wanli Ouyang
Title: Unifying Visual Contrastive Learning for Object Recognition from a Graph Perspective
Abstract:
Recent contrastive based unsupervised object recognition methods leverage a Siamese architecture, which has two branches composed of a backbone, a projector layer, and an optional predictor layer in each branch. To learn the parameters of the backbone, existing methods have a similar projector layer design, while the major difference among them lies in the predictor layer. In this paper, we propose to \underline{Uni}fy existing unsupervised \underline{V}isual \underline{C}ontrastive \underline{L}earning methods by using a GCN layer as the predictor layer (UniVCL), which deserves two merits to unsupervised learning in object recognition. First, by treating different designs of predictors in the existing methods as its special cases, our fair and comprehensive experiments reveal the critical importance of neighborhood aggregation in the GCN predictor. Second, by viewing the predictor from the graph perspective, we can bridge the vision self-supervised learning with the graph representation learning area, which facilitates us to introduce the augmentations from the graph representation learning to unsupervised object recognition and further improves the unsupervised object recognition accuracy. Extensive experiments on linear evaluation and the semi-supervised learning tasks demonstrate the effectiveness of UniVCL and the introduced graph augmentations. Code will be released upon acceptance.



Paperid:1096
Authors:Chun-Hsiao Yeh; Cheng-Yao Hong; Yen-Chi Hsu; Tyng-Luh Liu; Yubei Chen; Yann LeCun
Title: Decoupled Contrastive Learning
Abstract:
Contrastive learning (CL) is one of the most successful paradigms for self-supervised learning (SSL). In a principled way, it considers two augmented views of the same image as positive to be pulled closer, and all other images negative to be pushed further apart. However, behind the impressive success of CL-based techniques, their formulation often relies on heavy-computation settings, including large sample batches, extensive training epochs, etc. We are thus motivated to tackle these issues and establish a simple, efficient, yet competitive baseline of contrastive learning. Specifically, we identify, from theoretical and empirical studies, a noticeable negative-positive-coupling (NPC) effect in the widely used InfoNCE loss, leading to unsuitable learning efficiency concerning the batch size. By removing the NPC effect, we propose decoupled contrastive learning (DCL) loss, which removes the positive term from the denominator and significantly improves the learning efficiency. DCL achieves competitive performance with less sensitivity to sub-optimal hyperparameters, requiring neither large batches in SimCLR, momentum encoding in MoCo, or large epochs. We demonstrate with various benchmarks while manifesting robustness as much less sensitive to suboptimal hyperparameters. Notably, SimCLR with DCL achieves 68.2% ImageNet-1K top-1 accuracy using batch size 256 within 200 epochs pre-training, outperforming its SimCLR baseline by 6.4%. Further, DCL can be combined with the SOTA contrastive learning method, NNCLR, to achieve 72.3% ImageNet-1K top-1 accuracy with 512 batch size in 400 epochs, which represents a new SOTA in contrastive learning. We believe DCL provides a valuable baseline for future contrastive SSL studies.



Paperid:1097
Authors:Philip Müller; Georgios Kaissis; Congyu Zou; Daniel Rueckert
Title: Joint Learning of Localized Representations from Medical Images and Reports
Abstract:
Contrastive learning has proven effective for pre-training image models on unlabeled data with promising results for tasks such as medical image classification. Using paired text (like radiological reports) during pre-training improves the results even further. Still, most existing methods target image classification downstream tasks and may not be optimal for localized tasks like semantic segmentation or object detection. We therefore propose Localized representation learning from Vision and Text (LoVT), a text-supervised pre-training method that explicitly targets localized medical imaging tasks. Our method combines instance-level image-report contrastive learning with local contrastive learning on image region and report sentence representations. We evaluate LoVT and commonly used pre-training methods on an evaluation framework of 18 localized tasks on chest X-rays from five public datasets. LoVT performs best on 10 of the 18 studied tasks making it the preferred method of choice for localized tasks.



Paperid:1098
Authors:Senthil Purushwalkam; Pedro Morgado; Abhinav Gupta
Title: The Challenges of Continuous Self-Supervised Learning
Abstract:
Self-supervised learning (SSL) aims to eliminate one of the major bottlenecks in representation learning - the need for human annotations. As a result, SSL holds the promise to learn representations from data in-the-wild, i.e., without the need for finite and static datasets. Instead, true SSL algorithms should be able to exploit the continuous stream of data being generated on the internet or by agents exploring their environments. But do traditional self-supervised learning approaches work in this setup? In this work, we investigate this question by conducting experiments on the continuous self-supervised learning problem. While learning in the wild, we expect to see a continuous (infinite) non-IID data stream that follows a non-stationary distribution of visual concepts. The goal is to learn a representation that can be robust, adaptive yet not forgetful of concepts seen in the past. We show that a direct application of current methods to such continuous setup is 1) inefficient both computationally and in the amount of data required, 2) leads to inferior representations due to temporal correlations (non-IID data) in some sources of streaming data and 3) exhibits signs of catastrophic forgetting when trained on sources with non-stationary data distributions. We propose the use of replay buffers as an approach to alleviate the issues of inefficiency and temporal correlations. We further propose a novel method to enhance the replay buffer by maintaining the least redundant samples. Minimum redundancy (MinRed) buffers allow us to learn effective representations even in the most challenging streaming scenarios composed of sequential visual data obtained from a single embodied agent, and alleviates the problem of catastrophic forgetting when learning from data with non-stationary semantic distributions.



Paperid:1099
Authors:Zhixin Ling; Zhen Xing; Jian Zhou; Xiangdong Zhou
Title: Conditional Stroke Recovery for Fine-Grained Sketch-Based Image Retrieval
Abstract:
The key to Fine-Grained Sketch Based Image Retrieval (FG-SBIR) is to establish fine-grained correspondence between sketches and images. Since sketches only consist of abstract strokes, stroke recognition ability plays an important role in FG-SBIR. However, existing works usually ignore the unique feature of sketches and treat images and sketches equally. Targeting at this problem, we propose Conditional Stroke Recovery (CSR) to enhance stroke recognition ability for FG-SBIR, in which we introduce an auxiliary task that requires the network recover the strokes using the paired image as condition. In this way, the network learns better to match the strokes with corresponding image elements. To complete the auxiliary task, we propose an unsupervised stroke disorder algorithm, which does well in stroke extraction and sketch augmentation. In addition, we figure out two weaknesses of the common triplet loss and propose double-anchor InfoNCE loss to reduce cosine distances between sketch-image pairs. Comprehensive experiments using various backbones are conducted on four datasets (i.e., QMUL-Shoe, QMUL-Chair, QMUL-ShoeV2, and Sketchy). In terms of acc@1, our method outperforms previous works by a great margin.



Paperid:1100
Authors:Xuanyu Yi; Kaihua Tang; Xian-Sheng Hua; Joo-Hwee Lim; Hanwang Zhang
Title: Identifying Hard Noise in Long-Tailed Sample Distribution
Abstract:
Conventional de-noising methods rely on the assumption that the noisy samples are independent and identically distributed, so the resultant classifier, though disturbed by noise, can still easily identify the noises as outliers. However, the assumption is unrealistic in large-scale data that is inevitably long-tailed. Such imbalance makes a classifier less discriminative for the tail classes, whose previously “easy” noises are now turned into “hard” ones--they are almost as outliers as the tail samples. We introduce this new challenge as Noisy Long-Tailed Classification (NLT). Not surprisingly, we find that most de-noising methods fail to identify the hard noises, resulting in significant performance drop on the three proposed NLT benchmarks: ImageNet-NLT, Animal10-NLT, and Food101-NLT. To this end, we design an iterative noisy learning framework called Hard-to-Easy (H2E). Our bootstrapping philosophy is to first learn a classifier as noise identifier invariant to the class and context distributional change, reducing “hard” noises to “easy” ones, whose removal further improves the invariance. Experimental results show that our H2E outperforms state-of-the-art de-noising methods and their ablations on long-tailed settings while maintaining a stable performance on balanced ones. Codes are in Appendix.



Paperid:1101
Authors:Shixiang Tang; Feng Zhu; Lei Bai; Rui Zhao; Wanli Ouyang
Title: Relative Contrastive Loss for Unsupervised Representation Learning
Abstract:
Defining positive and negative samples is critical for learning visual variations of the semantic classes in an unsupervised manner. Previous methods either construct positive sample pairs as different data augmentations on the same image (i.e., single-instance-positive) or estimate a class prototype by clustering (i.e., prototype-positive), both ignoring the relative nature of positive/negative concepts in the real world. Motivated by the ability of humans in recognizing relatively positive/negative samples, we propose the Relative Contrastive Loss (RCL) to learn feature representation from relatively positive/negative pairs, which not only learns more real world semantic variations than the single-instance-positive methods but also respects positive-negative relativeness compared with absolute prototype-positive methods. The proposed RCL improves the linear evaluation for MoCo v3 by \textbf{+2.0\%} on ImageNet. Code will be released publicly upon acceptance.



Paperid:1102
Authors:Yang Jiao; Ning Xie; Yan Gao; Chien-chih Wang; Yi Sun
Title: Fine-Grained Fashion Representation Learning by Online Deep Clustering
Abstract:
Fashion designs are rich in visual details associated with various visual attributes at both global and local levels. As a result, effective modeling and analyzing fashion requires fine-grained representations for individual attributes. In this work, we present a deep learning based online clustering method to jointly learn fine-grained fashion representations for all attributes at both instance and cluster level, where the attribute-specific cluster centers are online estimated. Based on the similarity between fine-grained representations and cluster centers, attribute-specific embedding spaces are further segmented into class-specific embedding spaces for fine-grained fashion retrieval. To better regulate the learning process, we design a three-stage learning scheme, to progressively incorporate different supervisions at both instance and cluster level, from both original and augmented data, and with ground-truth and pseudo labels. Experiments on FashionAI and DARN datasets in retrieval task demonstrated the efficacy of our method compared with competing baselines.



Paperid:1103
Authors:Eric Yeats; Frank Liu; David Womble; Hai Li
Title: NashAE: Disentangling Representations through Adversarial Covariance Minimization
Abstract:
We present a self-supervised method to disentangle factors of variation in high-dimensional data that does not rely on prior knowledge of the underlying variation profile (e.g., no assumptions on the number or distribution of the individual variables to be extracted). In this method which we call NashAE, high-dimensional feature disentanglement is accomplished in the low-dimensional latent space of a standard autoencoder (AE) by promoting the discrepancy between each encoding element and information of the element recovered from all other encoding elements. Disentanglement is promoted efficiently by framing this as a minmax game between the AE and an ensemble of regression networks which each provide an estimate of an element conditioned on an observation of all other elements. We quantitatively compare our approach with leading disentanglement methods using existing disentanglement metrics. Furthermore, we show that NashAE has increased reliability and increased capacity to capture salient data characteristics in the learned latent representation.



Paperid:1104
Authors:Xuan Son Nguyen
Title: A Gyrovector Space Approach for Symmetric Positive Semi-Definite Matrix Learning
Abstract:
Representation learning with Symmetric Positive Semi-definite (SPSD) matrices has proven effective in many machine learning problems. Recently, some SPSD neural networks have been proposed and shown promising performance. While these works share a common idea of generalizing some basic operations in deep neural networks (DNNs) to the SPSD manifold setting, their proposed generalizations are usually designed in an ad hoc manner. In this work, we make an attempt to propose a principled framework for building such generalizations. Our method is motivated by the success of hyperbolic neural networks (HNNs) that have demonstrated impressive performance in a variety of applications. At the heart of HNNs is the theory of gyrovector spaces that provides a powerful tool for studying hyperbolic geometry. Here we consider connecting the theory of gyrovector spaces and the Riemannian geometry of SPSD manifolds. We first propose a method to define basic operations, i.e., binary operation and scalar multiplication in gyrovector spaces of (full-rank) Symmetric Positive Definite (SPD) matrices. We then extend these operations to the low-rank SPSD manifold setting. Finally, we present an approach for building SPSD neural networks. Experimentl evaluations on three benchmarks for human activity recognition demonstrate the efficacy of our proposed framework.



Paperid:1105
Authors:Haoxuan You; Luowei Zhou; Bin Xiao; Noel Codella; Yu Cheng; Ruochen Xu; Shih-Fu Chang; Lu Yuan
Title: Learning Visual Representation from Modality-Shared Contrastive Language-Image Pre-training
Abstract:
Large-scale multi-modal contrastive pre-training has demonstrated great utility to learn transferable features for a range of downstream tasks by mapping multiple modalities into a shared embedding space. Typically, this has employed separate encoders for each modality. However, recent work suggests that transformers can support learning across multiple modalities and allow knowledge sharing. Inspired by this, we investigate a variety of Modality-Shared Contrastive Language-Image Pre-training (MS-CLIP) frameworks. More specifically, we question how many parameters of a transformer model can be shared across modalities during contrastive pre-training, and rigorously examine architectural design choices that position the proportion of parameters shared along a spectrum. In studied conditions, we observe that a mostly unified encoder for vision and language signals outperforms all other variations that separate more parameters. Additionally, we find that light-weight modality-specific parallel modules further improve performance. Experimental results show that the proposed MS-CLIP approach outperforms vanilla CLIP by up to 13% relative in zero-shot ImageNet classification (pre-trained on YFCC-100M), while simultaneously supporting a reduction of parameters. In addition, our approach outperforms vanilla CLIP by 1.6 points in linear probing on a collection of 24 downstream vision tasks. Furthermore, we discover that sharing parameters leads to semantic concepts from different modalities being encoded more closely in the embedding space, facilitating the transferring of common semantic structure (e.g., attention patterns) from language to vision. Code is available at https://github.com/Hxyou/MSCLIP.



Paperid:1106
Authors:Artem Moskalev; Ivan Sosnovik; Volker Fischer; Arnold Smeulders
Title: Contrasting Quadratic Assignments for Set-Based Representation Learning
Abstract:
The standard approach to contrastive learning is to maximize the agreement between different views of the data. The views are ordered in pairs, such that they are either positive, encoding different views of the same object, or negative, corresponding to views of different objects. The supervisory signal comes from maximizing the total similarity over positive pairs, while the negative pairs are needed to avoid collapse. In this work, we note that the approach of considering individual pairs cannot account for both intra-set and inter-set similarities when the sets are formed from the views of the data. It thus limits the information content of the supervisory signal available to train representations. We propose to go beyond contrasting individual pairs of objects by focusing on contrasting objects as sets. For this, we use combinatorial quadratic assignment theory designed to evaluate set and graph similarities and derive set-contrastive objective as a regularizer for contrastive learning methods. We conduct experiments and demonstrate that our method improves learned representations for the tasks of metric learning and self-supervised classification.



Paperid:1107
Authors:Arjun Ashok; K J Joseph; Vineeth N Balasubramanian
Title: Class-Incremental Learning with Cross-Space Clustering and Controlled Transfer
Abstract:
In class-incremental learning, the model is expected to learn new classes continually while maintaining knowledge on previous classes. The challenge here lies in preserving the model’s ability to effectively represent prior classes in the feature space, while adapting it to represent incoming new classes. We propose two distillation-based objectives for class incremental learning that leverage the structure of the feature space to maintain accuracy on previous classes, as well as enable learning the new classes. In our first objective, termed cross-space clustering (CSC), we propose to use the feature space structure of the previous model to characterize directions of optimization that maximally preserve the class - directions that all instances of a specific class should collectively optimize towards, and those directions that they should collectively optimize away from. Apart from minimizing forgetting, such a class-level constraint indirectly encourages the model to reliably cluster all instances of a class in the current feature space, and further gives rise to a sense of “herd-immunity”, allowing all samples of a class to jointly combat the model from forgetting the class. Our second objective termed controlled transfer (CT) tackles incremental learning from an important and understudied perspective of inter-class transfer. CT explicitly approximates and conditions the current model on the semantic similarities between incrementally arriving classes and prior classes. This allows the model to learn the incoming classes in such a way that it maximizes positive forward transfer from similar prior classes, thus increasing plasticity, and minimizes negative backward transfer on dissimilar prior classes, whereby strengthening stability. We perform extensive experiments on two benchmark datasets, adding our method (CSCCT) on top of three prominent class-incremental learning methods. We observe consistent performance improvement on a variety of experimental settings.



Paperid:1108
Authors:Olivier J. Hénaff; Skanda Koppula; Evan Shelhamer; Daniel Zoran; Andrew Jaegle; Andrew Zisserman; João Carreira; Relja Arandjelović
Title: Object Discovery and Representation Networks
Abstract:
The promise of self-supervised learning (SSL) is to leverage large amounts of unlabeled data to solve complex tasks. While there has been excellent progress with simple, image-level learning, recent methods have shown the advantage of including knowledge of image structure. However, by introducing hand-crafted image segmentations to define regions of interest, or specialized augmentation strategies, these methods sacrifice the simplicity and generality that makes SSL so powerful. Instead, we propose a self-supervised learning paradigm that discovers this image structure by itself. Our method, Odin, couples object discovery and representation networks to discover meaningful image segmentations without any supervision. The resulting learning paradigm is simpler, less brittle, and more general, and achieves state-of-the-art transfer learning results for object detection and instance segmentation on COCO, and semantic segmentation on PASCAL and Cityscapes, while strongly surpassing supervised pre-training for video segmentation on DAVIS.



Paperid:1109
Authors:Jianqiao Zheng; Sameera Ramasinghe; Xueqian Li; Simon Lucey
Title: Trading Positional Complexity vs Deepness in Coordinate Networks
Abstract:
It is well noted that coordinate-based MLPs benefit---in terms of preserving high-frequency information---through the encoding of coordinate positions as an array of Fourier features. Hitherto, the rationale for the effectiveness of these \emph{positional encodings} has been mainly studied through a Fourier lens. In this paper, we strive to broaden this understanding by showing that alternative non-Fourier embedding functions can indeed be used for positional encoding. Moreover, we show that their performance is entirely determined by a trade-off between the stable rank of the embedded matrix and the distance preservation between embedded coordinates. We further establish that the now ubiquitous Fourier feature mapping of position is a special case that fulfills these conditions. Consequently, we present a more general theory to analyze positional encoding in terms of shifted basis functions. In addition, we argue that employing a more complex positional encoding---that scales exponentially with the number of modes---requires only a linear (rather than deep) coordinate function to achieve comparable performance. Counter-intuitively, we demonstrate that trading positional embedding complexity for network deepness is orders of magnitude faster than current state-of-the-art; despite the additional embedding complexity. To this end, we develop the necessary theoretical formulae and empirically verify that our theoretical claims hold in practice.



Paperid:1110
Authors:Jian Zhang; Lei Qi; Yinghuan Shi; Yang Gao
Title: MVDG: A Unified Multi-View Framework for Domain Generalization
Abstract:
Aiming to generalize the model trained in source domains to unseen target domains, domain generalization (DG) has attracted lots of attention recently. Since target domains can not be involved in training, overfitting to source domains is inevitable. As a popular regularization technique, the meta-learning training scheme has shown its ability to resist overfitting. However, in the training stage, current meta-learning-based methods utilize only one task along a single optimization trajectory, which might produce biased and noisy optimization direction. Beyond the training stage, overfitting could also cause unstable prediction in the test stage. In this paper, we propose a novel multi-view DG framework to effectively reduce the overfitting in both the training and test stage. Specifically, in the training stage, we develop a multi-view regularized meta-learning algorithm that employs multiple optimization trajectories to produce a suitable optimization direction for model updating. We also theoretically show the generalization bound could be reduced by increasing the number of tasks in each trajectory. In the test stage, to alleviate unstable prediction, we utilize multiple augmented images to yield a multi-view prediction, which significantly promotes model reliability. Extensive experiments on three benchmark datasets validate that our method can find a flat minimum to enhance generalization and outperform several state-of-the-art approaches.



Paperid:1111
Authors:Jingkang Yang; Yi Zhe Ang; Zujin Guo; Kaiyang Zhou; Wayne Zhang; Ziwei Liu
Title: Panoptic Scene Graph Generation
Abstract:
Existing research addresses scene graph generation (SGG), a critical technology to scene understanding in images, from the detection perspective, i.e., objects are detected using bounding boxes followed by prediction of their pairwise relationships. We argue that such a paradigm would cause several problems that impede the progress of the field. For instance, bounding box-based labels in current datasets usually contain redundant information like hairs and miss some background information that is crucial to the understanding of context. In this work, we introduce panoptic scene graph generation (PSG task), a new problem that requires the model to generate more comprehensive scene graph representations based on panoptic segmentations rather than rigid bounding boxes. A high-quality PSG dataset, which contains 51k well-annotated images from COCO and Visual Genome, is created for the community to keep track of the progress. For benchmarking, we build three two-stage models, which are modified from current state-of-the-arts in SGG, and another one-stage model called PSGTR, which is based on the efficient Transformer-based detector, i.e., DETR. We further propose an approach called PSGFormer, which achieves significant improvements with two novel extensions over PSGTR: 1) separate modeling of objects and relations in the form of queries in two Transformer decoders, and 2) a prompting-like interaction mechanism. In the end, we share insights on open challenges and future directions.



Paperid:1112
Authors:Qianyi Wu; Xian Liu; Yuedong Chen; Kejie Li; Chuanxia Zheng; Jianfei Cai; Jianmin Zheng
Title: Object-Compositional Neural Implicit Surfaces
Abstract:
The neural implicit representation has shown its effectiveness in novel view synthesis and high-quality 3D reconstruction from multi-view images. However, most approaches focus on holistic scene representation yet ignore individual objects inside it, thus limiting potential downstream applications. In order to learn object-compositional representation, a few works incorporate the 2D semantic map as a cue in training to grasp the difference between objects. But they neglect the strong connections between object geometry and instance semantic information, which leads to inaccurate modeling of individual instance. This paper proposes a novel framework, ObjectSDF, to build an object-compositional neural implicit representation with high fidelity in 3D reconstruction and object representation. Observing the ambiguity of conventional volume rendering pipelines, we model the scene by combining the Signed Distance Functions (SDF) of individual object to exert explicit surface constraint. The key in distinguishing different instances is to revisit the strong association between an individual object’s SDF and semantic label. Particularly, we convert the semantic information to a function of object SDF and develop a unified and compact representation for scene and objects. Experimental results show the superiority of ObjectSDF framework in representing both the holistic object-compositional scene and the individual instances. Code can be found at https://qianyiwu.github.io/objectsdf/"



Paperid:1113
Authors:Zhiqiang Yan; Kun Wang; Xiang Li; Zhenyu Zhang; Jun Li; Jian Yang
Title: RigNet: Repetitive Image Guided Network for Depth Completion
Abstract:
Depth completion deals with the problem of recovering dense depth maps from sparse ones, where color images are often used to facilitate this task. Recent approaches mainly focus on image guided learning frameworks to predict dense depth. However, blurry guidance in the image and unclear structure in the depth still impede the performance of the image guided frameworks. To tackle these problems, we explore a repetitive design in our image guided network to gradually and sufficiently recover depth values. Specifically, the repetition is embodied in both the image guidance branch and depth generation branch. In the former branch, we design a repetitive hourglass network to extract discriminative image features of complex environments, which can provide powerful contextual instruction for depth prediction. In the latter branch, we introduce a repetitive guidance module based on dynamic convolution, in which an efficient convolution factorization is proposed to simultaneously reduce its complexity and progressively model high-frequency structures. Extensive experiments show that our method achieves superior or competitive results on KITTI benchmark and NYUv2 dataset.



Paperid:1114
Authors:Hao Lu; Wenze Liu; Hongtao Fu; Zhiguo Cao
Title: FADE: Fusing the Assets of Decoder and Encoder for Task-Agnostic Upsampling
Abstract:
We consider the problem of task-agnostic feature upsampling in dense prediction where an upsampling operator is required to facilitate both region-sensitive tasks like semantic segmentation and detail-sensitive tasks such as image matting. Existing upsampling operators often can work well in either type of the tasks, but not both. In this work, we present FADE, a novel, plug-and-play, and task-agnostic upsampling operator. FADE benefits from three design choices: i) considering encoder and decoder features jointly in upsampling kernel generation; ii) an efficient semi-shift convolutional operator that enables granular control over how each feature point contributes to upsampling kernels; iii) a decoder-dependent gating mechanism for enhanced detail delineation. We first study the upsampling properties of FADE on toy data and then evaluate it on large-scale semantic segmentation and image matting. In particular, FADE reveals its effectiveness and task-agnostic characteristic by consistently outperforming recent dynamic upsampling operators in different tasks. It also generalizes well across convolutional and transformer architectures with little computational overhead. Our work additionally provides thoughtful insights on what makes for task-agnostic upsampling. Code is available at: http://lnkiy.in/fade_in"



Paperid:1115
Authors:Zeyu Hu; Xuyang Bai; Runze Zhang; Xin Wang; Guangyuan Sun; Hongbo Fu; Chiew-Lan Tai
Title: LiDAL: Inter-Frame Uncertainty Based Active Learning for 3D LiDAR Semantic Segmentation
Abstract:
We propose LiDAL, a novel active learning method for 3D LiDAR semantic segmentation by exploiting inter-frame uncertainty among LiDAR frames. Our core idea is that a well-trained model should generate robust results irrespective of viewpoints for scene scanning and thus the inconsistencies in model predictions across frames provide a very reliable measure of uncertainty for active sample selection. To implement this uncertainty measure, we introduce new inter-frame divergence and entropy formulations, which serve as the metrics for active selection. Moreover, we demonstrate additional performance gains by predicting and incorporating pseudo-labels, which are also selected using the proposed inter-frame uncertainty measure. Experimental results validate the effectiveness of LiDAL: we achieve 95% of the performance of fully supervised learning with less than 5% of annotations on the SemanticKITTI and nuScenes datasets, outperforming state-of-the-art active learning methods.



Paperid:1116
Authors:Youming Deng; Yansheng Li; Yongjun Zhang; Xiang Xiang; Jian Wang; Jingdong Chen; Jiayi Ma
Title: Hierarchical Memory Learning for Fine-Grained Scene Graph Generation
Abstract:
Regarding Scene Graph Generation (SGG), coarse and fine predicates mix in the dataset due to the crowd-sourced labeling, and the long-tail problem is also pronounced. Given this tricky situation, many existing SGG methods treat the predicates equally and learn the model under the supervision of mixed-granularity predicates in one stage, leading to relatively coarse predictions. In order to alleviate the impact of the suboptimum mixed-granularity annotation and long-tail effect problems, this paper proposes a novel Hierarchical Memory Learning (HML) framework to learn the model from simple to complex, which is similar to the human beings’ hierarchical memory learning process. After the autonomous partition of coarse and fine predicates, the model is first trained on the coarse predicates and then learns the fine predicates. In order to realize this hierarchical learning pattern, this paper, for the first time, formulates the HML framework using the new Concept Reconstruction (CR) and Model Reconstruction (MR) constraints. It is worth noticing that the HML framework can be taken as one general optimization strategy to improve various SGG models, and significant improvement can be achieved on the SGG benchmark.



Paperid:1117
Authors:Runyu Ding; Jihan Yang; Li Jiang; Xiaojuan Qi
Title: DODA: Data-Oriented Sim-to-Real Domain Adaptation for 3D Semantic Segmentation
Abstract:
Deep learning approaches achieve prominent success in 3D semantic segmentation. However, collecting densely annotated real-world 3D datasets is extremely time-consuming and expensive. Training models on synthetic data and generalizing on real-world scenarios becomes an appealing alternative, but unfortunately suffers from notorious domain shifts. In this work, we propose a Data-Oriented Domain Adaptation (DODA) framework to mitigate pattern and context gaps caused by different sensing mechanisms and layout placements across domains. Our DODA encompasses virtual scan simulation to imitate real-world point cloud patterns and tail-aware cuboid mixing to alleviate the interior context gap with a cuboid-based intermediate domain. The first unsupervised sim-to-real adaptation benchmark on 3D indoor semantic segmentation is also built on 3D-FRONT, ScanNet and S3DIS along with 8 popular Unsupervised Domain Adaptation (UDA) methods. Our DODA surpasses existing UDA approaches by over 13% on both 3DFRONT -> ScanNet and 3D-FRONT -> S3DIS. Code is available at https://github.com/CVMI-Lab/DODA.



Paperid:1118
Authors:Xiaogang Xu; Hengshuang Zhao; Vibhav Vineet; Ser-Nam Lim; Antonio Torralba
Title: MTFormer: Multi-task Learning via Transformer and Cross-Task Reasoning
Abstract:
In this paper, we explore the advantages of utilizing transformer structures for addressing multi-task learning (MTL). Specifically, we demonstrate that models with transformer structures are more appropriate for MTL than convolutional neural networks (CNNs), and we propose a novel transformer-based architecture named MTFormer for MTL. In the framework, multiple tasks share the same transformer encoder and transformer decoder, and lightweight branches are introduced to harvest task-specific outputs, which increases the MTL performance and reduces the time-space complexity. Furthermore, information from different task domains can benefit each other, and we conduct cross-task reasoning. We propose a cross-task attention mechanism for further boosting the MTL results. The cross-task attention mechanism brings little parameters and computations while introducing extra performance improvements. Besides, we design a self-supervised cross-task contrastive learning algorithm for further boosting the MTL performance. Extensive experiments are conducted on two multi-task learning datasets, on which MTFormer achieves state-of-the-art results with limited network parameters and computations. It also demonstrates significant superiorities for few-shot learning and zero-shot learning.



Paperid:1119
Authors:Runfa Li; Truong Nguyen
Title: MonoPLFlowNet: Permutohedral Lattice FlowNet for Real-Scale 3D Scene Flow Estimation with Monocular Images
Abstract:
Real-scale scene flow estimation has become increasingly important for 3D computer vision. Some works successfully estimate real-scale 3D scene flow with LiDAR. However, these ubiquitous and expensive sensors are still unlikely to be equipped widely for real application. Other works use monocular images to estimate scene flow, but their scene flow estimations are normalized with scale ambiguity, where additional depth or point cloud ground truth are required to recover the real scale. Even though they perform well in 2D, these works do not provide accurate and reliable 3D estimates. We present a deep learning architecture on permutohedral lattice - MonoPLFlowNet. Different from all previous works, our MonoPLFlowNet is the first work where only two consecutive monocular images are used as input, while both depth and 3D scene flow are estimated in real scale. Our real-scale scene flow estimation outperforms all state-of-the-art monocular-image based works recovered to real scale by ground truth, and is comparable to LiDAR approaches. As a by-product, our real-scale depth estimation also outperforms other state-of-the-art works.



Paperid:1120
Authors:Mutian Xu; Pei Chen; Haolin Liu; Xiaoguang Han
Title: TO-Scene: A Large-Scale Dataset for Understanding 3D Tabletop Scenes
Abstract:
Many basic indoor activities such as eating or writing are always conducted upon different tabletops (e.g., coffee tables, writing desks). It is indispensable to understanding tabletop scenes in 3D indoor scene parsing applications. Unfortunately, it is hard to meet this demand by directly deploying data-driven algorithms, since 3D tabletop scenes are rarely available in current datasets. To remedy this defect, we introduce TO-Scene, a large-scale dataset focusing on tabletop scenes, which contains 20,740 scenes with three variants. To acquire the data, we design an efficient and scalable framework, where a crowdsourcing UI is developed to transfer CAD objects from ModelNet and ShapeNet onto tables from ScanNet, then the output tabletop scenes are simulated into real scans and annotated automatically. Further, a tabletop-aware learning strategy is proposed for better perceiving the small-sized tabletop instances. Notably, we also provide a real scanned test set TO-Real to verify the practical value of TO-Scene. Experiments show that the algorithms trained on TO-Scene indeed work on the realistic test data, and our proposed tabletop-aware learning strategy greatly improves the state-of-the-art results on both 3D semantic segmentation and object detection tasks. Dataset and code are available at https://github.com/GAP-LAB-CUHK-SZ/TO-Scene.



Paperid:1121
Authors:Xinyi Wu; Zhenyao Wu; Jin Wan; Lili Ju; Song Wang
Title: Is It Necessary to Transfer Temporal Knowledge for Domain Adaptive Video Semantic Segmentation?
Abstract:
Video semantic segmentation is a fundamental and important task in computer vision, and it usually requires large-scale labeled data for training deep neural network models. To avoid laborious manual labeling, domain adaptive video segmentation approaches were recently introduced by transferring the knowledge from the source domain of self-labeled simulated videos to the target domain of unlabeled real-world videos. However, it leads to an interesting question -- while video-to-video adaptation is a natural idea, does the source data require to be videos? In this paper, we argue that it is not necessary to transfer temporal knowledge since the temporal continuity of video segmentation in the target domain can be estimated and enforced without reference to videos in the source domain. This motivates a new framework of Image-to-Video Domain Adaptive Semantic Segmentation (I2VDA), where the source domain is a set of images without temporal information. Under this setting, we bridge the domain gap via adversarial training based on only the spatial knowledge, and develop a novel temporal augmentation strategy, through which the temporal consistency in the target domain is well-exploited and learned. In addition, we introduce a new training scheme by leveraging a proxy network to produce pseudo-labels on-the-fly, which is very effective to improve the stability of adversarial training. Experimental results on two synthetic-to-real scenarios show that the proposed I2VDA method can achieve even better performance on video semantic segmentation than existing state-of-the-art video-to-video domain adaption approaches.



Paperid:1122
Authors:Li Xu; Haoxuan Qu; Jason Kuen; Jiuxiang Gu; Jun Liu
Title: Meta Spatio-Temporal Debiasing for Video Scene Graph Generation
Abstract:
Video scene graph generation (VidSGG) aims to parse the video content into scene graphs, which involves modeling the spatio-temporal contextual information in the video. However, due to the long-tailed training data in datasets, the generalization performance of existing VidSGG models can be affected by the spatio-temporal conditional bias problem. In this work, from the perspective of meta-learning, we propose a novel Meta Video Scene Graph Generation (MVSGG) framework to address such a bias problem. Specifically, to handle various types of spatio-temporal conditional biases, our framework first constructs a support set and a group of query sets from the training data, where the data distribution of each query set is different from that of the support set w.r.t. a type of conditional bias. Then, by performing a novel meta training and testing process to optimize the model to obtain good testing performance on these query sets after training on the support set, our framework can effectively guide the model to learn to well generalize against biases. Extensive experiments demonstrate the efficacy of our proposed framework.



Paperid:1123
Authors:Haoxuan Qu; Yanchao Li; Lin Geng Foo; Jason Kuen; Jiuxiang Gu; Jun Liu
Title: Improving the Reliability for Confidence Estimation
Abstract:
Confidence estimation, a task that aims to evaluate the trustworthiness of the model’s prediction output during deployment, has received lots of research attention recently, due to its importance for the safe deployment of deep models. Previous works have outlined two important qualities that a reliable confidence estimation model should possess, i.e., the ability to perform well under label imbalance and the ability to handle various out-of-distribution data inputs. In this work, we propose a meta-learning framework that can simultaneously improve upon both qualities in a confidence estimation model. Specifically, we first construct virtual training and testing sets with some intentionally designed distribution differences between them. Our framework then uses the constructed sets to train the confidence estimation model through a virtual training and testing scheme leading it to learn knowledge that generalizes to diverse distributions. We show the effectiveness of our framework on both monocular depth estimation and image classification.



Paperid:1124
Authors:Ao Zhang; Yuan Yao; Qianyu Chen; Wei Ji; Zhiyuan Liu; Maosong Sun; Tat-Seng Chua
Title: Fine-Grained Scene Graph Generation with Data Transfer
Abstract:
Scene graph generation (SGG) is designed to extract (subject, predicate, object) triplets in images. Recent works have made a steady progress on SGG, and provide useful tools for high-level vision and language understanding. However, due to the data distribution problems including long-tail distribution and semantic ambiguity, the predictions of current SGG models tend to collapse to several frequent but uninformative predicates (e.g., on, at), which limits practical application of these models in downstream tasks. To deal with the problems above, we propose a novel Internal and External Data Transfer (IETrans) method, which can be applied in a plug-and-play fashion and expanded to large SGG with 1,807 predicate classes. Our IETrans tries to relieve the data distribution problem by automatically creating an enhanced dataset that provides more sufficient and coherent annotations for all predicates. By applying our proposed method, a Neural Motif model doubles the macro performance for informative SGG. The code and data are publicly available at https://github.com/waxnkw/IETrans-SGG.pytorch.



Paperid:1125
Authors:Yinyu Nie; Angela Dai; Xiaoguang Han; Matthias Nießner
Title: Pose2Room: Understanding 3D Scenes from Human Activities
Abstract:
With wearable IMU sensors, one can estimate human poses from wearable devices without requiring visual input. In this work, we pose the question: Can we reason about object structure in real-world environments solely from human trajectory information? Crucially, we observe that human motion and interactions tend to give strong information about the objects in a scene -- for instance a person sitting indicates the likely presence of a chair or sofa. To this end, we propose P2R-Net to learn a probabilistic 3D model of the objects in a scene characterized by their class categories and oriented 3D bounding boxes, based on an input observed human trajectory in the environment. P2R-Net models the probability distribution of object class as well as a deep Gaussian mixture model for object boxes, enabling sampling of multiple, diverse, likely modes of object configurations from an observed human trajectory. In our experiments we show that P2R-Net can effectively learn multi-modal distributions of likely objects for human motions, and produce a variety of plausible object structures of the environment, even without any visual information. The results demonstrate that P2R-Net consistently outperforms the baselines on the PROX dataset and the VirtualHome platform.



Paperid:1126
Authors:Xubin Zhong; Changxing Ding; Zijian Li; Shaoli Huang
Title: Towards Hard-Positive Query Mining for DETR-Based Human-Object Interaction Detection
Abstract:
Human-Object Interaction (HOI) detection is a core task for high-level image understanding. Recently, Detection Transformer (DETR)-based HOI detectors have become popular due to their superior performance and efficient structure. However, these approaches typically adopt fixed HOI queries for all testing images, which is vulnerable to the location change of objects in one specific image. Accordingly, in this paper, we propose to enhance DETR’s robustness by mining hard-positive queries, which are forced to make correct predictions using partial visual cues. First, we explicitly compose hard-positive queries according to the ground-truth (GT) position of labeled human-object pairs for each training image. Specifically, we shift the GT bounding boxes of each labeled human-object pair so that the shifted boxes cover only a certain portion of the GT ones. We encode the coordinates of the shifted boxes for each labeled human-object pair into an HOI query. Second, we implicitly construct another set of hard-positive queries by masking the top scores in cross-attention maps of the decoder layers. The masked attention maps then only cover partial important cues for HOI predictions. Finally, an alternate strategy is proposed that efficiently combines both types of hard queries. In each iteration, both DETR’s learnable queries and one selected type of hard-positive queries are adopted for loss computation. Experimental results show that our proposed approach can be widely applied to existing DETR-based HOI detectors. Moreover, we consistently achieve state-of-the-art performance on three benchmarks: HICO-DET, V-COCO, and HOI-A.



Paperid:1127
Authors:Zhi Hou; Baosheng Yu; Dacheng Tao
Title: Discovering Human-Object Interaction Concepts via Self-Compositional Learning
Abstract:
A comprehensive understanding of human-object interaction (HOI) requires detecting not only a small portion of predefined HOI concepts (or categories) but also other reasonable HOI concepts, while current approaches usually fail to explore a huge portion of unknown HOI concepts (i.e., unknown but reasonable combinations of verbs and objects). In this paper, 1) we introduce a novel and challenging task for a comprehensive HOI understanding, which is termed as HOI Concept Discovery; and 2) we devise a self-compositional learning framework (or SCL) for HOI concept discovery. Specifically, we maintain an online updated concept confidence matrix during training: 1) we assign pseudo-labels for all composite HOI instances according to the concept confidence matrix for self-training; and 2) we update the concept confidence matrix using the predictions of all composite HOI instances. Therefore, the proposed method enables the learning on both known and unknown HOI concepts. We perform extensive experiments on several popular HOI datasets to demonstrate the effectiveness of the proposed method for HOI concept discovery, object affordance recognition and HOI detection. For example, the proposed self-compositional learning framework significantly improves the performance of 1) HOI concept discovery by over 10% on HICO-DET and over 3% on V-COCO, respectively; 2) object affordance recognition by over 9% mAP on MS-COCO and HICO-DET; and 3) rare-first and non-rare-first unknown HOI detection relatively over 30% and 20%, respectively. Code is publicly available at https://github.com/zhihou7/HOI-CL.



Paperid:1128
Authors:Yuwei Wu; Weixiao Liu; Sipu Ruan; Gregory S. Chirikjian
Title: Primitive-Based Shape Abstraction via Nonparametric Bayesian Inference
Abstract:
3D shape abstraction has drawn great interest over the years. Apart from low-level representations such as meshes and voxels, researchers also seek to semantically abstract complex objects with basic geometric primitives. Recent deep learning methods rely heavily on datasets, with limited generality to unseen categories. Furthermore, abstracting an object accurately yet with a small number of primitives still remains a challenge. In this paper, we propose a novel non-parametric Bayesian statistical method to infer an abstraction, consisting of an unknown number of geometric primitives, from a point cloud. We model the generation of points as observations sampled from an infinite mixture of Gaussian Superquadric Taper Models (GSTM). Our approach formulates the abstraction as a clustering problem, in which: 1) each point is assigned to a cluster via the Chinese Restaurant Process (CRP); 2) a primitive representation is optimized for each cluster, and 3) a merging post-process is incorporated to provide a concise representation. We conduct extensive experiments on two datasets. The results indicate that our method outperforms the state-of-the-art in terms of accuracy and is generalizable to various types of objects.



Paperid:1129
Authors:Chenghao Zhang; Kun Tian; Bolin Ni; Gaofeng Meng; Bin Fan; Zhaoxiang Zhang; Chunhong Pan
Title: Stereo Depth Estimation with Echoes
Abstract:
Stereo depth estimation is particularly amenable to local textured regions while echoes have good depth estimations for global textureless regions, thus the two modalities complement each other. Motivated by the reciprocal relationship between both modalities, in this paper, we propose an end-to-end framework named StereoEchoes for stereo depth estimation with echoes. A Cross-modal Volume Refinement module is designed to transfer the complementary knowledge of the audio modality to the visual modality at feature level. A Relative Depth Uncertainty Estimation module is further proposed to yield pixel-wise confidence for multimodal depth fusion at output space. As there is no dataset for this new problem, we introduce two Stereo-Echo datasets named Stereo-Replica and Stereo-Matterport3D for the first time. Remarkably, we show empirically that our StereoEchoes, on Stereo-Replica and Stereo-Matterport3D, outperforms stereo depth estimation methods by 25%/13.8% RMSE, and surpasses the state-of-the-art audio-visual depth prediction method by 25.3%/42.3% RMSE.



Paperid:1130
Authors:Hanrong Ye; Dan Xu
Title: Inverted Pyramid Multi-task Transformer for Dense Scene Understanding
Abstract:
Multi-task dense scene understanding is a thriving research domain that requires simultaneous perception and reasoning on a series of correlated tasks with pixel-wise prediction. Most existing works encounter a severe limitation of modeling in the locality due to heavy utilization of convolution operations, while learning interactions and inference in a global spatial-position and multi-task context is critical for this problem. In this paper, we propose a novel end-to-end Inverted Pyramid multi-task Transformer (InvPT) to perform simultaneous modeling of spatial positions and multiple tasks in a unified framework. To the best of our knowledge, this is the first work that explores designing a transformer structure for multi-task dense prediction for scene understanding. Besides, it is widely demonstrated that a higher spatial resolution is remarkably beneficial for dense predictions, while it is very challenging for existing transformers to go deeper with higher resolutions due to huge complexity to large spatial size. InvPT presents an efficient UP-Transformer block to learn multi-task feature interaction at gradually increased resolutions, which also incorporates effective self-attention message passing and multi-scale feature aggregation to produce task-specific prediction at a high resolution. Our method achieves superior multi-task performance on NYUD-v2 and PASCAL-Context datasets respectively, and significantly outperforms previous state-of-the-arts. The code is available at https://github.com/prismformore/InvPT.



Paperid:1131
Authors:Yingfei Liu; Tiancai Wang; Xiangyu Zhang; Jian Sun
Title: PETR: Position Embedding Transformation for Multi-View 3D Object Detection
Abstract:
In this paper, we develop position embedding transformation (PETR) for multi-view 3D object detection. PETR encodes the position information of 3D coordinates into image features, producing the 3D position-aware features. Object query can perceive the 3D position-aware features and perform end-to-end object detection. Till submission, PETR achieves state-of-the-art performance (50.4% NDS and 44.1% mAP) on standard nuScenes dataset and ranks 1st place on the benchmark. It can serve as a simple yet strong baseline for future research.



Paperid:1132
Authors:Xinshuo Weng; Junyu Nan; Kuan-Hui Lee; Rowan McAllister; Adrien Gaidon; Nicholas Rhinehart; Kris M. Kitani
Title: S2Net: Stochastic Sequential Pointcloud Forecasting
Abstract:
Predicting futures of surrounding agents is critical for autonomous systems such as self-driving cars. Instead of requiring accurate detection and tracking prior to trajectory prediction, an object agnostic Sequential Pointcloud Forecasting (SPF) task was proposed in prior work, which enables a forecast-then-detect pipeline effective for downstream detection and trajectory prediction. One limitation of prior work is that it forecasts only a deterministic sequence of future point clouds, despite the inherent uncertainty of dynamic scenes. In this work, we tackle the stochastic SPF problem by proposing a generative model with two main components: (1) a conditional variational recurrent neural network that models a temporally-dependent latent space; (2) a pyramid-LSTM that increases the fidelity of predictions with temporally-aligned skip connections. Through experiments on real-world autonomous driving datasets, our stochastic SPF model produces higher-fidelity predictions, reducing Chamfer distances by up to 56.6% compared to its deterministic counterpart. In addition, our model can estimate the uncertainty of predicted points, which can be helpful to downstream tasks.



Paperid:1133
Authors:Mu He; Le Hui; Yikai Bian; Jian Ren; Jin Xie; Jian Yang
Title: RA-Depth: Resolution Adaptive Self-Supervised Monocular Depth Estimation
Abstract:
Existing self-supervised monocular depth estimation methods can get rid of expensive annotations and achieve promising results. However, these methods suffer from severe performance degradation when directly adopting a model trained on a fixed resolution to evaluate at other different resolutions. In this paper, we propose a resolution adaptive self-supervised monocular depth estimation method (RA-Depth) by learning the scale invariance of the scene depth. Specifically, we propose a simple yet efficient data augmentation method to generate images with arbitrary scales for the same scene. Then, we develop a dual high-resolution network that uses the multi-path encoder and decoder with dense interactions to aggregate multi-scale features for accurate depth inference. Finally, to explicitly learn the scale invariance of the scene depth, we formulate a cross-scale depth consistency loss on depth predictions with different scales. Extensive experiments on the KITTI, Make3D and NYU-V2 datasets demonstrate that RA-Depth not only achieves state-of-the-art performance, but also exhibits a good ability of resolution adaptation.



Paperid:1134
Authors:Haobo Yuan; Xiangtai Li; Yibo Yang; Guangliang Cheng; Jing Zhang; Yunhai Tong; Lefei Zhang; Dacheng Tao
Title: PolyphonicFormer: Unified Query Learning for Depth-Aware Video Panoptic Segmentation
Abstract:
The Depth-aware Video Panoptic Segmentation (DVPS) is a new challenging vision problem that aims to predict panoptic segmentation and depth in a video simultaneously. The previous work solves this task by extending the existing panoptic segmentation method with an extra dense depth prediction and instance tracking head. However, the relationship between the depth and panoptic segmentation is not well explored -- simply combining existing methods leads to competition and needs carefully weight balancing. In this paper, we present PolyphonicFormer, a vision transformer to unify these sub-tasks under the DVPS task and lead to more robust results. Our principal insight is that the depth can be harmonized with the panoptic segmentation with our proposed new paradigm of predicting instance level depth maps with object queries. Then the relationship between the two tasks via query-based learning is explored. From the experiments, we demonstrate the benefits of our design from both depth estimation and panoptic segmentation aspects. Since each thing query also encodes the instance-wise information, it is natural to perform tracking directly with appearance learning. Our method achieves state-of-the-art results on two DVPS datasets (Semantic KITTI, Cityscapes), and ranks 1st on the ICCV-2021 BMTT Challenge video + depth track. Code is available at https://github.com/HarborYuan/PolyphonicFormer.



Paperid:1135
Authors:Qingyong Hu; Bo Yang; Guangchi Fang; Yulan Guo; Aleš Leonardis; Niki Trigoni; Andrew Markham
Title: SQN: Weakly-Supervised Semantic Segmentation of Large-Scale 3D Point Clouds
Abstract:
Labelling point clouds fully is highly time-consuming and costly. As larger point cloud datasets containing billions of points become more common, we ask whether the full annotation is even necessary, demonstrating that existing baselines designed under a fully annotated assumption only degrade slightly even when faced with 1% random point annotations. However, beyond this point, e.g. at 0.1% annotations, segmentation accuracy is unacceptably low. We observe that, as point clouds are samples of the 3D world, the distribution of points in a local neighbourhood is relatively homogeneous, exhibiting strong semantic similarity. Motivated by this, we propose a new weak supervision method to implicitly augment these highly sparse supervision signals. Extensive experiments demonstrate that the proposed Semantic Query Network (SQN) achieves state-of-the-art performance on seven large-scale open datasets under weak supervision schemes, while requiring only 0.1% randomly annotated points for training, greatly reducing annotation cost and effort.



Paperid:1136
Authors:Jaesung Choe; Chunghyun Park; Francois Rameau; Jaesik Park; In So Kweon
Title: PointMixer: MLP-Mixer for Point Cloud Understanding
Abstract:
MLP-Mixer has newly appeared as a new challenger against the realm of CNNs and Transformer. Despite its simplicity compared to Transformer, the concept of channel-mixing MLPs and token-mixing MLPs achieves noticeable performance in image recognition tasks. Unlike images, point clouds are inherently sparse, unordered and irregular, which limits the direct use of MLP-Mixer for point cloud understanding. To overcome these limitations, we propose PointMixer, a universal point set operator that facilitates information sharing among unstructured 3D point cloud. By simply replacing token-mixing MLPs with Softmax function, PointMixer can “mix"" features within/between point sets. By doing so, PointMixer can be broadly used for intra-set, inter-set, and hierarchical-set mixing. We demonstrate that various channel-wise feature aggregation in numerous point sets is better than self-attention layers or dense token-wise interaction in a view of parameter efficiency and accuracy. Extensive experiments show the competitive or superior performance of PointMixer in semantic segmentation, classification, and reconstruction against Transformer-based methods.



Paperid:1137
Authors:Xiaoming Zhao; Zhizhen Zhao; Alexander G. Schwing
Title: Initialization and Alignment for Adversarial Texture Optimization
Abstract:
While recovery of geometry from image and video data has received a lot of attention in computer vision, methods to capture the texture for a given geometry are less mature. Specifically, classical methods for texture generation often assume clean geometry and reasonably well-aligned image data. While very recent methods, e.g., adversarial texture optimization, better handle lower-quality data obtained from hand-held devices, we find them to still struggle frequently. To improve robustness, particularly of recent adversarial texture optimization, we develop an explicit initialization and an alignment procedure. It deals with complex geometry due to a robust mapping of the geometry to the texture map and a hard-assignment-based initialization. It deals with misalignment of geometry and images by integrating fast image-alignment into the texture refinement optimization. We demonstrate efficacy of our texture generation on a dataset of 11 scenes with a total of 2807 frames, observing 7.8% and 11.1% relative improvements regarding perceptual and sharpness measurements.



Paperid:1138
Authors:Fangao Zeng; Bin Dong; Yuang Zhang; Tiancai Wang; Xiangyu Zhang; Yichen Wei
Title: MOTR: End-to-End Multiple-Object Tracking with TRansformer
Abstract:
Temporal modeling of objects is a key challenge in multiple-object tracking (MOT). Existing methods track by associating detections through motion-based and appearance-based similarity heuristics. The post-processing nature of association prevents end-to-end exploitation of temporal variations in video sequence. In this paper, we propose MOTR, which extends DETR \cite{carion2020detr} and introduces “track query” to model the tracked instances in the entire video. Track query is transferred and updated frame-by-frame to perform iterative prediction over time. We propose tracklet-aware label assignment to train track queries and newborn object queries. We further propose temporal aggregation network and collective average loss to enhance temporal relation modeling. Experimental results on DanceTrack show that MOTR significantly outperforms state-of-the-art method, ByteTrack by 6.5\% on HOTA metric. On MOT17, MOTR outperforms our concurrent works, TrackFormer and TransTrack, on association performance. MOTR can serve as a stronger baseline for future research on temporal modeling and Transformer-based trackers. Code is available at \url{https://github.com/anonymous4669/MOTR}.



Paperid:1139
Authors:Sijie Zhu; Zhe Lin; Scott Cohen; Jason Kuen; Zhifei Zhang; Chen Chen
Title: GALA: Toward Geometry-and-Lighting-Aware Object Search for Compositing
Abstract:
Compositing-aware object search aims to find the most compatible objects for compositing given a background image and a query bounding box. Previous works focus on learning compatibility between the foreground object and background, but fail to learn other important factors from large-scale data, i.e. geometry and lighting. To move a step further, this paper proposes GALA (Geometry-and-Lighting-Aware), a generic foreground object search method with discriminative modeling on geometry and lighting compatibility for open-world image compositing. Remarkably, it achieves state-of-the-art results on the CAIS dataset and generalizes well on large-scale open-world datasets, i.e. Pixabay and Open Images. In addition, our method can effectively handle non-box scenarios, where users only provide background images without any input bounding box. Experiments are conducted on real-world images to showcase applications of the proposed method for compositing-aware search and automatic location/scale prediction for the foreground object.



Paperid:1140
Authors:Henry Howard-Jenkins; Victor Adrian Prisacariu
Title: LaLaLoc++: Global Floor Plan Comprehension for Layout Localisation in Unvisited Environments
Abstract:
We present LaLaLoc++, a method for floor plan localisation in unvisited environments through latent representations of room layout. We perform localisation by aligning room layout inferred from a panorama image with the floor plan of a scene. To process a floor plan prior, previous methods required that the plan first be used to construct an explicit 3D representation of the scene. This process requires that assumptions be made about the scene geometry and can result in expensive steps becoming necessary, such as rendering. LaLaLoc++ instead introduces a global floor plan comprehension module that is able to efficiently infer structure densely and directly from the 2D plan, removing any need for explicit modelling or rendering. On the Structured3D dataset this module alone improves localisation accuracy by more than 31%, all while increasing throughput by an order of magnitude. Combined with the further addition of a transformer-based panorama embedding module, LaLaLoc++ improves accuracy over earlier methods by more than 37% with dramatically faster inference.



Paperid:1141
Authors:Yu-Ting Yen; Chia-Ni Lu; Wei-Chen Chiu; Yi-Hsuan Tsai
Title: 3D-PL: Domain Adaptive Depth Estimation with 3D-Aware Pseudo-Labeling
Abstract:
For monocular depth estimation, acquiring ground truths for real data is not easy, and thus domain adaptation methods are commonly adopted using the supervised synthetic data. However, this may still incur a large domain gap due to the lack of supervision from the real data. In this paper, we develop a domain adaptation framework via generating reliable pseudo ground truths of depth from real data to provide direct supervisions. Specifically, we propose two mechanisms for pseudo-labeling: 1) 2D-based pseudo labels via measuring the consistency of depth predictions when images are with the same content but different styles; 2) 3D-aware pseudo labels via a point cloud completion network that learns to complete the depth values in the 3D space, thus providing more structural information in a scene to refine and generate more reliable pseudo labels. In experiments, we show that our pseudo-labeling methods improve depth estimation in various settings, including the usage of stereo pairs during training. Furthermore, the proposed method performs favorably against several state-of-the-art unsupervised domain adaptation approaches in real-world datasets.



Paperid:1142
Authors:Xiangtai Li; Shilin Xu; Yibo Yang; Guangliang Cheng; Yunhai Tong; Dacheng Tao
Title: Panoptic-PartFormer: Learning a Unified Model for Panoptic Part Segmentation
Abstract:
Panoptic Part Segmentation (PPS) aims to unify panoptic segmentation and part segmentation into one task. Previous work mainly utilizes separated approaches to handle thing, stuff, and part predictions individually without performing any shared computation and task association. In this work, we aim to unify these tasks at the architectural level, designing the first end-to-end unified method named Panoptic-PartFormer. In particular, motivated by the recent progress in Vision Transformer, we model things, stuff, and part as object queries and directly learn to optimize the all three predictions as unified mask prediction and classification problem. We design a decoupled decoder to generate part feature and thing/stuff feature respectively. Then we propose to utilize all the queries and corresponding features to perform reasoning jointly and iteratively. The final mask can be obtained via inner product between queries and the corresponding features. The extensive ablation studies and analysis prove the effectiveness of our framework. Our Panoptic-PartFormer achieves the new state-of-the-art results on both Cityscapes PPS and Pascal Context PPS datasets with around 70\% GFlops and 50\% parameters decrease. Given its effectiveness and conceptual simplicity, we hope the Panoptic-PartFormer can serve as a strong baseline and aid future research in PPS. Our code and models will be available at \url{https://github.com/lxtGH/Panoptic-PartFormer}.



Paperid:1143
Authors:Songlin Fan; Wei Gao; Ge Li
Title: Salient Object Detection for Point Clouds
Abstract:
This paper researches the unexplored task-point cloud salient object detection (SOD). Differing from SOD for images, we find the attention shift of point clouds may provoke saliency conflict, i.e., an object paradoxically belongs to salient and non-salient categories. To eschew this issue, we present a novel view-dependent perspective of salient objects, reasonably reflecting the most eye-catching objects in point cloud scenarios. Following this formulation, we introduce PCSOD, the first dataset proposed for point cloud SOD consisting of 2,872 in-/out-door 3D views. The samples in our dataset are labeled with hierarchical annotations, e.g., super-/sub-class, bounding box, and segmentation map, which endows the brilliant generalizability and broad applicability of our dataset verifying various conjectures. To evidence the feasibility of our solution, we further contribute a baseline model and benchmark five representative models for a comprehensive comparison. The proposed model can effectively analyze irregular and unordered points for detecting salient objects. Thanks to incorporating the task-tailored designs, our method shows visible superiority over other baselines, producing more satisfactory results. Extensive experiments and discussions reveal the promising potential of this research field, paving the way for further study.



Paperid:1144
Authors:Dongwan Kim; Yi-Hsuan Tsai; Yumin Suh; Masoud Faraki; Sparsh Garg; Manmohan Chandraker; Bohyung Han
Title: Learning Semantic Segmentation from Multiple Datasets with Label Shifts
Abstract:
While it is desirable to train segmentation models on an aggregation of multiple datasets, a major challenge is that the label space of each dataset may be in conflict with one another. To tackle this challenge, we propose UniSeg, an effective and model-agnostic approach to automatically train segmentation models across multiple datasets with heterogeneous label spaces, without requiring any manual relabeling efforts. Specifically, we introduce two new ideas that account for conflicting and co-occurring labels to achieve better generalization performance in unseen domains. First, we identify a gradient conflict in training incurred by mismatched label spaces and propose a class-independent binary cross-entropy loss to alleviate such label conflicts. Second, we propose a loss function that considers class-relationships across datasets for a better multi-dataset training scheme.xtensive quantitative and qualitative analyses on road-scene datasets show that UniSeg improves over multi-dataset baselines, especially on unseen datasets, e.g., achieving more than 8%p gain in IoU on KITTI. Furthermore, UniSeg achieves 39.4% IoU on the WildDash2 public benchmark, making it one of the strongest submissions in the zero-shot setting. Our project page is available at https://www.nec-labs.com/ mas/UniSeg.



Paperid:1145
Authors:Kangcheng Liu; Yuzhi Zhao; Qiang Nie; Zhi Gao; Ben M. Chen
Title: Weakly Supervised 3D Scene Segmentation with Region-Level Boundary Awareness and Instance Discrimination
Abstract:
Current state-of-the-art 3D scene understanding methods are merely designed in a full-supervised way. However, in the limited reconstruction cases, only limited 3D scenes can be reconstructed and annotated. We are in need of a framework that can concurrently be applied to 3D point cloud semantic segmentation and instance segmentation, particularly in circumstances where labels are rather scarce. The paper introduces an effective approach to tackle the 3D scene understanding problem when labeled scenes are limited. To leverage the boundary information, we propose a novel energy-based loss with boundary awareness benefiting from the region-level boundary labels predicted by the boundary prediction network. To encourage latent instance discrimination and guarantee efficiency, we propose the first unsupervised region-level semantic contrastive learning scheme for point clouds, which uses confident predictions of the network to discriminate the intermediate feature embeddings in multiple stages. In the limited reconstruction case, our proposed approach, termed WS3D, has pioneer performance on the large-scale ScanNet on semantic segmentation and instance segmentation. Also, our proposed WS3D achieves state-of-the-art performance on the other indoor and outdoor datasets S3DIS and SemanticKITTI.



Paperid:1146
Authors:Tao He; Lianli Gao; Jingkuan Song; Yuan-Fang Li
Title: Towards Open-Vocabulary Scene Graph Generation with Prompt-Based Finetuning
Abstract:
Scene graph generation (SGG) is a fundamental task aimed at detecting visual relations between objects in an image. The prevailing SGG methods require all object classes to be given in the training set. Such a closed setting limits the practical application of SGG. In this paper, we introduce open-vocabulary scene graph generation, a novel, realistic and challenging setting, in which a model is trained on a small set of base object classes but is required to infer relations for unseen target object classes. To this end, we propose a two-step method which firstly pre-trains on large amounts of coarse-grained region-caption data and then leverage two prompt-based techniques to finetune the pre-trained model without updating its parameters. Moreover, our method is able to support inference over completely unseen object classes, which existing methods are incapable of handling. On extensive experiments on three benchmark datasets, Visual Genome, GQA and Open-Image, our method significantly outperforms recent, strong SGG methods on the setting of Ov-SGG, as well as on the conventional closed SGG.



Paperid:1147
Authors:Pedro Hermosilla; Michael Schelling; Tobias Ritschel; Timo Ropinski
Title: Variance-Aware Weight Initialization for Point Convolutional Neural Networks
Abstract:
Appropriate weight initialization has been of key importance to successfully train neural networks. Recently, batch normalization has diminished the role of weight initialization by simply normalizing each layer based on batch statistics. Unfortunately, batch normalization has several drawbacks when applied to small batch sizes, as they are required to cope with memory limitations when learning on point clouds. While well-founded weight initialization strategies can render batch normalization unnecessary and thus avoid these drawbacks, no such approaches have been proposed for point convolutional networks. To fill this gap, we propose a framework to unify the multitude of continuous convolutions. This enables our main contribution, variance-aware weight initialization. We show that this initialization can avoid batch normalization while achieving similar and, in some cases, better performance.



Paperid:1148
Authors:Aaron Walsman; Muru Zhang; Klemen Kotar; Karthik Desingh; Ali Farhadi; Dieter Fox
Title: Break and Make: Interactive Structural Understanding Using LEGO Bricks
Abstract:
Visual understanding of geometric structures with complex spatial relationships is a fundamental component of human intelligence. As children, we learn how to reason about structure not only from observation, but also by interacting with the world around us - by taking things apart and putting them back together again. The ability to reason about structure and compositionality allows us to not only build things, but also understand and reverse-engineer complex systems. In order to advance research in interactive reasoning for part-based geometric understanding, we propose a challenging new assembly problem using LEGO bricks that we call Break and Make. In this problem an agent is given a LEGO model and attempts to understand its structure by interactively inspecting and disassembling it. After this inspection period, the agent must then prove its understanding by rebuilding the model from scratch using low-level action primitives. In order to facilitate research on this problem we have built \LTRON, a fully interactive 3D simulator that allows learning agents to assemble, disassemble and manipulate LEGO models. We pair this simulator with a new dataset of fan-made LEGO creations that have been uploaded to the internet in order to provide complex scenes containing over a thousand unique brick shapes. We take a first step towards solving this problem using sequence-to-sequence models that provide guidance for how to make progress on this challenging problem. Our simulator and data are available at github.com/aaronwalsman/ltron. Additional training code and PyTorch examples are available at github.com/aaronwalsman/ltron-torch-eccv22.



Paperid:1149
Authors:Wencan Cheng; Jong Hwan Ko
Title: Bi-PointFlowNet: Bidirectional Learning for Point Cloud Based Scene Flow Estimation
Abstract:
Scene flow estimation, which extracts point-wise motion between scenes, is becoming a crucial task in many computer vision tasks. However, all of the existing estimation methods utilize only the unidirectional features, restricting the accuracy and generality. This paper presents a novel scene flow estimation architecture using bidirectional flow embedding layers. The proposed bidirectional layer learns features along both forward and backward directions, enhancing the estimation performance. In addition, hierarchical feature extraction and warping improve the performance and reduce computational overhead. Experimental results show that the proposed architecture achieved a new state-of-the-art record by outperforming other approaches with large margin in both FlyingThings3D and KITTI benchmarks. Codes are available at https://github.com/cwc1260/BiFlow.



Paperid:1150
Authors:Runyu Mao; Chen Bai; Yatong An; Fengqing Zhu; Cheng Lu
Title: 3DG-STFM: 3D Geometric Guided Student-Teacher Feature Matching
Abstract:
We tackle the essential task of finding dense visual correspondences between a pair of images. This is a challenging problem due to various factors such as poor texture, repetitive patterns, illumination variation, and motion blur in practical scenarios. In contrast to methods that use dense correspondence ground-truths as direct supervision for local feature matching training, we train 3DG-STFM: a multi-modal matching model (Teacher) to enforce the depth consistency under 3D dense correspondence supervision and transfer the knowledge to 2D unimodal matching model (Student). Both teacher and student models consist of two transformer-based matching modules that obtain dense correspondences in a coarse-to-fine manner. The teacher model guides the student model to learn RGB-induced depth information for the matching purpose on both coarse and fine branches. We also evaluate 3DG-STFM on a model compression task. To the best of our knowledge, 3DG-STFM is the first student-teacher learning method for the local feature matching task. The experiments show that our method outperforms state-of-the-art methods on indoor and outdoor camera pose estimations, and homography estimation problems. Code is available at: https://github.com/Ryan-prime/3DG-STFM"



Paperid:1151
Authors:Prashant W Patil; Sunil Gupta; Santu Rana; Svetha Venkatesh
Title: Video Restoration Framework and Its Meta-Adaptations to Data-Poor Conditions
Abstract:
Restoration of weather degraded videos is a challenging problem due to diverse weather conditions e.g., rain, haze, snow, etc. Existing works handle video restoration for each weather using a different custom-designed architecture. This approach has many limitations. First, a custom-designed architecture for each weather condition requires domain-specific knowledge. Second, disparate network architectures across weather conditions prevent easy knowledge transfer to novel weather conditions where we do not have a lot of data to train a model from scratch. For example, while there is a lot of common knowledge to exploit between the models of different weather conditions at day or night time, it is difficult to do such adaptation. To this end, we propose a generic architecture that is effective for any weather condition due to the ability to extract robust feature maps without any domain-specific knowledge. This is achieved by novel components: spatio-temporal feature modulation, multi-level feature aggregation, and recurrent guidance decoder. Next, we propose a meta-learning based adaptation of our deep architecture to the restoration of videos in data-poor conditions (night-time videos). We show comprehensive results on video de-hazing and de-raining datasets in addition to the meta-learning based adaptation results on night-time video restoration tasks. Our results clearly outperform the state-of-the-art weather degraded video restoration methods.



Paperid:1152
Authors:Michaël Ramamonjisoa; Sinisa Stekovic; Vincent Lepetit
Title: MonteBoxFinder: Detecting and Filtering Primitives to Fit a Noisy Point Cloud
Abstract:
We present MonteBoxFinder, a method that, given an noisy input point cloud, detects a dense set of imperfect boxes, and employs a discrete optimization algorithm that efficiently explores the space of allbox arrangements in order to find the arrangement that best fits the pointcloud. Our method demonstrates significant superiority of our method over our discrete optimization baselines on the ScanNet dataset, both inefficiency and precision. This is achieved by leveraging the structure of the problem, which is that the fit quality of a cuboid arrangement is invariant to cuboid permutation. Finally, our solution search algorithm is general, as it can be extended to other challenging discrete optimization scenarii.



Paperid:1153
Authors:Darwin Bautista; Rowel Atienza
Title: Scene Text Recognition with Permuted Autoregressive Sequence Models
Abstract:
Context-aware STR methods typically use internal autoregressive (AR) language models (LM). Inherent limitations of AR models motivated two-stage methods which employ an external LM. The conditional independence of the external LM on the input image may cause it to erroneously rectify correct predictions, leading to significant inefficiencies. Our method, PARSeq, learns an ensemble of internal AR LMs with shared weights using Permutation Language Modeling. It unifies context-free non-AR and context-aware AR inference, and iterative refinement using bidirectional context. Using synthetic training data, PARSeq achieves state-of-the-art (SOTA) results in STR benchmarks (91.9% accuracy) and more challenging datasets. It establishes new SOTA results (96.0% accuracy) when trained on real data. PARSeq is optimal on accuracy vs parameter count, FLOPS, and latency because of its simple, unified structure and parallel token processing. Due to its extensive use of attention, it is robust on arbitrarily-oriented text which is common in real-world images. Code, pretrained weights, and data are available at: https://github.com/baudm/parseq.



Paperid:1154
Authors:Bohan Li; Ye Yuan; Dingkang Liang; Xiao Liu; Zhilong Ji; Jinfeng Bai; Wenyu Liu; Xiang Bai
Title: When Counting Meets HMER: Counting-Aware Network for Handwritten Mathematical Expression Recognition
Abstract:
Recently, most handwritten mathematical expression recognition (HMER) methods adopt the encoder-decoder networks, which directly predict the markup sequences from formula images with the attention mechanism. However, such methods may fail to accurately read formulas with complicated structure or generate long markup sequences, as the attention results are often inaccurate due to the large variance of writing styles or spatial layouts. To alleviate this problem, we propose an unconventional network for HMER named Counting-Aware Network (CAN), which jointly optimizes two tasks: HMER and symbol counting. Specifically, we design a weakly-supervised counting module that can predict the number of each symbol class without the symbol-level position annotations, and then plug it into a typical attention-based encoder-decoder model for HMER. Experiments on the benchmark datasets for HMER validate that both joint optimization and counting results are beneficial for correcting the prediction errors of encoder-decoder models, and CAN consistently outperforms the state-of-the-art methods. In particular, compared with an encoder-decoder model for HMER, the extra time cost caused by the proposed counting module is marginal. The source code is available at https://github.com/LBH1024/CAN.



Paperid:1155
Authors:Yuxin Wang; Hongtao Xie; Mengting Xing; Jing Wang; Shenggao Zhu; Yongdong Zhang
Title: Detecting Tampered Scene Text in the Wild
Abstract:
Text manipulation technologies cause serious worries in recent years, however, corresponding tampering detection methods have not been well explored. In this paper, we introduce a new task, named Tampered Scene Text Detection (TSTD), to localize text instances and recognize the texture authenticity in an end-to-end manner. Different from the general scene text detection (STD) task, TSTD further introduces the fine-grained classification, i.e. the tampered and real-world texts share a semantic space (text position and geometric structure) but have different local textures. To this end, we propose a simple yet effective modification strategy to migrate existing STD methods to TSTD task, keeping the semantic invariance while explicitly guiding the class-specific texture feature learning. Furthermore, we discuss the potential of frequency information for distinguishing feature learning, and propose a parallel-branch feature extractor to enhance the feature representation capability. To evaluate the effectiveness of our method, a new TSTD dataset (Tampered-IC13) is proposed and released at https://github.com/wangyuxin87/Tampered-IC13.



Paperid:1156
Authors:Jingqun Tang; Wenming Qian; Luchuan Song; Xiena Dong; Lan Li; Xiang Bai
Title: Optimal Boxes: Boosting End-to-End Scene Text Recognition by Adjusting Annotated Bounding Boxes via Reinforcement Learning
Abstract:
Text detection and recognition are the essential components of a modern OCR system. Most OCR approaches attempt to obtain accurate bounding boxes of text at the detection stage, which is used as the input of the text recognition stage. We observe that when using tight text bounding boxes as input, a text recognizer frequently fails to achieve optimal performance due to the inconsistency between bounding boxes and deep representations of text recognition. In this paper, we propose Box Adjuster, a reinforcement learning-based method for adjusting the shape of each text bounding box to make it more compatible with text recognition models. Additionally, when dealing with cross-domain problems such as synthetic-to-real, the proposed method significantly reduces mismatches in domain distribution between the source and target domains. Experiments demonstrate that the performance of end-to-end text recognition systems can be improved when using the adjusted bounding boxes as the ground truth for training. Specifically, on several benchmark datasets for scene text understanding, the proposed method outperforms state-of-the-art text spotters by an average of 2.0% F-Score on end-to-end text recognition tasks and 4.6% F-Score on domain adaptation tasks.



Paperid:1157
Authors:Roi Ronen; Shahar Tsiper; Oron Anschel; Inbal Lavi; Amir Markovitz; R. Manmatha
Title: GLASS: Global to Local Attention for Scene-Text Spotting
Abstract:
In recent years, the dominant paradigm for text spotting is to combine the tasks of text detection and recognition into a single end-to-end framework. Under this paradigm, both tasks are accomplished by operating over a shared global feature map extracted from the input image. Among the main challenges that end-to-end approaches face is the performance degradation when recognizing text across scale variations (smaller or larger text), and arbitrary word rotation angles. In this work, we address these challenges by proposing a novel global-to-local attention mechanism for text spotting, termed GLASS, that fuses together global and local features. The global features are extracted from the shared backbone, preserving contextual information from the entire image, while the local features are computed individually on resized, high resolution rotated word crops. The information extracted from the local crops alleviates much of the inherent difficulties with scale and word rotation. We show a performance analysis across scales and angles, highlighting improvement over scale and angle extremities. In addition, we introduce an orientation-aware loss term supervising the detection task, and show its contribution to both detection and recognition performance across all angles. Finally, we show that GLASS is general by incorporating it into other leading text spotting architectures, improving their text spotting performance. Our method achieves state-of-the-art results on multiple benchmarks, including the newly released TextOCR.



Paperid:1158
Authors:Jeonghun Baek; Yusuke Matsui; Kiyoharu Aizawa
Title: COO: Comic Onomatopoeia Dataset for Recognizing Arbitrary or Truncated Texts
Abstract:
Recognizing irregular texts has been a challenging topic in text recognition. To encourage research on this topic, we provide a novel comic onomatopoeia dataset (COO), which consists of onomatopoeia texts in Japanese comics. COO has many arbitrary texts, such as extremely curved, partially shrunk texts, or arbitrarily placed texts. Furthermore, some texts are separated into several parts. Each part is a truncated text and is not meaningful by itself. These parts should be linked to represent the intended meaning. Thus, we propose a novel task that predicts the link between truncated texts. We conduct three tasks to detect the onomatopoeia region and capture its intended meaning: text detection, text recognition, and link prediction. Through extensive experiments, we analyze the characteristics of the COO. Our data and code are available at https://github.com/ku21fan/COO-Comic-Onomatopoeia.



Paperid:1159
Authors:Chuhui Xue; Wenqing Zhang; Yu Hao; Shijian Lu; Philip H. S. Torr; Song Bai
Title: Language Matters: A Weakly Supervised Vision-Language Pre-training Approach for Scene Text Detection and Spotting
Abstract:
Recently, Vision-Language Pre-training (VLP) techniques have greatly benefited various vision-language tasks by jointly learning visual and textual representations, which intuitively helps in Optical Character Recognition (OCR) tasks due to the rich visual and textual information in scene text images. However, these methods cannot well cope with OCR tasks because of the difficulty in both instance-level text encoding and image-text pair acquisition (i.e. images and captured texts in them). This paper presents a weakly supervised pre-training method, oCLIP, which can acquire effective scene text representations by jointly learning and aligning visual and textual information. Our network consists of an image encoder and a character-aware text encoder that extract visual and textual features, respectively, as well as a visual-textual decoder that models the interaction among textual and visual features for learning effective scene text representations. With the learning of textual features, the pre-trained model can attend texts in images well with character awareness. Besides, these designs enable the learning from weakly annotated texts (i.e. partial texts in images without text bounding boxes) which mitigates the data annotation constraint greatly. Experiments over the weakly annotated images in ICDAR2019-LSVT show that our pre-trained model improves F-score by +2.5\% and +4.8\% while transferring its weights to other text detection and spotting networks, respectively. In addition, the proposed method outperforms existing pre-training techniques consistently across multiple public datasets (e.g., +3.2\% and +1.3\% for Total-Text and CTW1500).



Paperid:1160
Authors:Xudong Xie; Ling Fu; Zhifei Zhang; Zhaowen Wang; Xiang Bai
Title: Toward Understanding WordArt: Corner-Guided Transformer for Scene Text Recognition
Abstract:
Artistic text recognition is an extremely challenging task with a wide range of applications. However, current scene text recognition methods mainly focus on irregular text while have not explored artistic text specifically. The challenges of artistic text recognition include the various appearance with special-designed fonts and effects, the complex connections and overlaps between characters, and the severe interference from background patterns. To alleviate these problems, we propose to recognize the artistic text at three levels. Firstly, corner points are applied to guide the extraction of local features inside characters, considering the robustness of corner structures to appearance and shape. In this way, the discreteness of the corner points cuts off the connection between characters, and the sparsity of them improves the robustness for background interference. Secondly, we design a character contrastive loss to model the character-level feature, improving the feature representation for character classification. Thirdly, we utilize Transformer to learn the global feature on image-level and model the global relationship of the corner points, with the assistance of a corner-query cross-attention mechanism. Besides, we provide an artistic text dataset to benchmark the performance. Experimental results verify the significant superiority of our proposed method on artistic text recognition and also achieve state-of-the-art performance on several blurred and perspective datasets.



Paperid:1161
Authors:Cheng Da; Peng Wang; Cong Yao
Title: Levenshtein OCR
Abstract:
A novel scene text recognizer based on Vision-Language Transformer (VLT) is presented. Inspired by Levenshtein Transformer in the area of NLP, the proposed method (named Levenshtein OCR, and LevOCR for short) explores an alternative way for automatically transcribing textual content from cropped natural images. Specifically, we cast the problem of scene text recognition as an iterative sequence refinement process. The initial prediction sequence produced by a pure vision model is encoded and fed into a cross-modal transformer to interact and fuse with the visual features, to progressively approximate the ground truth. The refinement process is accomplished via two basic characterlevel operations: deletion and insertion, which are learned with imitation learning and allow for parallel decoding, dynamic length change and good interpretability. The quantitative experiments clearly demonstrate that LevOCR achieves state-of-the-art performances on standard benchmarks and the qualitative analyses verify the effectiveness and advantage of the proposed LevOCR algorithm.



Paperid:1162
Authors:Peng Wang; Cheng Da; Cong Yao
Title: Multi-Granularity Prediction for Scene Text Recognition
Abstract:
Scene text recognition (STR) has been an active research topic in computer vision for years. To tackle this challenging problem, numerous innovative methods have been successively proposed and incorporating linguistic knowledge into STR models has recently become a prominent trend. In this work, we first draw inspiration from the recent progress in Vision Transformer (ViT) to construct a conceptually simple yet powerful vision STR model, which is built upon ViT and outperforms previous state-of-the-art models for scene text recognition, including both pure vision models and language-augmented methods. To integrate linguistic knowledge, we further propose a Multi-Granularity Prediction strategy to inject information from the language modality into the model in an implicit way, i.e., subword representations (BPE and WordPiece) widely-used in NLP are introduced into the output space, in addition to the conventional character level representation, while no independent language model (LM) is adopted. The resultant algorithm (termed MGP-STR) is able to push the performance envelop of STR to an even higher level. Specifically, it achieves an average recognition accuracy of 93.35% on standard benchmarks.



Paperid:1163
Authors:Ying Chen; Liang Qiao; Zhanzhan Cheng; Shiliang Pu; Yi Niu; Xi Li
Title: Dynamic Low-Resolution Distillation for Cost-Efficient End-to-End Text Spotting
Abstract:
End-to-end text spotting has attached great attention recently due to its benefits on global optimization and high maintainability for real applications. However, the input scale has always been a tough trade-off since recognizing a small text instance usually requires enlarging the whole image, which brings high computational costs. In this paper, to address this problem, we propose a novel cost-efficient Dynamic Low-resolution Distillation (DLD) text spotting framework, which aims to infer images in different small but recognizable resolutions and achieve a better balance between accuracy and efficiency. Concretely, we adopt a resolution selector to dynamically decide the input resolutions for different images, which is constraint by both inference accuracy and computational cost. Another sequential knowledge distillation strategy is conducted on the text recognition branch, making the low-res input obtains comparable performance to a high-res image. The proposed method can be optimized end-to-end and adopted in any current text spotting framework to improve their practicabilities. Extensive experiments on several text spotting benchmarks show that the proposed method vastly improves the usability of low-res models. The code is available.



Paperid:1164
Authors:Chuhui Xue; Jiaxing Huang; Wenqing Zhang; Shijian Lu; Changhu Wang; Song Bai
Title: Contextual Text Block Detection towards Scene Text Understanding
Abstract:
Most existing scene text detectors focus on detecting characters or words that only capture partial text messages due to missing contextual information. For a better understanding of text in scenes, it is more desired to detect contextual text blocks (CTBs) which consist of one or multiple integral text units (e.g., characters, words, or phrases) in natural reading order and transmit certain complete text messages. This paper presents contextual text detection, a new setup that detects CTBs for better understanding of texts in scenes. We formulate the new setup by a dual detection task which first detects integral text units and then groups them into a CTB. To this end, we design a novel scene text clustering technique that treats integral text units as tokens and groups them (belonging to the same CTB) into an ordered token sequence. In addition, we create two datasets SCUT-CTW-Context and ReCTS-Context to facilitate future research, where each CTB is well annotated by an ordered sequence of integral text units. Further, we introduce three metrics that measure contextual text detection in local accuracy, continuity, and global accuracy. Extensive experiments show that our method accurately detects CTBs which effectively facilitates downstream tasks such as text classification and translation. The project is available at \url{https://sg-vilab.github.io/publication/xue2022contextual/}.



Paperid:1165
Authors:Wenqi Zhao; Liangcai Gao
Title: CoMER: Modeling Coverage for Transformer-Based Handwritten Mathematical Expression Recognition
Abstract:
The Transformer-based encoder-decoder architecture has recently made significant advances in recognizing handwritten mathematical expressions. However, the transformer model still suffers from the lack of coverage problem, making its expression recognition rate (ExpRate) inferior to its RNN counterpart. Coverage information, which records the alignment information of the past steps, has proven effective in the RNN models. In this paper, we propose CoMER, a model that adopts the coverage information in the transformer decoder. Specifically, we propose a novel Attention Refinement Module (ARM) to refine the attention weights with past alignment information without hurting its parallelism. Furthermore, we take coverage information to the extreme by proposing self-coverage and cross-coverage, which utilize the past alignment information from the current and previous layers. Experiments show that CoMER improves the ExpRate by 0.61%/2.09%/1.59% compared to the current state-of-the-art model, and reaches 59.33%/59.81%/62.97% on the CROHME 2014/2016/2019 test sets.



Paperid:1166
Authors:Chongyu Liu; Lianwen Jin; Yuliang Liu; Canjie Luo; Bangdong Chen; Fengjun Guo; Kai Ding
Title: Don't Forget Me: Accurate Background Recovery for Text Removal via Modeling Local-Global Context
Abstract:
Text removal has attracted increasingly attention due to its various applications on privacy protection, document restoration, and text editing. It has shown significant progress with deep neural network. However, most of the existing methods often generate inconsistent results for complex background. To address this issue, we propose a Contextual-guided Text Removal Network, termed as CTRNet. CTRNet explores both low-level structure and high-level discriminative context feature as prior knowledge to guide the process of background restoration. We further propose a Local-global Content Modeling (LGCM) block with CNNs and Transformer-Encoder to capture local features and establish the long-term relationship among pixels globally. Finally, we incorporate LGCM with context guidance for feature modeling and decoding. Experiments on benchmark datasets, SCUT-EnsText and SCUT-Syn show that CTRNet significantly outperforms the existing state-of-the-art methods. Furthermore, a qualitative experiment on examination papers also demonstrates the generalization ability of our method. The code of CTRNet is available at https://github.com/lcy0604/CTRNet.



Paperid:1167
Authors:Oren Nuriel; Sharon Fogel; Ron Litman
Title: TextAdaIN: Paying Attention to Shortcut Learning in Text Recognizers
Abstract:
Leveraging the characteristics of convolutional layers, neural networks are extremely effective for pattern recognition tasks. However in some cases, their decisions are based on unintended information leading to high performance on standard benchmarks but also to a lack of generalization to challenging testing conditions and unintuitive failures. Recent work has termed this “shortcut learning” and addressed its presence in multiple domains. In text recognition, we reveal another such shortcut, whereby recognizers overly depend on local image statistics. Motivated by this, we suggest an approach to regulate the reliance on local statistics that improves text recognition performance. Our method, termed TextAdaIN, creates local distortions in the feature map which prevent the network from overfitting to local statistics. It does so by viewing each feature map as a sequence of elements and deliberately mismatching fine-grained feature statistics between elements in a mini-batch. DespiteTextAdaIN’s simplicity, extensive experiments show its effectiveness compared to other, more complicated methods. TextAdaIN achieves state-of-the-art results on standard handwritten text recognition benchmarks. It generalizes to multiple architectures and to the domain of scene text recognition. Furthermore, we demonstrate that integrating TextAdaIN improves robustness towards more challenging testing conditions.



Paperid:1168
Authors:Byeonghu Na; Yoonsik Kim; Sungrae Park
Title: Multi-modal Text Recognition Networks: Interactive Enhancements between Visual and Semantic Features
Abstract:
Linguistic knowledge has brought great benefits to scene text recognition by providing semantics to refine character sequences. However, since linguistic knowledge has been applied individually on the output sequence, previous methods have not fully utilized the semantics to understand visual clues for text recognition. This paper introduces a novel method, called Multi-modAl Text Recognition Network (MATRN), that enables interactions between visual and semantic features for better recognition performances. Specifically, MATRN identifies visual and semantic feature pairs and encodes spatial information into semantic features. Based on the spatial encoding, visual and semantic features are enhanced by referring to related features in the other modality. Furthermore, MATRN stimulates combining semantic features into visual features by hiding visual clues related to the character in the training phase. Our experiments demonstrate that MATRN achieves state-of-the-art performances on seven benchmarks with large margins, while naive combinations of two modalities show less-effective improvements. Further ablative studies prove the effectiveness of our proposed components. Our implementation is available at https://github.com/wp03052/MATRN.



Paperid:1169
Authors:Dajian Zhong; Shujing Lyu; Palaiahnakote Shivakumara; Bing Yin; Jiajia Wu; Umapada Pal; Yue Lu
Title: SGBANet: Semantic GAN and Balanced Attention Network for Arbitrarily Oriented Scene Text Recognition
Abstract:
Scene text recognition is a challenging task due to the complex backgrounds and diverse variations of text instances. In this paper, we propose a novel Semantic GAN and Balanced Attention Network (SGBANet) to recognize the texts in scene images. The proposed method first generates the simple semantic feature using Semantic GAN and then recognizes the scene text with the Balanced Attention Module. The Semantic GAN aims to align the semantic feature distribution between the support domain and target domain. Different from the conventional image-to-image translation methods that perform at the image level, the Semantic GAN performs the generation and discrimination on the semantic level with the Semantic Generator Module (SGM) and Semantic Discriminator Module (SDM). For target images (scene text images), the Semantic Generator Module generates simple semantic features that share the same feature distribution with support images (clear text images). The Semantic Discriminator Module is used to distinguish the semantic features between the support domain and target domain. In addition, a Balanced Attention Module is designed to alleviate the problem of attention drift. The Balanced Attention Module first learns a balancing parameter based on the visual glimpse vector and semantic glimpse vector, and then performs the balancing operation for obtaining a balanced glimpse vector. Experiments on six benchmarks, including regular datasets, i.e., IIIT5K, SVT, ICDAR2013, and irregular datasets, i.e., ICDAR2015, SVTP, CUTE80, validate the effectiveness of our proposed method.



Paperid:1170
Authors:Yew Lee Tan; Adams Wai-Kin Kong; Jung-Jae Kim
Title: Pure Transformer with Integrated Experts for Scene Text Recognition
Abstract:
Scene text recognition (STR) involves the task of reading text in cropped images of natural scenes. Conventional models in STR employ convolutional neural network (CNN) followed by recurrent neural network in an encoder-decoder framework. In recent times, the transformer architecture is being widely adopted in STR as it shows strong capability in capturing long-term dependency which appears to be prominent in scene text images. Many researchers utilized transformer as part of a hybrid CNN-transformer encoder, often followed by a transformer decoder. However, such methods only make use of the long-term dependency mid-way through the encoding process. Although the vision transformer (ViT) is able to capture such dependency at an early stage, its utilization remains largely unexploited in STR. This work proposes the use of a transformer-only model as a simple baseline which outperforms hybrid CNN-transformer models. Furthermore, two key areas for improvement were identified. Firstly, the first decoded character has the lowest prediction accuracy. Secondly, images of different original aspect ratios react differently to the patch resolutions while ViT only employ one fixed patch resolution. To explore these areas, Pure Transformer with Integrated Experts (PTIE) is proposed. PTIE is a transformer model that can process multiple patch resolutions and decode in both the original and reverse character orders. It is examined on 7 commonly used benchmarks and compared with over 20 state-of-the-art methods. The experimental results show that the proposed method outperforms them and obtains state-of-the-art results in most benchmarks.



Paperid:1171
Authors:Geewook Kim; Teakgyu Hong; Moonbin Yim; JeongYeon Nam; Jinyoung Park; Jinyeong Yim; Wonseok Hwang; Sangdoo Yun; Dongyoon Han; Seunghyun Park
Title: OCR-Free Document Understanding Transformer
Abstract:
Understanding document images (e.g., invoices) is a core but challenging task since it requires complex functions such as reading text and a holistic understanding of the document. Current Visual Document Understanding (VDU) methods outsource the task of reading text to off-the-shelf Optical Character Recognition (OCR) engines and focus on the understanding task with the OCR outputs. Although such OCR-based approaches have shown promising performance, they suffer from 1) high computational costs for using OCR; 2) inflexibility of OCR models on languages or types of documents; 3) OCR error propagation to the subsequent process. To address these issues, in this paper, we introduce a novel OCR-free VDU model named Donut, which stands for Document understanding transformer. As the first step in OCR-free VDU research, we propose a simple architecture (i.e., Transformer) with a pre-training objective (i.e., cross-entropy loss). Donut is conceptually simple yet effective. Through extensive experiments and analyses, we show a simple OCR-free VDU model, Donut, achieves state-of-the-art performances on various VDU tasks in terms of both speed and accuracy. In addition, we offer a synthetic data generator that helps the model pre-training to be flexible in various languages and domains. The code, trained model, and synthetic data are available at https://github.com/clovaai/donut.



Paperid:1172
Authors:Ye Huang; Di Kang; Liang Chen; Xuefei Zhe; Wenjing Jia; Linchao Bao; Xiangjian He
Title: CAR: Class-Aware Regularizations for Semantic Segmentation
Abstract:
Recent segmentation methods, such as OCR and CPNet, utilizing “class level” information in addition to pixel features, have achieved notable success for boosting the accuracy of existing network modules. However, the extracted class-level information was simply concatenated to pixel features, without explicitly being exploited for better pixel representation learning. Moreover, these approaches learn soft class centers based on coarse mask prediction, which is prone to error accumulation. In this paper, aiming to use class level information more effectively, we propose a universal Class-Aware Regularization (CAR) approach to optimize the intra-class variance and inter-class distance during feature learning, motivated by the fact that humans can recognize an object by itself no matter which other objects it appears with. Three novel loss functions are proposed. The first loss function encourages more compact class representations within each class, the second directly maximizes the distance between different class centers, and the third further pushes the distance between inter-class centers and pixels. Furthermore, the class center in our approach is directly generated from ground truth instead of from the error-prone coarse prediction. Our method can be easily applied to most existing segmentation models during training, including OCR and CPNet, and can largely improve their accuracy at no additional inference overhead. Extensive experiments and ablation studies conducted on multiple benchmark datasets demonstrate that the proposed CAR can boost the accuracy of all baseline models by up to 2.23% mIOU with superior generalization ability. The complete code is available at https://github.com/edwardyehuang/CAR.



Paperid:1173
Authors:Yuyang Zhao; Zhun Zhong; Na Zhao; Nicu Sebe; Gim Hee Lee
Title: Style-Hallucinated Dual Consistency Learning for Domain Generalized Semantic Segmentation
Abstract:
In this paper, we study the task of synthetic-to-real domain generalized semantic segmentation, which aims to learn a model that is robust to unseen real-world scenes using only synthetic data. The large domain shift between synthetic and real-world data, including the limited source environmental variations and the large distribution gap between synthetic and real-world data, significantly hinders the model performance on unseen real-world scenes. In this work, we propose the Style-HAllucinated Dual consistEncy learning (SHADE) framework to handle such domain shift. Specifically, SHADE is constructed based on two consistency constraints, Style Consistency (SC) and Retrospection Consistency (RC). SC enriches the source situations and encourages the model to learn consistent representation across style-diversified samples. RC leverages real-world knowledge to prevent the model from overfitting to synthetic data and thus largely keeps the representation consistent between the synthetic and real-world models. Furthermore, we present a novel style hallucination module (SHM) to generate style-diversified samples that are essential to consistency learning. SHM selects basis styles from the source distribution, enabling the model to dynamically generate diverse and realistic samples during training. Experiments show that our SHADE yields significant improvement and outperforms state-of-the-art methods by 5.05% and 8.35% on the average mIoU of three real-world datasets on single- and multi-source settings, respectively.



Paperid:1174
Authors:Junfeng Wu; Yi Jiang; Song Bai; Wenqing Zhang; Xiang Bai
Title: SeqFormer: Sequential Transformer for Video Instance Segmentation
Abstract:
In this work, we present SeqFormer for video instance segmentation. SeqFormer follows the principle of vision transformer that models instance relationships among video frames. Nevertheless, we observe that a stand-alone instance query suffices for capturing a time sequence of instances in a video, but attention mechanisms shall be done with each frame independently. To achieve this, SeqFormer locates an instance in each frame and aggregates temporal information to learn a powerful representation of a video-level instance, which is used to predict the mask sequences on each frame dynamically. Instance tracking is achieved naturally without tracking branches or post-processing. On YouTube-VIS, SeqFormer achieves 47.4 AP with a ResNet-50 backbone and 49.0 AP with a ResNet-101 backbone without bells and whistles. Such achievement significantly exceeds the previous state-of-the-art performance by 4.6 and 4.4, respectively. In addition, integrated with the recently-proposed Swin transformer, SeqFormer achieves a much higher AP of 59.3. We hope SeqFormer could be a strong baseline that fosters future research in video instance segmentation, and in the meantime, advances this field with a more robust, accurate, neat model. The code is available at https://github.com/wjf5203/SeqFormer.



Paperid:1175
Authors:Wenhu Zhang; Liangli Zheng; Huanyu Wang; Xintian Wu; Xi Li
Title: Saliency Hierarchy Modeling via Generative Kernels for Salient Object Detection
Abstract:
Salient Object Detection (SOD) is a challenging problem that aims to precisely recognize and segment the salient objects. In ground-truth maps, all pixels belonging to the salient objects are positively annotated with the same value. However, the saliency level should be a relative quantity, which varies among different regions in a given sample and different samples. The conflict between various saliency levels and single saliency value in ground-truth, results in learning difficulty. To alleviate the problem, we propose a Saliency Hierarchy Network (SHNet), modeling saliency patterns via generative kernels from two perspectives: region-level and sample-level. Specifically, we construct a Saliency Hierarchy Module to explicitly model saliency levels of different regions in a given sample with the guide of prior knowledge. Moreover, considering the sample-level divergence, we introduce a Hyper Kernel Generator to capture the global contexts and adaptively generate convolution kernels for various inputs. As a result, extensive experiments on five standard benchmarks demonstrate our SHNet outperforms other state-of-the-art methods in both terms of performance and efficiency.



Paperid:1176
Authors:Junfeng Wu; Qihao Liu; Yi Jiang; Song Bai; Alan Yuille; Xiang Bai
Title: In Defense of Online Models for Video Instance Segmentation
Abstract:
In recent years, video instance segmentation (VIS) has been largely advanced by offline models, while online models gradually attracted less attention possibly due to their inferior performance. However, online methods have their inherent advantage in handling long video sequences and ongoing videos while offline models fail due to the limit of computational resources. Therefore, it would be highly desirable if online models can achieve comparable or even better performance than offline models. By dissecting current online models and offline models, we demonstrate that the main cause of the performance gap is the error-prone association between frames caused by the similar appearance among different instances in the feature space. Observing this, we propose an online framework based on contrastive learning that is able to learn more discriminative instance embeddings for association and fully exploit history information for stability. Despite its simplicity, our method outperforms all online and offline methods on three benchmarks. Specifically, we achieve 49.5 AP on YouTube-VIS 2019, a significant improvement of 13.2 AP and 2.1 AP over the prior online and offline art, respectively. Moreover, we achieve 30.2 AP on OVIS, a more challenging dataset with significant crowding and occlusions, surpassing the prior art by 14.8 AP. The proposed method won first place in the video instance segmentation track of the 4th Large-scale Video Object Segmentation Challenge (CVPR2022). We hope the simplicity and effectiveness of our method, as well as our insight on current methods, could shed light on the exploration of VIS models. The code is available at https://github.com/wjf5203/VNext.



Paperid:1177
Authors:Chufeng Tang; Lingxi Xie; Gang Zhang; Xiaopeng Zhang; Qi Tian; Xiaolin Hu
Title: Active Pointly-Supervised Instance Segmentation
Abstract:
The requirement of expensive annotations is a major burden for training a well-performed instance segmentation model. In this paper, we present an economic active learning setting, named active pointly-supervised instance segmentation (APIS), which starts with box-level annotations and iteratively samples a point within the box and asks if it falls on the object. The key of APIS is to find the most desirable points to maximize the segmentation accuracy with limited annotation budgets. We formulate this setting and propose several uncertainty-based sampling strategies. The model developed with these strategies yields consistent performance gain on the challenging MS-COCO dataset, compared against other learning strategies. The results suggest that APIS, integrating the advantages of active learning and point-based supervision, is an effective learning paradigm for label-efficient instance segmentation.



Paperid:1178
Authors:Bowen Shi; Dongsheng Jiang; Xiaopeng Zhang; Han Li; Wenrui Dai; Junni Zou; Hongkai Xiong; Qi Tian
Title: A Transformer-Based Decoder for Semantic Segmentation with Multi-level Context Mining
Abstract:
Transformers have recently shown superior performance than CNN on semantic segmentation. However, previous works mostly focus on the deliberate design of the encoder, while seldom considering the decoder part. In this paper, we find that a light weighted decoder counts for segmentation, and propose a pure transformer-based segmentation decoder, named SegDeformer, to seamlessly incorporate into current varied transformer-based encoders. The highlight is that SegDeformer is able to conveniently utilize the tokenized input and the attention mechanism of the transformer for effective context mining. This is achieved by two key component designs, i.e., the internal and external context mining modules. The former is equipped with internal attention within an image to better capture global-local context, while the latter introduces external tokens from other images to enhance current representation. To enable SegDeformer in a scalable way, we further provide performance/efficiency optimization modules for flexible deployment. Experiments on widely used benchmarks ADE20K, COCO-Stuff and Cityscapes and different transformer encoders (e.g., ViT, MiT and Swin) demonstrate that SegDeformer can bring consistent performance gains.



Paperid:1179
Authors:Ho Kei Cheng; Alexander G. Schwing
Title: XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model
Abstract:
We present XMem, a video object segmentation architecture for long videos with unified feature memory stores inspired by the Atkinson-Shiffrin memory model. Prior work on video object segmentation typically only uses one type of feature memory. For videos longer than a minute, a single feature memory model tightly links memory consumption and accuracy. In contrast, following the Atkinson-Shiffrin model, we develop an architecture that incorporates multiple independent yet deeply-connected feature memory stores: a rapidly updated sensory memory, a high-resolution working memory, and a compact thus sustained long-term memory. Crucially, we develop a memory potentiation algorithm that routinely consolidates actively used working memory elements into the long-term memory, which avoids memory explosion and minimizes performance decay for long-term prediction. Combined with a new memory reading mechanism, XMem greatly exceeds state-of-the-art performance on long-video datasets while being on par with state-of-the-art methods (that do not work on long videos) on short-video datasets. Code is available at https://hkchengrex.github.io/XMem"



Paperid:1180
Authors:Jiale Li; Hang Dai; Yong Ding
Title: Self-Distillation for Robust LiDAR Semantic Segmentation in Autonomous Driving
Abstract:
We propose a new and effective self-distillation framework with our new Test-Time Augmentation (TTA) and Transformer based Voxel Feature Encoder (TransVFE) for robust LiDAR semantic segmentation in autonomous driving, where the robustness is mission-critical but usually neglected. The proposed framework enables the knowledge to be distilled from a teacher model instance to a student model instance, while the two model instances are with the same network architecture for jointly learning and evolving. This requires a strong teacher model to evolve in training. Our TTA strategy effectively reduces the uncertainty in the inference stage of the teacher model. Thus, we propose to equip the teacher model with TTA for providing privileged guidance while the student continuously updates the teacher with better network parameters learned by itself. To further enhance the teacher model, we propose a TransVFE to improve the point cloud encoding by modeling and preserving the local relationship among the points inside each voxel via multi-head attention. The proposed modules are generally designed to be instantiated with different backbones. Evaluations on SemanticKITTI and nuScenes datasets show that our method achieves state-of-the-art performance. Our code will be made publicly available.



Paperid:1181
Authors:Xu Yan; Jiantao Gao; Chaoda Zheng; Chao Zheng; Ruimao Zhang; Shuguang Cui; Zhen Li
Title: 2DPASS: 2D Priors Assisted Semantic Segmentation on LiDAR Point Clouds
Abstract:
As camera and LiDAR sensors capture complementary information used in autonomous driving, great efforts have been made to develop semantic segmentation algorithms through multi-modality data fusion. However, fusion-based approaches require paired data, i.e., LiDAR point clouds and camera images with strict point-to-pixel mappings, as the inputs in both training and inference, which seriously hinders their application in practical scenarios. Thus, in this work, we propose the 2D Priors Assisted Semantic Segmentation (2DPASS), a general training scheme, to boost the representation learning on point clouds, by fully taking advantage of 2D images with rich appearance. In practice, by leveraging an auxiliary modal fusion and multi-scale fusion-to-single knowledge distillation (MSFSKD), 2DPASS acquires richer semantic and structural information from the multi-modal data, which are then online distilled to the pure 3D network. As a result, equipped with 2DPASS, our baseline shows significant improvement with only point cloud inputs. Specifically, it achieves the state-of-the-arts on two large-scale benchmarks (i.e. SemanticKITTI and NuScenes), including top-1 results in both single and multiple scan(s) competitions of SemanticKITTI.



Paperid:1182
Authors:Chong Zhou; Chen Change Loy; Bo Dai
Title: Extract Free Dense Labels from CLIP
Abstract:
Contrastive Language-Image Pre-training (CLIP) has made a remarkable breakthrough in open-vocabulary zero-shot image recognition. Many recent studies leverage the pre-trained CLIP models for image-level classification and manipulation. In this paper, we wish examine the intrinsic potential of CLIP for pixel-level dense prediction, specifically in semantic segmentation. To this end, with minimal modification, we show that MaskCLIP yields compelling segmentation results on open concepts across various datasets in the absence of annotations and fine-tuning. By adding pseudo labeling and self-training, MaskCLIP+ surpasses SOTA transductive zero-shot semantic segmentation methods by large margins, e.g., mIoUs of unseen classes on PASCAL VOC/PASCAL Context/COCO Stuff are improved from 35.6/20.7/30.3 to 86.1/66.7/54.7. We also test the robustness of MaskCLIP under input corruption and evaluate its capability in discriminating fine-grained objects and novel concepts. Our finding suggests that MaskCLIP can serve as a new reliable source of supervision for dense prediction tasks to achieve annotation-free segmentation. Source code is available at https://github.com/chongzhou96/MaskCLIP.



Paperid:1183
Authors:Muhammad Ferjad Naeem; Evin Pınar Örnek; Yongqin Xian; Luc Van Gool; Federico Tombari
Title: 3D Compositional Zero-Shot Learning with DeCompositional Consensus
Abstract:
Parts represent a basic unit of geometric and semantic similarity across different objects. We argue that part knowledge should be composable beyond the observed object classes. Towards this, we present 3D Compositional Zero-shot Learning as a problem of part generalization from seen to unseen object classes for semantic segmentation. We provide a structured study through benchmarking the task with the proposed Compositional-PartNet dataset. This dataset is created by processing the original PartNet to maximize part overlap across different objects. The existing point cloud part segmentation methods fail to generalize to unseen object classes in this setting. As a solution, we propose DeCompositional Consensus, which combines a part segmentation network with a part scoring network. The key intuition to our approach is that a segmentation mask over some parts should have a consensus with its part scores when each part is taken apart. The two networks reason over different part combinations defined in a per-object part prior to generate the most suitable segmentation mask. We demonstrate that our method allows compositional zero-shot segmentation and generalized zero-shot classification and establishes the state of the art on both tasks.



Paperid:1184
Authors:Lei Ke; Henghui Ding; Martin Danelljan; Yu-Wing Tai; Chi-Keung Tang; Fisher Yu
Title: Video Mask Transfiner for High-Quality Video Instance Segmentation
Abstract:
While Video Instance Segmentation (VIS) has seen rapid progress, current approaches struggle to predict high-quality masks with accurate boundary details. Moreover, the predicted segmentations often fluctuate over time, suggesting that temporal consistency cues are neglected or not fully utilized. In this paper, we set out to tackle these issues, with the aim of achieving highly detailed and more temporally stable mask predictions for VIS. We first propose the Video Mask Transfiner (VMT) method, capable of leveraging fine-grained high-resolution features thanks to a highly efficient video transformer structure. Our VMT detects and groups sparse error-prone spatio-temporal regions of each tracklet in the video segment, which are then refined using both local and instance-level cues. Second, we identify that the coarse boundary annotations of the popular YouTube-VIS dataset constitute a major limiting factor. Based on our VMT architecture, we therefore design an automated annotation refinement approach by iterative training and self-correction. To benchmark high-quality mask predictions for VIS, we introduce the HQ-YTVIS dataset, consisting of a manually re-annotated test set and our automatically refined training data. We compare VMT with the most recent state-of-the-art methods on the HQ-YTVIS, as well as the Youtube-VIS, OVIS and BDD100K MOTS benchmarks. Experimental results clearly demonstrate the efficacy and effectiveness of our method on segmenting complex and dynamic objects, by capturing precise details.



Paperid:1185
Authors:Wentong Li; Wenyu Liu; Jianke Zhu; Miaomiao Cui; Xian-Sheng Hua; Lei Zhang
Title: Box-Supervised Instance Segmentation with Level Set Evolution
Abstract:
In contrast to the fully supervised methods using pixel-wise mask labels, box-supervised instance segmentation takes advantage of the simple box annotations, which has recently attracted a lot of research attentions. In this paper, we propose a novel single-shot box-supervised instance segmentation approach, which integrates the classical level set model with deep neural network delicately. Specifically, our proposed method iteratively learns a series of level sets through a continuous Chan-Vese energy-based function in an end-to-end fashion. A simple mask supervised SOLOv2 model is adapted to predict the instance-aware mask map as the level set for each instance. Both the input image and its deep features are employed as the input data to evolve the level set curves, where a box projection function is employed to obtain the initial boundary. By minimizing the fully differentiable energy function, the level set for each instance is iteratively optimized within its corresponding bounding box annotation. The experimental results on four challenging benchmarks demonstrate the leading performance of our proposed approach to robust instance segmentation in various scenarios. The code is available at: https://github.com/LiWentomng/boxlevelset.



Paperid:1186
Authors:Hao Wen; Yunze Liu; Jingwei Huang; Bo Duan; Li Yi
Title: Point Primitive Transformer for Long-Term 4D Point Cloud Video Understanding
Abstract:
This paper proposes a 4D backbone for long-term point cloud video understanding. A typical way to capture spatial-temporal context is using 4Dconv or transformer without hierarchy. However, those methods are neither effective nor efficient enough due to camera motion, scene changes, sampling patterns, and complexity of 4D data. To address those issues, we leverage the primitive plane as mid-level representation to capture the long-term spatial-temporal context in 4D point cloud videos, and propose a novel hierarchical backbone named Point Primitive Transformer(PPTr), which is mainly composed of intra-primitive point transformers and primitive transformers. Extensive experiments show that PPTr outperforms the previous state of the arts on different tasks.



Paperid:1187
Authors:Yuan Wang; Rui Sun; Zhe Zhang; Tianzhu Zhang
Title: Adaptive Agent Transformer for Few-Shot Segmentation
Abstract:
Few-shot segmentation (FSS) aims to segment objects in a given query image with only a few labelled support images. The limited support information makes it an extremely challenging task. Most previous best-performing methods adopt prototypical learning or affinity learning. Nevertheless, they either neglect to further utilize support pixels for facilitating segmentation and lose spatial information, or are not robust to noisy pixels and computationally expensive. In this work, we propose a novel end-to-end adaptive agent transformer (AAFormer) to integrate prototypical and affinity learning to exploit the complementarity between them via a transformer encoder-decoder architecture, including a representation encoder, an agent learning decoder and an agent matching decoder. The proposed AAFormer enjoys several merits. First, to learn agent tokens well without any explicit supervision, and to make agent tokens capable of dividing different objects into diverse parts in an adaptive manner, we customize the agent learning decoder according to the three characteristics of context awareness, spatial awareness and diversity. Second, the proposed agent matching decoder is responsible for decomposing the direct pixel-level matching matrix into two more computationally-friendly matrices to suppress the noisy pixels. Extensive experimental results on two standard benchmarks demonstrate that our AAFormer performs favorably against state-of-the-art FSS methods.



Paperid:1188
Authors:Jieru Mei; Alex Zihao Zhu; Xinchen Yan; Hang Yan; Siyuan Qiao; Yukun Zhu; Liang-Chieh Chen; Henrik Kretzschmar
Title: Waymo Open Dataset: Panoramic Video Panoptic Segmentation
Abstract:
Panoptic image segmentation is the computer vision task of finding groups of pixels in an image and assigning semantic classes and object instance identifiers to them. Research in image segmentation has become increasingly popular due to its critical applications in robotics and autonomous driving. The research community thereby relies on publicly available benchmark dataset to advance the state-of-the-art in computer vision. Due to the high costs of densely labeling the images, however, there is a shortage of publicly available ground truth labels that are suitable for panoptic segmentation. The high labeling costs also make it challenging to extend existing datasets to the video domain and to multi-camera setups. We therefore present the Waymo Open Dataset: Panoramic Video Panoptic Segmentation dataset, a large-scale dataset that offers high-quality panoptic segmentation labels for autonomous driving. We generate our dataset using the publicly available Waymo Open Dataset (WOD), leveraging the diverse set of camera images. Our labels are consistent over time for video processing and consistent across multiple cameras mounted on the vehicles for full panoramic scene understanding. Specifically, we offer labels for 28 semantic categories and 2,860 temporal sequences that were captured by five cameras mounted on autonomous vehicles driving in three different geographical locations, leading to a total of 100k labeled camera images. To the best of our knowledge, this makes our dataset an order of magnitude larger than existing datasets that offer video panoptic segmentation labels. We further propose a new benchmark for Panoramic Video Panoptic Segmentation and establish a number of strong baselines based on the DeepLab family of models. We have made the benchmark and the code publicly available, which we hope will facilitate future research on holistic scene understanding. Our dataset can be found at: https://waymo.com/open/"



Paperid:1189
Authors:Zhaoyuan Yin; Pichao Wang; Fan Wang; Xianzhe Xu; Hanling Zhang; Hao Li; Rong Jin
Title: TransFGU: A Top-down Approach to Fine-Grained Unsupervised Semantic Segmentation
Abstract:
Unsupervised semantic segmentation aims to obtain high-level semantic representation on low-level visual features without manual annotations. Most existing methods are bottom-up approaches that try to group pixels into regions based on their visual cues or certain predefined rules. As a result, it is difficult for these bottom-up approaches to generate fine-grained semantic segmentation when coming to complicated scenes with multiple objects and some objects sharing similar visual appearance. In contrast, we propose the first top-down unsupervised semantic segmentation framework for fine-grained segmentation in extremely complicated scenarios. Specifically, we first obtain rich high-level structured semantic concept information from large-scale vision data in a self-supervised learning manner, and use such information as a prior to discover potential semantic categories presented in target datasets. Secondly, the discovered high-level semantic categories are mapped to low-level pixel features by calculating the class activate map (CAM) with respect to certain discovered semantic representation. Lastly, the obtained CAMs serve as pseudo labels to train the segmentation module and produce final semantic segmentation. Experimental results on multiple semantic segmentation benchmarks show that our top-down unsupervised segmentation is robust to both object-centric and scene-centric datasets under different settings of semantic granularity level, and outperforms all the current state-of-the-art bottom-up methods.



Paperid:1190
Authors:Yian Wang; Ruihai Wu; Kaichun Mo; Jiaqi Ke; Qingnan Fan; Leonidas J. Guibas; Hao Dong
Title: AdaAfford: Learning to Adapt Manipulation Affordance for 3D Articulated Objects via Few-Shot Interactions
Abstract:
Perceiving and interacting with 3D articulated objects, such as cabinets, doors, and faucets, pose particular challenges for future home-assistant robots performing daily tasks in human environments. Besides parsing the articulated parts and joint parameters, researchers recently advocate learning manipulation affordance over the input shape geometry which is more task-aware and geometrically fine-grained. However, taking only passive observations as inputs, these methods ignore many hidden but important kinematic constraints (e.g., joint location and limits) and dynamic factors (e.g., joint friction and restitution), therefore losing significant accuracy for test cases with such uncertainties. In this paper, we propose a novel framework, named AdaAfford, that learns to perform very few test-time interactions for quickly adapting the affordance priors to more accurate instance-specific posteriors. We conduct large-scale experiments using the PartNet-Mobility dataset and prove that our system performs better than baselines. We will release our code and data upon paper acceptance.



Paperid:1191
Authors:Sunghwan Hong; Seokju Cho; Jisu Nam; Stephen Lin; Seungryong Kim
Title: Cost Aggregation with 4D Convolutional Swin Transformer for Few-Shot Segmentation
Abstract:
We present a novel cost aggregation network, called Volumetric Aggregation with Transformers (VAT), for few-shot segmentation. The use of transformers can benefit correlation map aggregation through self-attention over a global receptive field. However, the tokenization of a correlation map for transformer processing can be detrimental, because the discontinuity at token boundaries reduces the local context available near the token edges and decreases inductive bias. To address this problem, we propose a 4D Convolutional Swin Transformer, where a high-dimensional Swin Transformer is preceded by a series of small-kernel convolutions that impart local context to all pixels and introduce convolutional inductive bias. We additionally boost aggregation performance by applying transformers within a pyramidal structure, where aggregation at a coarser level guides aggregation at a finer level. Error in the transformer output is then filtered in the subsequent decoder with the help of the query’s appearance embedding. With this model, a new state-of-the-art is set for all the standard benchmarks in few-shot segmentation. It is shown that VAT attains state-of-the-art performance for semantic correspondence as well, where cost aggregation also plays a central role.



Paperid:1192
Authors:Lingzhi Zhang; Shenghao Zhou; Simon Stent; Jianbo Shi
Title: "Fine-Grained Egocentric Hand-Object Segmentation: Dataset, Model, and Applications"
Abstract:
Egocentric videos offer fine-grain information for high-fidelity modeling of human behaviors. Hands and interacting objects are one crucial aspect of understanding viewer’s behaviors and intentions. We provide a labeled dataset consisting of 11,235 egocentric images with per-pixel segmentation labels of hands and the interacting objects in diverse daily activities. Our dataset is the first to label detailed interacting hand-object contact boundaries. We introduce a context-aware compositional data augmentation technique to adapt to out-of-the-distribution YouTube egocentric video. We show that our robust hand-object segmentation model and dataset can serve as a foundation tool to boost or enable several downstream vision applications, such as: Hand state classification, video activity recognition, 3D mesh reconstruction of hand-object interaction, and Seeing through the hand with video inpainting in egocentric videos. All of our data and code will be released to the public.



Paperid:1193
Authors:Lingzhi Zhang; Yuqian Zhou; Connelly Barnes; Sohrab Amirghodsi; Zhe Lin; Eli Shechtman; Jianbo Shi
Title: Perceptual Artifacts Localization for Inpainting
Abstract:
Image inpainting is an essential task for multiple practical applications like object removal and image editing. Deep GAN-based models greatly improve the inpainting performance in structures and textures within the hole, but might also generate unexpected artifacts like broken structures or color blobs. Users perceive these artifacts to judge the effectiveness of inpainting models, and retouch these imperfect areas to inpaint again in a typical retouching workflow. Inspired by this workflow, we propose a new learning task of automatic segmentation of inpainting perceptual artifacts, and apply the model for inpainting model evaluation and iterative refinement. Specifically, we first construct a new inpainting artifacts dataset by manually annotating perceptual artifacts in the results of state-of-the-art inpainting models. Then we train advanced segmentation networks on this dataset to reliably localize inpainting artifacts within inpainted images. Second, we propose a new interpretable evaluation metric called Perceptual Artifact Ratio (PAR), which is the ratio of objectionable inpainted regions to the entire inpainted area. PAR demonstrates a strong correlation with real user preference. Finally, we further apply the generated masks for iterative image inpainting by combining our approach with multiple recent inpainting methods. Extensive experiments demonstrate the consistent decrease of artifact regions and inpainting quality improvement across the different methods.



Paperid:1194
Authors:Zhixuan Li; Weining Ye; Tingting Jiang; Tiejun Huang
Title: 2D Amodal Instance Segmentation Guided by 3D Shape Prior
Abstract:
Amodal instance segmentation aims to predict the complete mask of the occluded instance, including both visible and invisible regions. Existing 2D AIS methods learn and predict the complete silhouettes of target instances in 2D space. However, masks in 2D space are only some observations and samples from the 3D model in different viewpoints and thus can not represent the real complete physical shape of the instances. With the 2D masks learned, 2D amodal methods are hard to generalize to new viewpoints not included in the training dataset. To tackle these problems, we are motivated by observations that (1) a 2D amodal mask is the projection of a 3D complete model, and (2) the 3D complete model can be recovered and reconstructed from the occluded 2D object instances. This paper builds a bridge to link the 2D occluded instances with the 3D complete models by 3D reconstruction and utilizes 3D shape prior for 2D AIS. To deal with the diversity of 3D shapes, our method is pretrained on large 3D reconstruction datasets for high-quality results. And we adopt the unsupervised 3D reconstruction method to avoid relying on 3D annotations. In this approach, our method can reconstruct 3D models from occluded 2D object instances and generalize to new unseen 2D viewpoints of the 3D object. Experiments demonstrate that our method outperforms all existing 2D AIS methods. Our code will be released.



Paperid:1195
Authors:Ping-Chung Yu; Cheng Sun; Min Sun
Title: Data Efficient 3D Learner via Knowledge Transferred from 2D Model
Abstract:
Collecting and labeling 3D data is costly. As a result, 3D resources for training are typically limited in quantity compared to the 2D images counterpart. In this work, we deal with the data scarcity challenge of 3D tasks by transferring knowledge from strong 2D models via abundant RGB-D images. Specifically, we utilize an strong and well-trained semantic segmentation model for 2D images to augment abundant RGB-D images with pseudo-label. The augmented dataset can then be used to pre-train 3D models. Finally, by simply fine-tuning on a few labeled 3D instances, our method already outperforms existing state-of-the-art that is tailored for 3D label efficiency. We further improve our results by using simple semi-supervised techniques ({\it i.e.}, mean teacher and entropy minimization). We verify the effectiveness of our pre-training on two popular 3D models on three different tasks. On ScanNet official evaluation, we establish new state-of-the-art semantic segmentation results on the data-efficient track.



Paperid:1196
Authors:Tong Wu; Guangyu Gao; Junshi Huang; Xiaolin Wei; Xiaoming Wei; Chi Harold Liu
Title: Adaptive Spatial-BCE Loss for Weakly Supervised Semantic Segmentation
Abstract:
For Weakly-Supervised Semantic Segmentation (WSSS) with image-level annotation, mostly relies on the classification network to generate initial segmentation pseudo-labels. However, the optimization target of classification networks usually neglects the discrimination between different pixels, like insignificant foreground and background regions. In this paper, we propose an adaptive Spatial Binary Cross-Entropy (Spatial-BCE) Loss for WSSS, which aims to enhance the discrimination between pixels. In Spatial-BCE Loss, we calculate the loss independently for each pixel and heuristically assign the optimization directions for foreground and background pixels separately. An auxiliary self-supervised task is also proposed to guarantee the Spatial-BCE Loss working as envisaged. Meanwhile, to enhance the network’s generalization for different data distributions, we design an alternate training strategy to adaptively generate thresholds to divide the foreground and background. Benefiting from high-quality initial pseudo-labels by Spatial-BCE Loss, our method also reduce the reliance on post-processing, thereby simplifying the pipeline of WSSS. Our method is validated on the PASCAL VOC 2012 and COCO2014 datasets and achieves the new state-of-the-art.



Paperid:1197
Authors:Joakim Johnander; Johan Edstedt; Michael Felsberg; Fahad Shahbaz Khan; Martin Danelljan
Title: Dense Gaussian Processes for Few-Shot Segmentation
Abstract:
Few-shot segmentation is a challenging dense prediction task, which entails segmenting a novel query image given only a small annotated support set. The key problem is thus to design a method that aggregates detailed information from the support set, while being robust to large variations in appearance and context. To this end, we propose a few-shot segmentation method based on dense Gaussian process (GP) regression. Given the support set, our dense GP learns the mapping from local deep image features to mask values, capable of capturing complex appearance distributions. Furthermore, it provides a principled means of capturing uncertainty, which serves as another powerful cue for the final segmentation, obtained by a CNN decoder. Instead of a one-dimensional mask output, we further exploit the end-to-end learning capabilities of our approach to learn a high-dimensional output space for the GP. Our approach sets a new state-of-the-art on the PASCAL-5i and COCO-20i benchmarks, achieving an absolute gain of +8.4 mIoU in the COCO-20i 5-shot setting. Furthermore, the segmentation quality of our approach scales gracefully when increasing the support set size, while achieving robust cross-dataset transfer.



Paperid:1198
Authors:Yizheng Wu; Min Shi; Shuaiyuan Du; Hao Lu; Zhiguo Cao; Weicai Zhong
Title: 3D Instances as 1D Kernels
Abstract:
We introduce a 3D instance representation, termed instance kernels, where instances are represented by one-dimensional vectors that encode the semantic, positional, and shape information of 3D instances. We show that instance kernels enable easy mask inference by simply scanning kernels over the entire scenes, avoiding the heavy reliance on proposals or heuristic clustering algorithms in standard 3D instance segmentation pipelines. The idea of instance kernel is inspired by recent success of dynamic convolutions in 2D/3D instance segmentation. However, we find it non-trivial to represent 3D instances due to the disordered and unstructured nature of point cloud data, e.g., poor instance localization can significantly degrade instance representation. To remedy this, we construct a novel 3D instance encoding paradigm. First, potential instance centroids are localized as candidates. Then, a candidate merging scheme is devised to simultaneously aggregate duplicated candidates and collect context around the merged centroids to form the instance kernels. Once instance kernels are available, instance masks can be reconstructed via dynamic convolutions whose weights are conditioned on instance kernels. The whole pipeline is instantiated with a dynamic kernel network (DKNet). Results show that DKNet outperforms the state of the arts on both ScanNetV2 and S3DIS datasets with better instance localization. Code is available: https://github.com/W1zheng/DKNet.



Paperid:1199
Authors:Huanqia Cai; Fanglei Xue; Lele Xu; Lili Guo
Title: TransMatting: Enhancing Transparent Objects Matting with Transformers
Abstract:
Image matting refers to predicting the alpha values of unknown foreground areas from natural images. Prior methods have focused on propagating alpha values from known to unknown regions. However, not all natural images have a specifically known foreground. Images of transparent objects, like glass, smoke, web, etc., have less or no known foreground. In this paper, we propose a Transformer-based network, TransMatting, to model transparent objects with a big receptive field. Specifically, we redesign the trimap as three learnable tri-tokens for introducing advanced semantic features into the self-attention mechanism. A small convolutional network is proposed to utilize the global feature and non-background mask to guide the multi-scale feature propagation from encoder to decoder for maintaining the contexture of transparent objects. In addition, we create a high-resolution matting dataset of transparent objects with small known foreground areas. Experiments on several matting benchmarks demonstrate the superiority of our proposed method over the current state-of-the-art methods.



Paperid:1200
Authors:Jiayuan Zhou; Lijun Wang; Huchuan Lu; Kaining Huang; Xinchu Shi; Bocong Liu
Title: MVSalNet:Multi-View Augmentation for RGB-D Salient Object Detection
Abstract:
RGB-D salient object detection (SOD) enjoys significant advantages in understanding 3D geometry of the scene. However, the geometry information conveyed by depth maps are mostly under-explored in existing RGB-D SOD methods. In this paper, we propose a new framework to address this issue. We augment the input image with multiple different views rendered using the depth maps, and cast the conventional single-view RGB-D SOD into a multi-view setting. Since different views captures complementary context of the 3D scene, the accuracy can be significantly improved through multi-view aggregation. We further design a multi-view saliency detection network (MVSalNet), which firstly performs saliency prediction for each view separately and incorporates multi-view outputs through a fusion model to produce final saliency prediction. A dynamic filtering module is also designed to facilitate more effective and flexible feature extraction. Extensive experiments on 6 widely used datasets demonstrate that our approach compares favorably against state-of-the-art approaches.



Paperid:1201
Authors:Qihang Yu; Huiyu Wang; Siyuan Qiao; Maxwell Collins; Yukun Zhu; Hartwig Adam; Alan Yuille; Liang-Chieh Chen
Title: $k$-Means Mask Transformer
Abstract:
The rise of transformers in vision tasks not only advances network backbone designs, but also starts a brand-new page to achieve end-to-end image recognition (e.g., object detection and panoptic segmentation). Originated from Natural Language Processing (NLP), transformer architectures, consisting of self-attention and cross-attention, effectively learn long-range interactions between elements in a sequence. However, we observe that most existing transformer-based vision models simply borrow the idea from NLP, neglecting the crucial difference between languages and images, particularly the extremely large sequence length of spatially flattened pixel features. This subsequently impedes the learning in cross-attention between pixel features and object queries. In this paper, we rethink the relationship between pixels and object queries and propose to reformulate the cross-attention learning as a clustering process. Inspired by the traditional k-means clustering algorithm, we develop a k-means Mask Xformer (kMaX-DeepLab) for segmentation tasks, which not only improves the state-of-the-art, but also enjoys a simple and elegant design. As a result, our kMaX-DeepLab achieves a new state-of-the-art performance on COCO val set with 58.0% PQ, and Cityscapes val set with 68.4% PQ, 44.0% AP, and 83.5% mIoU without test-time augmentation or external dataset. We hope our work can shed some light on designing transformers tailored for vision tasks. Code and models are available at https://github.com/google-research/deeplab2"



Paperid:1202
Authors:Jindong Gu; Hengshuang Zhao; Volker Tresp; Philip H. S. Torr
Title: SegPGD: An Effective and Efficient Adversarial Attack for Evaluating and Boosting Segmentation Robustness
Abstract:
Deep neural network-based image classifications are vulnerable to adversarial perturbations. The image classifications can be easily fooled by adding artificial small and imperceptible perturbations to input images. As one of the most effective defense strategies, adversarial training was proposed to address the vulnerability of classification models, where the adversarial examples are created and injected into training data during training. The attack and defense of classification models have been intensively studied in past years. Semantic segmentation, as an extension of classifications, has also received great attention recently. Recent work shows a large number of attack iterations are required to create effective adversarial examples to fool segmentation models. The observation makes both robustness evaluation and adversarial training on segmentation models challenging. In this work, we propose an effective and efficient segmentation attack method, dubbed SegPGD. Besides, we provide a convergence analysis to show the proposed SegPGD can create more effective adversarial examples than PGD under the same number of attack iterations. Furthermore, we propose to apply our SegPGD as the underlying attack method for segmentation adversarial training. Since SegPGD can create more effective adversarial examples, the adversarial training with our SegPGD can boost the robustness of segmentation models. Our proposals are also verified with experiments on popular Segmentation model architectures and standard segmentation datasets.



Paperid:1203
Authors:Sung-Hoon Yoon; Hyeokjun Kweon; Jegyeong Cho; Shinjeong Kim; Kuk-Jin Yoon
Title: Adversarial Erasing Framework via Triplet with Gated Pyramid Pooling Layer for Weakly Supervised Semantic Segmentation
Abstract:
Weakly supervised semantic segmentation (WSSS) has employed Class Activation Maps (CAMs) to localize the objects. However, the CAMs typically do not fit along the object boundaries and highlight only the most-discriminative regions. To resolve the problems, we propose a Gated Pyramid Pooling (GPP) layer which is a substitute for a Global Average Pooling (GAP) layer, and an Adversarial Erasing Framework via Triplet (AEFT). In the GPP layer, a feature pyramid is obtained by pooling the CAMs at multiple spatial resolutions, and then be aggregated into an attention for class prediction by gated convolution. With the process, CAMs are trained not only to capture the global context but also to preserve fine-details from the image. Meanwhile, the AEFT targets an over-expansion, a chronic problem of Adversarial Erasing (AE). Although AE methods expand CAMs by erasing the discriminative regions, they usually suffer from the over-expansion due to an absence of guidelines on when to stop erasing. We experimentally verify that the over-expansion is due to rigid classification, and metric learning can be a flexible remedy for it. AEFT is devised to learn the concept of erasing with the triplet loss between the input image, erased image, and negatively sampled image. With the GPP and AEFT, we achieve new state-of-the-art both on the PASCAL VOC 2012 val/test and MS-COCO 2014 val set by 70.9%/71.7% and 44.8% in mIoU, respectively.



Paperid:1204
Authors:Zihan Lin; Zilei Wang; Yixin Zhang
Title: Continual Semantic Segmentation via Structure Preserving and Projected Feature Alignment
Abstract:
Deep networks have been shown to suffer from catastrophic forgetting. In this work, we try to alleviate this phenomenon in the field of continual semantic segmentation (CSS). We observe that two main problems lie in existing arts. First, attention is only paid to designing constraints for encoder (i.e., the backbone of segmentation network) or output probabilities. But we find that forgetting also happens in the decoder head and harms the performance greatly. Second, old and new knowledge are entangled in intermediate features when learning new categories, making existing practices hard to balance between plasticity and stability. On these bases, we propose a framework driven by two novel constraints to address the aforementioned problems. First, a structure preserving loss is applied to the decoder’s output to maintain the discriminative power of old classes from two different granularities in embedding space. Second, a feature projection module is adopted to disentangle the process of preserving old knowledge from learning new classes. Extensive evaluations on VOC2012 and ADE20K datasets show the effectiveness of our approach, which significantly outperforms existing state-of-the-art CSS methods.



Paperid:1205
Authors:Atsuro Okazawa
Title: Interclass Prototype Relation for Few-Shot Segmentation
Abstract:
Traditional semantic segmentation requires a large labeled image dataset and can only be predicted within predefined classes. Solving this problem of few-shot segmentation, which requires only a handful of annotations for the new target class, is important. However, with few-shot segmentation, the target class data distribution in the feature space is sparse and has low coverage because of the slight variations in the sample data. Setting the classification boundary that properly separates the target class from other classes is an impossible task. In particular, it is difficult to classify classes that are similar to the target class near the boundary. This study proposes the Interclass Prototype Relation Network (IPRNet), which improves the separation performance by reducing the similarity between other classes. We conducted extensive experiments with Pascal-5i and COCO-20i and showed that IPRNet provides the best segmentation performance compared with previous research.



Paperid:1206
Authors:Kunyang Han; Jun Hao Liew; Jiashi Feng; Huawei Tian; Yao Zhao; Yunchao Wei
Title: Slim Scissors: Segmenting Thin Object from Synthetic Background
Abstract:
Existing interactive segmentation algorithms typically fail when segmenting objects with elongated thin structures (bicycle spokes). Though some recent efforts attempt to address this challenge by introducing a new synthetic dataset and a three-stream network design, they suffer two limitations: 1) large performance gap when tested on real image domain; 2) still requiring extensive amounts of user interactions (clicks) if the thin structures are not well segmented. To solve them, we develop Slim Scissors, which enables quick extraction of elongated thin parts by simply brushing some coarse scribbles. Our core idea is to segment thin parts by learning to compare the original image to a synthesized background without thin structures. Our method is model-agnostic and seamlessly applicable to existing state-of-the-art interactive segmentation models. To further reduce the annotation burden, we devise a similarity detection module, which enables the model to automatically synthesize background for other similar thin structures from only one or two scribbles. Extensive experiments on COIFT, HRSOD and ThinObject-5K clearly demonstrate the superiority of Slim Scissors for thin object segmentation: it outperforms TOS-Net by 5.9% IoU_thin and 3.5% F score on the real dataset HRSOD.



Paperid:1207
Authors:Stephan Alaniz; Massimiliano Mancini; Anjan Dutta; Diego Marcos; Zeynep Akata
Title: Abstracting Sketches through Simple Primitives
Abstract:
Humans show high-level of abstraction capabilities in games that require quickly communicating object information. They decompose the message content into multiple parts and communicate them in an interpretable protocol. Toward equipping machines with such capabilities, we propose the Primitive-based Sketch Abstraction task where the goal is to represent sketches using a fixed set of drawing primitives under the influence of a budget. To solve this task, our Primitive-Matching Network (PMN), learns interpretable abstractions of a sketch in a self supervised manner. Specifically, PMN maps each stroke of a sketch to its most similar primitive in a given set, predicting an affine transformation that aligns the selected primitive to the target stroke. We learn this stroke-to-primitive mapping end-to-end with a distance-transform loss that is minimal when the original sketch is precisely reconstructed with the predicted primitives. Our PMN abstraction empirically achieves the highest performance on sketch recognition and sketch-based image retrieval given a communication budget, while at the same time being highly interpretable. This opens up new possibilities for sketch analysis, such as comparing sketches by extracting the most relevant primitives that define an object category. Code is available at https://github.com/ExplainableML/sketch-primitives.



Paperid:1208
Authors:Theodoros Pissas; Claudio S. Ravasio; Lyndon Da Cruz; Christos Bergeles
Title: Multi-Scale and Cross-Scale Contrastive Learning for Semantic Segmentation
Abstract:
This work considers supervised contrastive learning for semantic segmentation. We apply contrastive learning to enhance the discriminative power of the multi-scale features extracted by semantic segmentation networks. Our key methodological insight is to leverage samples from the feature spaces emanating from multiple stages of a model’s encoder itself requiring neither data augmentation nor online memory banks to obtain a diverse set of samples. To allow for such an extension we introduce an efficient and effective sampling process, that enables applying contrastive losses over the encoder’s features at multiple scales. Furthermore, by first mapping the encoder’s multi-scale representations to a common feature space, we instantiate a novel form of supervised local-global constraint by introducing cross-scale contrastive learning linking high-resolution local features to low-resolution global features. Combined, our multi-scale and cross-scale contrastive losses boost performance of various models (DeepLabv3, HRNet, OCRNet, UPerNet) with both CNN and Transformer backbones, when evaluated on 4 diverse datasets from natural (Cityscapes, PascalContext, ADE20K) but also surgical (CaDIS) domains. Our code is available at https://github.com/RViMLab/MS_CS_ContrSeg.



Paperid:1209
Authors:Hongje Seong; Seoung Wug Oh; Brian Price; Euntai Kim; Joon-Young Lee
Title: One-Trimap Video Matting
Abstract:
Recent studies made great progress in video matting by extending the success of trimap-based image matting to the video domain. In this paper, we push this task toward a more practical setting and propose One-Trimap Video Matting network (OTVM) that performs video matting robustly using only one user-annotated trimap. A key of OTVM is the joint modeling of trimap propagation and alpha prediction. Starting from baseline trimap propagation and alpha prediction networks, our OTVM combines the two networks with an alpha-trimap refinement module to facilitate information flow. We also present an end-to-end training strategy to take full advantage of the joint model. Our joint modeling greatly improves the temporal stability of trimap propagation compared to the previous decoupled methods. We evaluate our model on two latest video matting benchmarks, Deep Video Matting and VideoMatting108, and outperform state-of-the-art by significant margins (MSE improvements of 56.4% and 56.7%, respectively). The source code and model are available online: https://github.com/Hongje/OTVM.



Paperid:1210
Authors:Tsung-Han Wu; Yi-Syuan Liou; Shao-Ji Yuan; Hsin-Ying Lee; Tung-I Chen; Kuan-Chih Huang; Winston H. Hsu
Title: D2ADA: Dynamic Density-Aware Active Domain Adaptation for Semantic Segmentation
Abstract:
In the field of domain adaptation, a trade-off exists between the model performance and the number of target domain annotations. Active learning, maximizing model performance with few informative labeled data, comes in handy for such a scenario. In this work, we present D2ADA, a general active domain adaptation framework for semantic segmentation. To adapt the model to the target domain with minimum queried labels, we propose acquiring labels of the samples with high probability density in the target domain yet with low probability density in the source domain, complementary to the existing source domain labeled data. To further facilitate labeling efficiency, we design a dynamic scheduling policy to adjust the labeling budgets between domain exploration and model uncertainty over time. Extensive experiments show that our method outperforms existing active learning and domain adaptation baselines on two benchmarks, GTA5 -> Cityscapes and SYNTHIA -> Cityscapes. With less than 5% target domain annotations, our method reaches comparable results with that of full supervision.



Paperid:1211
Authors:Yong Liu; Ran Yu; Fei Yin; Xinyuan Zhao; Wei Zhao; Weihao Xia; Yujiu Yang
Title: Learning Quality-Aware Dynamic Memory for Video Object Segmentation
Abstract:
Recently, several spatial-temporal memory-based methods have verified that storing intermediate frames and their masks as memory are helpful to segment target objects in videos. However, they mainly focus on better matching between the current frame and the memory frames without explicitly paying attention to the quality of the memory. Therefore, frames with poor segmentation masks are prone to be memorized, which leads to an segmentation mask error accumulation problem and further affect the segmentation performance. In addition, the linear increase of memory frames with the growth of frame number also limits the ability of the models to handle long videos. To this end, we propose a Quality-aware Dynamic Memory Network (QDMN) to evaluate the segmentation quality of each frame, allowing the memory bank to selectively store accurately segmented frames to prevent the error accumulation problem. Then, we combine the segmentation quality with temporal consistency to dynamically update the memory bank to improve the practicability of the models. Without any bells and whistles, our QDMN achieves new state-of-the-art performance on both DAVIS and YouTube-VOS benchmarks. Moreover, extensive experiments demonstrate that the proposed Quality Assessment Module (QAM) can be applied to memory-based methods as generic plugins and significantly improves performance. Our source code is available at https://github.com/workforai/QDMN.



Paperid:1212
Authors:Hanzhe Hu; Yinbo Chen; Jiarui Xu; Shubhankar Borse; Hong Cai; Fatih Porikli; Xiaolong Wang
Title: Learning Implicit Feature Alignment Function for Semantic Segmentation
Abstract:
Integrating high-level context information with low-level details is of central importance in semantic segmentation. Towards this end, most existing segmentation models apply bilinear up-sampling and convolutions to feature maps of different scales, and then align them at the same resolution. However, bilinear up-sampling blurs the precise information learned in these feature maps and convolutions incur extra computation costs. To address these issues, we propose the Implicit Feature Alignment function (IFA). Our method is inspired by the rapidly expanding topic of implicit neural representations, where coordinate-based neural networks are used to designate fields of signals. In IFA, feature vectors are viewed as representing a 2D field of information. Given a query coordinate, nearby feature vectors with their relative coordinates are taken from the multi-level feature maps and then fed into an MLP to generate the corresponding output. As such, IFA implicitly aligns the feature maps at different levels and is capable of producing segmentation maps in arbitrary resolutions. We demonstrate the efficacy of IFA on multiple datasets, including Cityscapes, PASCAL Context, and ADE20K. Our method can be combined with improvement on various architectures, and it achieves state-of-the-art computation-accuracy trade-off on common benchmarks.



Paperid:1213
Authors:Federica Arrigoni; Willi Menapace; Marcel Seelbach Benkner; Elisa Ricci; Vladislav Golyanik
Title: Quantum Motion Segmentation
Abstract:
Motion segmentation is a challenging problem that seeks to identify independent motions in two or several input images. This paper introduces the first algorithm for motion segmentation that relies on adiabatic quantum optimization of the objective function. The proposed method achieves on-par performance with the state of the art on problem instances which can be mapped to modern quantum annealers.



Paperid:1214
Authors:Feng Zhu; Zongxin Yang; Xin Yu; Yi Yang; Yunchao Wei
Title: Instance As Identity: A Generic Online Paradigm for Video Instance Segmentation
Abstract:
Modeling temporal information for both detection and tracking in a unified framework has been proved a promising solution to video instance segmentation (VIS). However, how to effectively incorporate the temporal information into an online model remains an open problem. In this work, we propose a new online VIS paradigm named Instance As Identity (IAI), which models temporal information for both detection and tracking in an efficient way. In detail, IAI employs a novel identification module to predict identification number for tracking instances explicitly. For passing temporal information cross frame, IAI utilizes an association module which combines current features and past embeddings. Notably, IAI can be integrated with different image models. We conduct extensive experiments on three VIS benchmarks. IAI outperforms all the online competitors on YouTube-VIS-2019 (ResNet-101 41.9 mAP) and YouTube-VIS-2021 (ResNet-50 37.7 mAP). Surprisingly, on the more challenging OVIS, IAI achieves SOTA performance (20.3 mAP). Code is available at https://github.com/zfonemore/IAI keywords: Video Instance Segmentation"



Paperid:1215
Authors:Xiao-Juan Li; Jie Yang; Fang-Lue Zhang
Title: Laplacian Mesh Transformer: Dual Attention and Topology Aware Network for 3D Mesh Classification and Segmentation
Abstract:
Deep learning-based approaches for shape understanding and processing tasks have attracted considerable attention. Despite the great progress that has been made, the existing approaches fail to efficiently capture sophisticated structure information and critical part features simultaneously, limiting their capability of providing discriminative deep shape features. To address the above issue, we proposed a novel deep learning framework, Laplacian Mesh Transformer, to extract the critical structure and geometry features. We introduce a dual attention mechanism, where the $1^{\rm st}$ level self-attention mechanism is used to capture the structure and critical partial geometric information on the entire mesh, and the $2^{\rm nd}$ level is to learn the importance of structure and geometric information. More particularly, Laplacian spectral decomposition is adopted as our basic structure representation given its ability to describe shape topology. Our approach builds a hierarchical structure to process shape features from fine to coarse using the dual attention mechanism, which is stable under the isometric transformations. It enables an effective feature extraction that can tackle 3D meshes with complex structure and geometry efficiently in various shape analysis tasks, such as shape segmentation and classification. Extensive experiments on the standard benchmarks show that our method outperforms state-of-the-art methods.



Paperid:1216
Authors:Tuan Ngo; Khoi Nguyen
Title: Geodesic-Former: A Geodesic-Guided Few-Shot 3D Point Cloud Instance Segmenter
Abstract:
This paper introduces a new problem in 3D point cloud: few-shot instance segmentation. Given a few annotated point clouds exemplified a target class, our goal is to segment all instances of this target class in a query point cloud. This problem has a wide range of practical applications where point-wise instance segmentation annotation is prohibitively expensive to collect. To address this problem, we present Geodesic-Former -- the first geodesic-guided transformer for 3D point cloud instance segmentation. The key idea is to leverage the geodesic distance to tackle the density imbalance of LiDAR 3D point clouds. The LiDAR 3D point clouds are dense near the object surface and sparse or empty elsewhere making the Euclidean distance less effective to distinguish different objects. The geodesic distance, on the other hand, is more suitable since it encodes the scene’s geometry which can be used as a guiding signal for the attention mechanism in a transformer decoder to generate kernels representing distinct features of instances. These kernels are then used in a dynamic convolution to obtain the final instance masks. To evaluate Geodesic-Former on the new task, we propose new splits of the two common 3D point cloud instance segmentation datasets: ScannetV2 and S3DIS. Geodesic-Former consistently outperforms strong baselines adapted from state-of-the-art 3D point cloud instance segmentation approaches with a significant margin. Code is available at https://github.com/VinAIResearch/GeoFormer.



Paperid:1217
Authors:Zongyao Li; Ren Togo; Takahiro Ogawa; Miki Haseyama
Title: Union-Set Multi-source Model Adaptation for Semantic Segmentation
Abstract:
This paper solves a generalized version of the problem of multi-source model adaptation for semantic segmentation. Model adaptation is proposed as a new domain adaptation problem which requires access to a pre-trained model instead of data for the source domain. A general multi-source setting of model adaptation assumes strictly that each source domain shares a common label space with the target domain. As a relaxation, we allow the label space of each source domain to be a subset of that of the target domain and require the union of the source-domain label spaces to be equal to the target-domain label space. For the new setting named union-set multi-source model adaptation, we propose a method with a novel learning strategy named model-invariant feature learning, which takes full advantage of the diverse characteristics of the source-domain models, thereby improving the generalization in the target domain. We conduct extensive experiments in various adaptation settings to show the superiority of our method.



Paperid:1218
Authors:Ardian Umam; Cheng-Kun Yang; Yung-Yu Chuang; Jen-Hui Chuang; Yen-Yu Lin
Title: Point MixSwap: Attentional Point Cloud Mixing via Swapping Matched Structural Divisions
Abstract:
Data augmentation is developed for increasing the amount and diversity of training data to enhance model learning. Compared to 2D images, point clouds, with the 3D geometric nature as well as the high collection and annotation costs, pose great challenges and potentials for augmentation. This paper presents a 3D augmentation method that explores the structural variance across multiple point clouds, and generates more diverse point clouds to enrich the training set. Specifically, we propose an attention module that decomposes a point cloud into several disjoint point subsets, called divisions, in a way where each division has a corresponding division in another point cloud. The augmented point clouds are synthesized by swapping matched divisions. They exhibit high diversity since both intra- and inter-cloud variations are explored, hence useful for downstream tasks. The proposed method for augmentation can act as a module and be integrated into a point-based network. The resultant framework is end-to-end trainable. The experiments show that it achieves state-of-the-art performance on the ModelNet40 and ModelNet10 benchmarks. The code for this work is publicly available.



Paperid:1219
Authors:Ye Yu; Jialin Yuan; Gaurav Mittal; Li Fuxin; Mei Chen
Title: BATMAN: Bilateral Attention Transformer in Motion-Appearance Neighboring Space for Video Object Segmentation
Abstract:
Video Object Segmentation (VOS) is fundamental to video understanding. Transformer-based methods show significant performance improvement on semi-supervised VOS. However, existing work faces challenges segmenting visually similar objects in close proximity of each other. In this paper, we propose a novel Bilateral Attention Transformer in Motion-Appearance Neighboring space (BATMAN) for semi-supervised VOS. It captures object motion in the video via a novel optical flow calibration module that fuses the segmentation mask with optical flow estimation to improve within-object optical flow smoothness and reduce noise at object boundaries. This calibrated optical flow is then employed in our novel bilateral attention, which computes the correspondence between the query and reference frames in the neighboring bilateral space considering both motion and appearance. Extensive experiments validate the effectiveness of BATMAN architecture by outperforming all existing state-of-the-art on all four popular VOS benchmarks: Youtube-VOS 2019 (85.0%), Youtube-VOS 2018 (85.3%), DAVIS 2017Val/Test-dev (86.2%/82.2%), and DAVIS 2016 (92.5%).



Paperid:1220
Authors:Minhyeok Lee; Chaewon Park; Suhwan Cho; Sangyoun Lee
Title: SPSN: Superpixel Prototype Sampling Network for RGB-D Salient Object Detection
Abstract:
RGB-D salient object detection (SOD) has been in the spotlight recently because it is an important preprocessing operation for various vision tasks. However, despite advances in deep learning-based methods, RGB-D SOD is still challenging due to the large domain gap between an RGB image and the depth map and low-quality depth maps. To solve this problem, we propose a novel superpixel prototype sampling network (SPSN) architecture. The proposed model splits the input RGB image and depth map into component superpixels to generate component prototypes. We design a prototype sampling network so that the network only samples prototypes corresponding to salient objects. In addition, we propose a reliance selection module to recognize the quality of each RGB and depth feature map and adaptively weight them in proportion to their reliability. The proposed method makes the model robust to inconsistencies between RGB images and depth maps and eliminates the influence of non-salient objects. Our method is evaluated on five popular datasets, achieving state-of-the-art performance. We prove the effectiveness of the proposed method through comparative experiments"



Paperid:1221
Authors:Yong Liu; Ran Yu; Jiahao Wang; Xinyuan Zhao; Yitong Wang; Yansong Tang; Yujiu Yang
Title: Global Spectral Filter Memory Network for Video Object Segmentation
Abstract:
This paper studies semi-supervised video object segmentation through boosting intra-frame interaction. Recent memory network-based methods focus on exploiting inter-frame temporal reference while paying little attention to intra-frame spatial dependency. Specifically, these segmentation model tends to be susceptible to interference from unrelated nontarget objects in a certain frame. To this end, we propose Global Spectral Filter Memory network (GSFM), which improves intra-frame interaction through learning long-term spatial dependencies in the spectral domain. The key components of GSFM is 2D (inverse) discrete Fourier transform for spatial information mixing. Besides, we empirically find low frequency feature should be enhanced in encoder (backbone) while high frequency for decoder (segmentation head). We attribute this to semantic information extracting role for encoder and fine-grained details highlighting role for decoder. Thus, Low (High)-Frequency Module is proposed to fit this circumstance. Extensive experiments on the popular DAVIS and YouTube-VOS benchmarks demonstrate that GSFM noticeably outperforms the baseline method and achieves state-of-the-art performance. Besides, extensive analysis shows that the proposed modules are reasonable and of great generalization ability. Our source code is available at https://github.com/workforai/GSFM.



Paperid:1222
Authors:Omkar Thawakar; Sanath Narayan; Jiale Cao; Hisham Cholakkal; Rao Muhammad Anwer; Muhammad Haris Khan; Salman Khan; Michael Felsberg; Fahad Shahbaz Khan
Title: Video Instance Segmentation via Multi-Scale Spatio-Temporal Split Attention Transformer
Abstract:
State-of-the-art transformer-based video instance segmentation (VIS) approaches typically utilize either single-scale spatio-temporal features or per-frame multi-scale features during the attention computations. We argue that such an attention computation ignores the multi-scale spatio-temporal feature relationships that are crucial to tackle target appearance deformations in videos. To address this issue, we propose a transformer-based VIS framework, named MS-STS VIS, that comprises a novel multi-scale spatio-temporal split (MS-STS) attention module in the encoder. The proposed MS-STS module effectively captures spatio-temporal feature relationships at multiple scales across frames in a video. We further introduce an attention block in the decoder to enhance the temporal consistency of the detected instances in different frames of a video. Moreover, an auxiliary discriminator is introduced during training to ensure better foreground-background separability within the multi-scale spatio-temporal feature space. We conduct extensive experiments on two benchmarks: Youtube-VIS (2019 and 2021). Our MS-STS VIS achieves state-of-the-art performance on both benchmarks. When using the ResNet50 backbone, our MS-STS achieves a mask AP of 50.1%, outperforming the best reported results in literature by 2.7% and by 4.8% at higher overlap threshold of AP75, while being comparable in model size and speed on Youtube-VIS 2019 val. set. When using the Swin Transformer backbone, MS-STS VIS achieves mask AP of 61.0% on Youtube-VIS 2019 val. set. Our code and models will be made public.



Paperid:1223
Authors:Haodi He; Yuhui Yuan; Xiangyu Yue; Han Hu
Title: RankSeg: Adaptive Pixel Classification with Image Category Ranking for Segmentation
Abstract:
The segmentation task has traditionally been formulated as a complete-label pixel classification task to predict a class for each pixel from a fixed number of predefined semantic categories shared by all images or videos. Yet, following this formulation, standard architectures will inevitably encounter various challenges under more realistic settings where the scope of categories scales up (e.g., beyond the level of 1k). On the other hand, in a typical image or video, only a few categories, i.e., a small subset of the complete label are present. Motivated by this intuition, in this paper, we propose to decompose segmentation into two sub-problems: (i) image-level or video-level multi-label classification and (ii) pixel-level rank-adaptive selected-label classification. Given an input image or video, our framework first conducts multi-label classification over the complete label, then sorts the complete label and selects a small subset according to their class confidence scores. We then use a rank-adaptive pixel classifier to perform the pixel-wise classification over only the selected labels, which uses a set of rank-oriented learnable temperature parameters to adjust the pixel classifications scores. Our approach is conceptually general and can be used to improve various existing segmentation frameworks by simply using a lightweight multi-label classification head and rank-adaptive pixel classifier. We demonstrate the effectiveness of our framework with competitive experimental results across four tasks, including image semantic segmentation, image panoptic segmentation, video instance segmentation, and video semantic segmentation. Especially, with our RankSeg, Mask2Former gains +0.8%/+0.7%/+0.7% on ADE20K panoptic segmentation/YouTubeVIS 2019 video instance segmentation/VSPW video semantic segmentation benchmarks respectively.



Paperid:1224
Authors:Saumya Gupta; Xiaoling Hu; James Kaan; Michael Jin; Mutshipay Mpoy; Katherine Chung; Gagandeep Singh; Mary Saltz; Tahsin Kurc; Joel Saltz; Apostolos Tassiopoulos; Prateek Prasanna; Chao Chen
Title: Learning Topological Interactions for Multi-Class Medical Image Segmentation
Abstract:
Deep learning methods have achieved impressive performance for multi-class medical image segmentation. However, they are limited in their ability to encode topological interactions among different classes (e.g., containment and exclusion). These constraints naturally arise in biomedical images and can be crucial in improving segmentation quality. In this paper, we introduce a novel topological interaction module to encode the topological interactions into a deep neural network. The implementation is completely convolution-based and thus can be very efficient. This empowers us to incorporate the constraints into end-to-end training and enrich the feature representation of neural networks. The efficacy of the proposed method is validated on different types of interactions. We also demonstrate the generalizability of the method on both proprietary and public challenge datasets, in both 2D and 3D settings, as well as across different modalities such as CT and Ultrasound. Code is available at: https://github.com/TopoXLab/TopoInteraction"



Paperid:1225
Authors:Honglin Chen; Rahul Venkatesh; Yoni Friedman; Jiajun Wu; Joshua B. Tenenbaum; Daniel L. K. Yamins; Daniel M. Bear
Title: Unsupervised Segmentation in Real-World Images via Spelke Object Inference
Abstract:
Self-supervised, category-agnostic segmentation of real-world images is a challenging open problem in computer vision. Here, we show how to learn static grouping priors from motion self-supervision by building on the cognitive science concept of a Spelke Object: a set of physical stuff that moves together. We introduce the Excitatory-Inhibitory Segment Extraction Network (EISEN), which learns to extract pairwise affinity graphs for static scenes from motion-based training signals. EISEN then produces segments from affinities using a novel graph propagation and competition network. During training, objects that undergo correlated motion (such as robot arms and the objects they move) are decoupled by a bootstrapping process: EISEN explains away the motion of objects it has already learned to segment. We show that EISEN achieves a substantial improvement in the state of the art for self-supervised image segmentation on challenging synthetic and real-world robotics datasets.



Paperid:1226
Authors:Mengde Xu; Zheng Zhang; Fangyun Wei; Yutong Lin; Yue Cao; Han Hu; Xiang Bai
Title: A Simple Baseline for Open-Vocabulary Semantic Segmentation with Pre-trained Vision-Language Model
Abstract:
Recently, open-vocabulary image classification by vision language pre-training has demonstrated incredible achievements, that the model can classify arbitrary categories without seeing additional annotated images of that category. However, it is still unclear how to make the open-vocabulary recognition work well on broader vision problems. This paper targets open-vocabulary semantic segmentation by building it on an off-the-shelf pre-trained vision-language model, i.e., CLIP. However, semantic segmentation and the CLIP model perform on different visual granularity, that semantic segmentation processes on pixels while CLIP performs on images. To remedy the discrepancy in processing granularity, we refuse the use of the prevalent one-stage FCN based framework, and advocate a two-stage semantic segmentation framework, with the first stage extracting generalizable mask proposals and the second stage leveraging an image based CLIP model to perform open-vocabulary classification on the masked image crops which are generated in the first stage. Our experimental results show that this two-stage framework can achieve superior performance than FCN when trained only on COCO Stuff dataset and evaluated on other datasets without fine-tuning. Moreover, this simple framework also surpasses previous state-of-the-arts of zero-shot semantic segmentation by a large margin: +29.5 hIoU on the Pascal VOC 2012 dataset, and +8.9 hIoU on the COCO Stuff dataset. With its simplicity and strong performance, we hope this framework to serve as a baseline to facilitate future research. The code are made publicly available at https://github.com/MendelXu/zsseg.baseline.



Paperid:1227
Authors:Bengisu Ozbay; Octavia Camps; Mario Sznaier
Title: Fast Two-View Motion Segmentation Using Christoffel Polynomials
Abstract:
We address the problem of segmenting moving rigid objects based on two-view image correspondences under a perspective camera model. While this is a well understood problem, existing methods scale poorly with the number of correspondences. In this paper we propose a fast segmentation algorithm that scales linearly with the number of correspondences and show that on benchmark datasets it offers the best trade-off between error and computational time: it is at least one order of magnitude faster than the best method (with comparable or better accuracy), with the ratio growing up to three orders of magnitude for larger number of correspondences. We approach the problem from an algebraic perspective by exploiting the fact that all points belonging to a given object lie in the same quadratic surface. The proposed method is based on a characterization of each surface in terms of the Christoffel polynomial associated with the probability that a given point belongs to the surface. This allows for efficiently segmenting points “one surface at a time” in O(number of points).



Paperid:1228
Authors:Xiaowen Ying; Mooi Choo Chuah
Title: UCTNet: Uncertainty-Aware Cross-Modal Transformer Network for Indoor RGB-D Semantic Segmentation
Abstract:
In this paper, we tackle the problem of RGB-D Semantic Segmentation. The key challenges in solving this problem lie in 1) how to extract features from depth sensor data and 2) how to effectively fuse the features extracted from the two modalities. For the first challenge, we found that the depth information obtained from the sensor is not always reliable (e.g. objects with reflective or dark surfaces typically have inaccurate or void sensor readings), and existing methods that extract depth features using ConvNets did not explicitly consider the reliability of depth value at different pixel locations. To tackle this challenge, we propose a novel mechanism, namely Uncertainty-Aware Self-Attention that explicitly controls the information flow from unreliable depth pixels to confident depth pixels during feature extraction. For the second challenge, we propose an effective and scalable fusion module based on Cross-Attention that can perform adaptive and asymmetric information exchange between the RGB and depth encoder. Our proposed framework, namely UCTNet, is an encoder-decoder network that naturally incorporates these two key designs for robust and accurate RGB-D Segmentation. Experimental results show that UCTNet outperforms existing works and achieves state-of-the-art performances on two RGB-D Semantic Segmentation benchmarks.



Paperid:1229
Authors:Geon Lee; Chanho Eom; Wonkyung Lee; Hyekang Park; Bumsub Ham
Title: Bi-directional Contrastive Learning for Domain Adaptive Semantic Segmentation
Abstract:
We present a novel unsupervised domain adaptation method for semantic segmentation that generalizes a model trained with source images and corresponding ground-truth labels to a target domain. A key to domain adaptive semantic segmentation is to learn domain-invariant and discriminative features without target ground-truth labels. To this end, we propose a bi-directional pixel-prototype contrastive learning framework that minimizes intra-class variations of features for the same object class, while maximizing inter-class variations for different ones, regardless of domains. Specifically, our framework aligns pixel-level features and a prototype of the same object class in target and source images (i.e., positive pairs), respectively, sets them apart for different classes (i.e., negative pairs), and performs the alignment and separation processes toward the other direction with pixel-level features in the source image and a prototype in the target image. The cross-domain matching encourages domain-invariant feature representations, while the bidirectional pixel-prototype correspondences aggregate features for the same object class, providing discriminative features. To establish training pairs for contrastive learning, we propose to generate dynamic pseudo labels of target images using a non-parametric label transfer, that is, pixel-prototype correspondences across different domains. We also present a calibration method compensating class-wise domain biases of prototypes gradually during training. Experimental results on standard benchmarks including GTA5 to Cityscapes and SYNTHIA to Cityscapes demonstrate the effectiveness of our framework.



Paperid:1230
Authors:Shichao Dong; Guosheng Lin; Tzu-Yi Hung
Title: Learning Regional Purity for Instance Segmentation on 3D Point Clouds
Abstract:
3D instance segmentation is a fundamental task for scene understanding, with a variety of applications in robotics and AR/VR. Many proposal-free methods have been proposed recently for this task, with remarkable results and high efficiency. However, these methods heavily rely on instance centroid regression and do not explicitly detect object boundaries, thus may mistakenly group nearby objects into the same clusters in some scenarios. In this paper, we define a novel concept of ”regional purity"" as the percentage of neighboring points belonging to the same instance within a fixed-radius 3D space. Intuitively, it indicates the likelihood of a point belonging to the boundary area. To evaluate the feasibility of predicting regional purity, we design a strategy to build a random scene toy dataset based on existing training data. Besides, using toy data is a ”free"" way of data augmentation on learning regional purity, which eliminates the burdens of additional real data. We propose Regional Purity Guided Network (RPGN), which has separate branches for predicting semantic class, regional purity, offset, and size. Predicted regional purity information is utilized to guide our clustering algorithm. Experimental results demonstrate that using regional purity can simultaneously prevent under-segmentation and over-segmentation problems during clustering.



Paperid:1231
Authors:Shuo Lei; Xuchao Zhang; Jianfeng He; Fanglan Chen; Bowen Du; Chang-Tien Lu
Title: Cross-Domain Few-Shot Semantic Segmentation
Abstract:
Few-shot semantic segmentation aims at learning to segment a novel object class with only a few annotated examples. Most existing methods consider a setting where base classes are sampled from the same domain as the novel classes. However, in many applications, collecting sufficient training data for meta-learning is infeasible or impossible. In this paper, we extend few-shot semantic segmentation to a new task, called Cross-Domain Few-Shot Semantic Segmentation (CD-FSS), which aims to generalize the meta-knowledge from domains with sufficient training labels to low-resource domains. Moreover, a new benchmark for the CD-FSS task is established and characterized by a task difficulty measurement. We evaluate both representative few-shot segmentation methods and transfer learning based methods on the proposed benchmark and find that current few-shot segmentation methods fail to address CD-FSS. To tackle the challenging CD-FSS problem, we propose a novel Pyramid-Anchor-Transformation based few-shot segmentation network (PATNet), in which domain-specific features are transformed into domain-agnostic ones for downstream segmentation modules to fast adapt to unseen domains. Our model outperforms the state-of-the-art few-shot segmentation method in CD-FSS by 8.49% and 10.61% average accuracies in 1-shot and 5-shot, respectively.



Paperid:1232
Authors:Yuehui Han; Le Hui; Haobo Jiang; Jianjun Qian; Jin Xie
Title: Generative Subgraph Contrast for Self-Supervised Graph Representation Learning
Abstract:
Contrastive learning has shown great promise in the field of graph representation learning. By manually constructing positive/negative samples, most graph contrastive learning methods rely on the vector inner product based similarity metric to distinguish the samples for graph representation. However, the handcrafted sample construction (e.g., the perturbation on the nodes or edges of the graph) may not effectively capture the intrinsic local structures of the graph. Also, the vector inner product based similarity metric cannot fully exploit the local structures of the graph to characterize the graph difference well. To this end, in this paper, we propose a novel adaptive subgraph generation based contrastive learning framework for efficient and robust self-supervised graph representation learning, and the optimal transport distance is utilized as the similarity metric between the subgraphs. It aims to generate contrastive samples by capturing the intrinsic structures of the graph and distinguish the samples based on the features and structures of subgraphs simultaneously. Specifically, for each center node, by adaptively learning relation weights to the nodes of the corresponding neighborhood, we first develop a network to generate the interpolated subgraph. We then construct the positive and negative pairs of subgraphs from the same and different nodes, respectively. Finally, we employ two types of optimal transport distances (i.e., Wasserstein distance and Gromov-Wasserstein distance) to construct the structured contrastive loss. Extensive node classification experiments on benchmark datasets verify the effectiveness of our graph contrastive learning method.



Paperid:1233
Authors:Yabo Chen; Yuchen Liu; Dongsheng Jiang; Xiaopeng Zhang; Wenrui Dai; Hongkai Xiong; Qi Tian
Title: SdAE: Self-Distillated Masked Autoencoder
Abstract:
With the development of generative-based self-supervised learning (SSL) approaches like BeiT and MAE, how to learn good representations by masking random patches of the input image and reconstructing the missing information has grown in concern. However, BeiT and PeCo need a “pre-pretraining” stage to produce discrete codebooks for masked patches representing. MAE does not require a pre-training codebook process, but setting pixels as reconstruction targets may introduce an optimization gap between pre-training and downstream tasks that good reconstruction quality may not always lead to the high descriptive capability for the model. Considering the above issues, in this paper, we propose a simple Self-distillated masked AutoEncoder network, namely SdAE. SdAE consists of a student branch using an encoder-decoder structure to reconstruct the missing information, and a teacher branch producing latent representation of masked tokens. We also analyze how to build good views for the teacher branch to produce latent representation from the perspective of information bottleneck. After that, we propose a multi-fold masking strategy to provide multiple masked views with balanced information for boosting the performance, which can also reduce the computational complexity. Our approach generalizes well: with only 300 epochs pre-training, a vanilla ViT-Base model achieves an 84.1% fine-tuning accuracy on ImageNet-1k classification, 48.6 mIOU on ADE20K segmentation, and 48.9 mAP on COCO detection with only 300 epochs pre-training, which surpasses other methods by a considerable margin. Code is available at https://github.com/AbrahamYabo/SdAE.



Paperid:1234
Authors:Mehmet Aygün; Oisin Mac Aodha
Title: Demystifying Unsupervised Semantic Correspondence Estimation
Abstract:
We explore semantic correspondence estimation through the lens of unsupervised learning. We thoroughly evaluate several recently proposed unsupervised methods across multiple challenging datasets using a standardized evaluation protocol where we vary factors such as the backbone architecture, the pre-training strategy, and the pre-training and finetuning datasets. To better understand the failure modes of these methods, and in order to provide a clearer path for improvement, we provide a new diagnostic framework along with a new performance metric that is better suited to the semantic matching task. Finally, we introduce a new unsupervised correspondence approach which utilizes the strength of pre-trained features while encouraging better matches during training. This results in significantly better matching performance compared to current state-of-the-art methods.



Paperid:1235
Authors:Yen-Cheng Liu; Chih-Yao Ma; Xiaoliang Dai; Junjiao Tian; Peter Vajda; Zijian He; Zsolt Kira
Title: Open-Set Semi-Supervised Object Detection
Abstract:
Recent developments for Semi-Supervised Object Detection (SSOD) have shown the promise of leveraging unlabeled data to improve an object detector. However, thus far these methods have assumed that the unlabeled data does not contain out-of-distribution (OOD) classes, which is unrealistic with larger-scale unlabeled datasets. In this paper, we consider a more practical yet challenging problem, Open-Set Semi-Supervised Object Detection (OSSOD). We first find the existing SSOD method obtains a lower performance gain in open-set conditions, and this is caused by the semantic expansion, where the distracting OOD objects are mispredicted as in-distribution pseudo-labels for the semi-supervised training. To address this problem, we consider online and offline OOD detection modules, which are integrated with SSOD methods. With the extensive studies, we found that leveraging an offline OOD detector based on a self-supervised vision transformer performs favorably against online OOD detectors due to its robustness to the interference of pseudo-labeling. In the experiment, our proposed framework effectively addresses the semantic expansion issue and shows consistent improvements on many OSSOD benchmarks, including large-scale COCO-OpenImages. We also verify the effectiveness of our framework under different OSSOD conditions, including varying numbers of in-distribution classes, different degrees of supervision, and different combinations of unlabeled sets.



Paperid:1236
Authors:Hengtong Hu; Lingxi Xie; Xinyue Huo; Richang Hong; Qi Tian
Title: Vibration-Based Uncertainty Estimation for Learning from Limited Supervision
Abstract:
We investigate the problem of estimating uncertainty for training data, so that deep neural networks can make use of the results for learning from limited supervision. However, both prediction probability and entropy estimate uncertainty from the instantaneous information. In this paper, we present a novel approach that measures uncertainty from the vibration of sequential data, e.g., the output probability during the training procedure. The key observation is that, a training sample that suffers heavier vibration often offers richer information when it is manually labeled. Motivated by Bayesian theory, we sample the sequences from the latter part of training. We make use of the Fourier Transformation to measure the extent of vibration, deriving a powerful tool that can be used for semi-supervised, active learning, and one-bit supervision. Experiments on the CIFAR10, CIFAR100, mini-ImageNet and ImageNet datasets validate the effectiveness of our approach.



Paperid:1237
Authors:Jogendra Nath Kundu; Suvaansh Bhambri; Akshay Kulkarni; Hiran Sarkar; Varun Jampani; R. Venkatesh Babu
Title: Concurrent Subsidiary Supervision for Unsupervised Source-Free Domain Adaptation
Abstract:
The prime challenge in unsupervised domain adaptation (DA) is to mitigate the domain shift between the source and target domains. Prior DA works show that pretext tasks could be used to mitigate this domain shift by learning domain invariant representations. However, in practice, we find that most existing pretext tasks are ineffective against other established techniques. Thus, we theoretically analyze how and when a subsidiary pretext task could be leveraged to assist the goal task of a given DA problem and develop objective subsidiary task suitability criteria. Based on this criteria, we devise a novel process of sticker intervention and cast sticker classification as a supervised subsidiary DA problem concurrent to the goal task unsupervised DA. Our approach not only improves goal task adaptation performance, but also facilitates privacy-oriented source-free DA i.e. without concurrent source-target access. Experiments on the standard Office-31, Office-Home, DomainNet, and VisDA benchmarks demonstrate our superiority for both single-source and multi-source source-free DA. Our approach also complements existing non-source-free works, achieving leading performance.



Paperid:1238
Authors:Jun Wei; Sheng Wang; S. Kevin Zhou; Shuguang Cui; Zhen Li
Title: Weakly Supervised Object Localization through Inter-class Feature Similarity and Intra-Class Appearance Consistency
Abstract:
Weakly supervised object localization (WSOL) aims at detecting objects through only image-level labels. Class activation maps (CAMs) are the commonly used features for WSOL. However, existing CAM-based methods tend to excessively pursue discriminative features for object recognition and hence ignore the feature similarities among different categories, thereby leading to CAMs incomplete for object localization. In addition, CAMs are sensitive to background noise due to over-dependence on the holistic classification. In this paper, we propose a simple but effective WSOL model (named ISIC) through Inter-class feature Similarity and Intra-class appearance Consistency. In practice, our ISIC model first proposes the inter-class feature similarity (ICFS) loss against the original cross entropy loss. Such an ICFS loss sufficiently leverages the shared features together with the discriminative features between different categories, which significantly reduces the model over-fitting risk to background noise and brings more complete object masks. Besides, instead of CAMs, a non-negative matrix factorization mask module is applied to extract object masks from multiple intra-class images. Thanks to intra-class appearance consistency, the achieved pseudo masks are more complete and robust. As a result, extensive experiments confirm that our ISIC model achieves state-of-the-art on both CUB-200 and ImageNet-1K benchmarks i.e., 97.3% and 70.0% GT-Known localization accuracy, respectively.



Paperid:1239
Authors:Huy V. Vo; Oriane Siméoni; Spyros Gidaris; Andrei Bursuc; Patrick Pérez; Jean Ponce
Title: Active Learning Strategies for Weakly-Supervised Object Detection
Abstract:
Object detectors trained with weak annotations are affordable alternatives to fully-supervised counterparts. However, there is still a significant performance gap between them. We propose to narrow this gap by fine-tuning a base pre-trained weakly-supervised detector with a few fully-annotated samples automatically selected from the training set using “box-in-box” (BiB), a novel active learning strategy designed specifically to address the well-documented failure modes of weakly-supervised detectors. Experiments on the VOC07 and COCO benchmarks show that BiB outperforms other active learning techniques and significantly improves the base weakly-supervised detector’s performance with only a few fully-annotated images per class. BiB reaches 97% of the performance of fully-supervised Fast RCNN with only 10% of fully-annotated images on VOC07. On COCO, using on average 10 fully-annotated images per class, or equivalently 1% of the training set, BiB also reduces the performance gap (in AP) between the weakly-supervised detector and the fully-supervised Fast RCNN by over 70%, showing a good trade-off between performance and data efficiency. Our code is publicly available at https://github.com/huyvvo/BiB.



Paperid:1240
Authors:Xiaotong Li; Yixiao Ge; Kun Yi; Zixuan Hu; Ying Shan; Ling-Yu Duan
Title: Mc-BEiT: Multi-Choice Discretization for Image BERT Pre-training
Abstract:
Image BERT pre-training with masked image modeling (MIM) becomes a popular practice to cope with self-supervised representation learning. A seminal work, BEiT, casts MIM as a classification task with a visual vocabulary, tokenizing the continuous visual signals into discrete vision tokens using a pre-learned dVAE. Despite a feasible solution, the improper discretization hinders further improvements of image pre-training. Since image discretization has no ground-truth answers, we believe that the masked patch should not be assigned with a unique token id even if a better “tokenizer” can be obtained. In this work, we introduce an improved BERT-style image pre-training method, namely mc-BEiT, which performs MIM proxy tasks towards eased and refined multi-choice training objectives. Specifically, the multi-choice supervision for the masked image patches is formed by the soft probability vectors of the discrete token ids, which are predicted by the off-the-shelf image “tokenizer” and further refined by high-level inter-patch perceptions resorting to the observation that similar patches should share their choices. Extensive experiments on classification, segmentation, and detection tasks demonstrate the superiority of our method, e.g., the pre-trained ViT-B achieves 84.1% top-1 fine-tuning accuracy on ImageNet-1K classification, 49.2% AP^b and 44.0% AP^m of object detection and instance segmentation on COCO, 50.8% mIOU on ADE20K semantic segmentation, outperforming the competitive counterparts. The code is available at https://github.com/lixiaotong97/mc-BEiT.



Paperid:1241
Authors:Xiaoyi Dong; Jianmin Bao; Ting Zhang; Dongdong Chen; Weiming Zhang; Lu Yuan; Dong Chen; Fang Wen; Nenghai Yu
Title: Bootstrapped Masked Autoencoders for Vision BERT Pretraining
Abstract:
We propose bootstrapped masked autoencoders (BootMAE), a new approach for vision BERT pretraining. BootMAE improves the original masked autoencoders (MAE) with two core designs: 1) momentum encoder that provides online feature as extra BERT prediction targets; 2) target-aware decoder that tries to reduce the pressure on the encoder to memorize target-specific information in BERT pretraining. The first design is motivated by the observation that using a pretrained MAE to extract the features as the BERT prediction target for masked tokens can achieve better pretraining performance. Therefore, we add a momentum encoder in parallel with the original MAE encoder, which bootstraps the pretraining performance by using its own representation as the BERT prediction target. In the second design, we introduce target-specific information (e.g., pixel values of unmasked patches) from the encoder directly to the decoder to reduce the pressure on the encoder of memorizing the target-specific information. Thus, the encoder focuses on semantic modeling, which is the goal of BERT pretraining, and does not need to waste its capacity in memorizing the information of unmasked tokens related to the prediction target. Through extensive experiments, our BootMAE achieves 84.2% Top-1 accuracy on ImageNet-1K with ViT-B backbone, outperforming MAE by +0.8% under the same pre-training epochs. BootMAE also gets +1.0 mIoU improvements on semantic segmentation on ADE20K and +1.3 box AP, +1.4 mask AP improvement on object detection and segmentation on COCO dataset. Code is released at https://github.com/LightDXY/BootMAE"



Paperid:1242
Authors:Bo Pang; Yifan Zhang; Yaoyi Li; Jia Cai; Cewu Lu
Title: Unsupervised Visual Representation Learning by Synchronous Momentum Grouping
Abstract:
In this paper, we propose a genuine group-level contrastive visual representation learning method whose linear evaluation performance on ImageNet surpasses the vanilla supervised learning. Two mainstream unsupervised learning schemes are the instance-level contrastive framework and clustering-based schemes. The former adopts the extremely fine-grained instance-level discrimination whose supervisory signal is not efficient due to the false negatives. Though the latter solves this, they commonly come with some restrictions affecting the performance. To integrate their advantages, we design the SMoG method. SMoG follows the framework of contrastive learning but replaces the contrastive unit from instance to group, mimicking clustering-based methods. To achieve this, we propose the momentum grouping scheme which synchronously conducts feature grouping with representation learning. In this way, SMoG solves the problem of supervisory signal hysteresis which the clustering-based method usually faces, and reduces the false negatives of instance contrastive methods. We conduct exhaustive experiments to show that SMoG works well on both CNN and Transformer backbones. Results prove that SMoG has surpassed the current SOTA unsupervised representation learning methods. Moreover, its linear evaluation results surpass the performances obtained by vanilla supervised learning and the representation can be well transferred to downstream tasks.



Paperid:1243
Authors:Oindrila Saha; Zezhou Cheng; Subhransu Maji
Title: Improving Few-Shot Part Segmentation Using Coarse Supervision
Abstract:
A significant bottleneck in training deep networks for part segmentation is the cost of obtaining detailed annotations. We propose a framework to exploit coarse labels such as figure-ground masks and keypoint locations that are readily available for some categories to improve part segmentation models. A key challenge is that these annotations were collected for different tasks and with different labeling styles and cannot be readily mapped to the part labels. To this end, we propose to jointly learn the dependencies between labeling styles and the part segmentation model, allowing us to utilize supervision from diverse labels. To evaluate our approach we develop a benchmark on the Caltech-UCSD birds and OID Aircraft dataset. Our approach outperforms baselines based on multi-task learning, semi-supervised learning, and competitive methods relying on loss functions manually designed to exploit sparse-supervision.



Paperid:1244
Authors:Ioannis Kakogeorgiou; Spyros Gidaris; Bill Psomas; Yannis Avrithis; Andrei Bursuc; Konstantinos Karantzalos; Nikos Komodakis
Title: What to Hide from Your Students: Attention-Guided Masked Image Modeling
Abstract:
Transformers and masked language modeling are quickly being adopted and explored in computer vision as vision transformers and masked image modeling (MIM). In this work, we argue that image token masking differs from token masking in text, due to the amount and correlation of tokens in an image. In particular, to generate a challenging pretext task for MIM, we advocate a shift from random masking to informed masking. We develop and exhibit this idea in the context of distillation-based MIM, where a teacher transformer encoder generates an attention map, which we use to guide masking for the student. We thus introduce a novel masking strategy, called attention-guided masking (AttMask), and we demonstrate its effectiveness over random masking for dense distillation-based MIM as well as plain distillation-based self-supervised learning on classification tokens. We confirm that AttMask accelerates the learning process and improves the performance on a variety of downstream tasks. We provide the implementation code at https://github.com/gkakogeorgiou/attmask.



Paperid:1245
Authors:Junsong Fan; Zhaoxiang Zhang; Tieniu Tan
Title: Pointly-Supervised Panoptic Segmentation
Abstract:
In this paper, we propose a new approach to applying point-level annotations for weakly-supervised panoptic segmentation. Instead of the dense pixel-level labels used by fully supervised methods, point-level labels only provide a single point for each target as supervision, significantly reducing the annotation burden. We formulate the problem in an end-to-end framework by simultaneously generating panoptic pseudo-masks from point-level labels and learning from them. To tackle the core challenge, i.e., panoptic pseudo-mask generation, we propose a principled approach to parsing pixels by minimizing pixel-to-point traversing costs, which model semantic similarity, low-level texture cues, and high-level manifold knowledge to discriminate panoptic targets. We conduct experiments on the Pascal VOC and the MS COCO datasets to demonstrate the approach’s effectiveness and show state-of-the-art performance in the weakly-supervised panoptic segmentation problem. Codes are available at https://github.com/BraveGroup/PSPS.git.



Paperid:1246
Authors:Longhui Wei; Lingxi Xie; Wengang Zhou; Houqiang Li; Qi Tian
Title: MVP: Multimodality-Guided Visual Pre-training
Abstract:
Recently, masked image modeling (MIM) has become a promising direction for visual pre-training. In the context of vision transformers, MIM learns effective visual representation by aligning the token-level features with a pre-defined space (e,g,, BEIT used a d-VAE trained on a large image corpus as the tokenizer). In this paper, we go one step further by introducing guidance from other modalities and validating that such additional knowledge leads to impressive gains for visual pre-training. The proposed approach is named Multimodality-guided Visual Pre-training (MVP), in which we replace the tokenizer with the vision branch of CLIP, a vision-language model pre-trained on 400 million image-text pairs. We demonstrate the effectivenss of MVP by performing standard experiments, i.e., pre-training the ViT models on ImageNet and fine-tuning them on a series of downstream visual recognition tasks. In particular, pre-training ViT-Base/16 for 300 epochs, MVP reports a 52.4% mIOU on ADE-20K, surpassing BEIT (the baseline and previous state-of-the-art) with an impressive margin of 6.8%.



Paperid:1247
Authors:Wen-Yan Lin; Zhonghang Liu; Siying Liu
Title: Locally Varying Distance Transform for Unsupervised Visual Anomaly Detection
Abstract:
Unsupervised anomaly detection on image data is notoriously unstable. We believe this is because many classical anomaly detectors implicitly assume data is low dimensional. However, image data is always high dimensional. Images can be projected to a low dimensional embedding but such projections rely on global transformations that truncate minor variations. As anomalies are rare, the final embedding often lacks the key variations needed to distinguish anomalies from normal instances. This paper proposes a new embedding using a set of locally varying data projections, with each projection responsible for persevering the variations that distinguish a local cluster of instances from all other instances. The locally varying embedding ensures the variations that distinguish anomalies are preserved, while simultaneously allowing the probability that an instance belongs to a cluster, to be statistically inferred from the one-dimensional, local projection associated with the cluster. Statistical agglomeration of an instance’s cluster membership probabilities, creates a global measure of its affinity to the dataset and causes anomalies to emerge, as instances whose affinity scores are surprisingly low.



Paperid:1248
Authors:Lukas Hoyer; Dengxin Dai; Luc Van Gool
Title: HRDA: Context-Aware High-Resolution Domain-Adaptive Semantic Segmentation
Abstract:
Unsupervised domain adaptation (UDA) aims to adapt a model trained on the source domain (e.g. synthetic data) to the target domain (e.g. real-world data) without requiring further annotations on the target domain. This work focuses on UDA for semantic segmentation as real-world pixel-wise annotations are particularly expensive to acquire. As UDA methods for semantic segmentation are usually GPU memory intensive, most previous methods operate only on downscaled images. We question this design as low-resolution predictions often fail to preserve fine details. The alternative of training with random crops of high-resolution images alleviates this problem but falls short in capturing long-range, domain-robust context information. Therefore, we propose HRDA, a multi-resolution training approach for UDA, that combines the strengths of small high-resolution crops to preserve fine segmentation details and large low-resolution crops to capture long-range context dependencies with a learned scale attention, while maintaining a manageable GPU memory footprint. HRDA enables adapting small objects and preserving fine segmentation details. It significantly improves the state-of-the-art performance by 5.5 mIoU for GTA-to-Cityscapes and 4.9 mIoU for Synthia-to-Cityscapes, resulting in unprecedented 73.8 and 65.8 mIoU, respectively. The implementation is available at https://github.com/lhoyer/HRDA.



Paperid:1249
Authors:Yang Zou; Jongheon Jeong; Latha Pemula; Dongqing Zhang; Onkar Dabeer
Title: SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation
Abstract:
Visual anomaly detection is commonly used in industrial quality inspection. In this paper, we present a new dataset as well as a new self-supervised learning method for ImageNet pre-training to improve anomaly detection and segmentation in 1-class and 2-class 5/10/high-shot training setups. We release the Visual Anomaly (VisA) Dataset consisting of 10,821 high-resolution color images (9,621 normal and 1,200 anomalous samples) covering 12 objects in 3 domains, making it the largest industrial anomaly detection dataset to date. Both image and pixel-level labels are provided. We also propose a new self-supervised framework - SPot-the-difference (SPD) - which can regularize contrastive self-supervised pre-training, such as SimSiam, MoCo and SimCLR, to be more suitable for anomaly detection tasks. Our experiments on VisA and MVTec-AD dataset show that SPD consistently improves these contrastive pre-training baselines and even the supervised pre-training. For example, SPD improves Area Under the Precision-Recall curve (AU-PR) for anomaly segmentation by 5.9% and 6.8% over SimSiam and supervised pre-training respectively in the 2-class high-shot regime. We open-source the project at http://github.com/amazon-research/spot-diff.



Paperid:1250
Authors:Yuhui Quan; Xinran Qin; Tongyao Pang; Hui Ji
Title: Dual-Domain Self-Supervised Learning and Model Adaption for Deep Compressive Imaging
Abstract:
Deep learning has been one promising tool for compressive imaging whose task is to reconstruct latent images from their compressive measurements. Aiming at addressing the limitations of supervised deep learning-based methods caused by their prerequisite on the ground truths of latent images, this paper proposes an unsupervised approach that trains a deep image reconstruction model using only a set of compressive measurements. The training is self-supervised in the domain of measurements and the domain of images, using a double-head noise-injected loss with a sign-flipping-based noise generator. In addition, the proposed scheme can also be used for efficiently adapting a trained model to a test sample for further improvement, with much less overhead than existing internal learning methods. Extensive experiments show that the proposed approach provides noticeable performance gain over existing unsupervised methods and competes well against the supervised ones.



Paperid:1251
Authors:Xudong Wang; Long Lian; Stella X. Yu
Title: Unsupervised Selective Labeling for More Effective Semi-Supervised Learning
Abstract:
Given an unlabeled dataset and an annotation budget, we study how to selectively label a fixed number of instances so that semi-supervised learning (SSL) on such a partially labeled dataset is most effective. We focus on selecting the right data to label, in addition to usual SSL’s propagating labels from given labeled data to the rest unlabeled data. This instance selection task is challenging, as without any labeled data we don’t know what the objective of learning should be. Intuitively, no matter what the downstream task is, instances to be labeled must be representative and diverse: The former would facilitate label propagation to unlabeled data, whereas the latter would ensure coverage of the entire dataset. We capture this idea by selecting cluster prototypes, either in a pretrained feature space, or along with the feature to be optimized for instance discrimination, both without labels. Our unsupervised selective labeling consistently improves SSL methods over state-of-the-art active learning given labeled data, by 8-25x in label efficiency. For example, it boosts FixMatch by 10%(14%) in accuracy on CIFAR-10(ImageNet-1K) with 0.08%(0.2%) labeled data, demonstrating that small computation spent on what data to label brings significant gain especially under a low annotation budget. Our work sets a new standard for practical SSL.



Paperid:1252
Authors:Simone Rossetti; Damiano Zappia; Marta Sanzari; Marco Schaerf; Fiora Pirri
Title: Max Pooling with Vision Transformers Reconciles Class and Shape in Weakly Supervised Semantic Segmentation
Abstract:
Weakly Supervised Semantic Segmentation (WSSS) research has explored many directions to improve the typical pipeline CNN plus class activation maps (CAM) plus refinements, given the image-class label as the only supervision. Though the gap with the fully supervised methods is reduced, further abating the spread seems unlikely within this framework. On the other hand, WSSS methods based on Vision Transformers (ViT) have not yet explored valid alternatives to CAM. ViT features have been shown to retain a scene layout, and object boundaries in self-supervised learning. To confirm these findings, we prove that the advantages of transformers in self-supervised methods are further strengthened by Global Max Pooling (GMP), which can leverage patch features to negotiate pixel-label probability with class probability. This work proposes a new WSSS method dubbed ViT-PCM (ViT Patch-Class Mapping), not based on CAM. The end-to-end presented network learns with a single optimization process, refined shape and proper localization for segmentation masks. Our model outperforms the state-of-the-art on baseline pseudo-masks (BPM), where we achieve 69.3\% mIoU on PascalVOC 2012 validation set. We show that our approach has the least set of parameters, though obtaining higher accuracy than all other approaches. In a sentence, quantitative and qualitative results of our method reveal that ViT-PCM is an excellent alternative to CNN-CAM based architectures.



Paperid:1253
Authors:Wenwei Zhang; Jiangmiao Pang; Kai Chen; Chen Change Loy
Title: Dense Siamese Network for Dense Unsupervised Learning
Abstract:
This paper presents Dense Siamese Network (DenseSiam), a simple unsupervised learning framework for dense prediction tasks. It learns visual representations by maximizing the similarity between two views of one image with two types of consistency, i.e., pixel consistency and region consistency. Concretely, DenseSiam first maximizes the pixel level spatial consistency according to the exact location correspondence in the overlapped area. It also extracts a batch of region embeddings that correspond to some sub-regions in the overlapped area to be contrasted for region consistency. In contrast to previous methods that require negative pixel pairs, momentum encoders or heuristic masks, DenseSiam benefits from the simple Siamese network and optimizes the consistency of different granularities. It also proves that the simple location correspondence and interacted region embeddings are effective enough to learn the similarity. We apply DenseSiam on ImageNet and obtain competitive improvements on various downstream tasks. We also show that only with some extra task-specific losses, the simple framework can directly conduct dense prediction tasks. On an existing unsupervised semantic segmentation benchmark, it surpasses state-of-the-art segmentation methods by 2.1 mIoU with 28% training costs. Code and models are released at https://github.com/ZwwWayne/DenseSiam"



Paperid:1254
Authors:Jie Qin; Jie Wu; Ming Li; Xuefeng Xiao; Min Zheng; Xingang Wang
Title: Multi-Granularity Distillation Scheme towards Lightweight Semi-Supervised Semantic Segmentation
Abstract:
Albeit with varying degrees of progress in the field of Semi-Supervised Semantic Segmentation, most of its recent successes are involved in unwieldy models and the lightweight solution is still not yet explored. We find that existing knowledge distillation techniques pay more attention to pixel-level concepts from labeled data, which fails to take more informative cues within unlabeled data into account. Consequently, we offer the first attempt to provide lightweight SSSS models via a novel multi-granularity distillation (MGD) scheme, where multi-granularity is captured from three aspects: i) complementary teacher structure; ii) labeled-unlabeled data cooperative distillation; iii) hierarchical and multi-levels loss setting. Specifically, MGD is formulated as a labeled-unlabeled data cooperative distillation scheme, which helps to take full advantage of diverse data characteristics that are essential in the semi-supervised setting. Image-level semantic-sensitive loss, region-level content-aware loss, and pixel-level consistency loss are set up to enrich hierarchical distillation abstraction via structurally complementary teachers. Experimental results on PASCAL VOC2012 and Cityscapes reveal that MGD can outperform the competitive approaches by a large margin under diverse partition protocols. For example, the performance of ResNet-18 and MobileNet-v2 backbone is boosted by 11.5% and 4.6% respectively under 1/16 partition protocol on Cityscapes. Although the FLOPs of the model backbone is compressed by 3.4-5.3× (ResNet-18) and 38.7-59.6× (MobileNetv2), the model manages to achieve satisfactory segmentation results.



Paperid:1255
Authors:Feng Wang; Huiyu Wang; Chen Wei; Alan Yuille; Wei Shen
Title: CP2: Copy-Paste Contrastive Pretraining for Semantic Segmentation
Abstract:
Recent advances in self-supervised contrastive learning yield good image-level representation, which favors classification tasks but usually neglects pixel-level detailed information, leading to unsatisfactory transfer performance to dense prediction tasks such as semantic segmentation. In this work, we propose a pixel-wise contrastive learning method called CP2 (Copy-Paste Contrastive Pretraining), which facilitates both image- and pixel-level representation learning and therefore is more suitable for downstream dense prediction tasks. In detail, we copy-paste a random crop from an image (the foreground) onto different background images and pretrain a semantic segmentation model with the objective of 1) distinguishing the foreground pixels from the background pixels, and 2) identifying the composed images that share the same foreground. Experiments show the strong performance of CP2 in downstream semantic segmentation: By finetuning CP2 pretrained models on PASCAL VOC 2012, we obtain 78.6% mIoU with a ResNet-50 and 79.5% with a ViT-S.



Paperid:1256
Authors:Qi Wei; Haoliang Sun; Xiankai Lu; Yilong Yin
Title: Self-Filtering: A Noise-Aware Sample Selection for Label Noise with Confidence Penalization
Abstract:
Sample selection is an effective strategy to mitigate the effect of label noise in robust learning. Typical strategies commonly apply the small-loss criterion to identify clean samples. However, those samples lying around the decision boundary with large losses usually entangle with noisy examples, which would be discarded with this criterion, leading to the heavy degeneration of the generalization performance. In this paper, we propose a novel selection strategy, \textbf{S}elf-\textbf{F}il\textbf{t}ering (SFT), that utilizes the fluctuation of noisy examples in historical predictions to filter them, which can avoid the selection bias of the small-loss criterion for the boundary examples. Specifically, we introduce a memory bank module that stores the historical predictions of each example and dynamically updates to support the selection for the subsequent learning iteration. Besides, to reduce the accumulated error of the sample selection bias of SFT, we devise a regularization term to penalize the confident output distribution. By increasing the weight of the misclassified categories with this term, the loss function is robust to label noise in mild conditions. We conduct extensive experiments on three benchmarks with variant noise types and achieve the new state-of-the-art. Ablation studies and further analysis demonstrate the virtue of SFT for sample selection in robust learning.



Paperid:1257
Authors:Yue Duan; Lei Qi; Lei Wang; Luping Zhou; Yinghuan Shi
Title: RDA: Reciprocal Distribution Alignment for Robust Semi-Supervised Learning
Abstract:
In this work, we propose Reciprocal Distribution Alignment (RDA) to address semi-supervised learning (SSL), which is a hyperparameter-free framework that is independent of confidence threshold and works with both the matched (conventionally) and the mismatched class distributions. Distribution mismatch is an often overlooked but more general SSL scenario where the labeled and the unlabeled data do not fall into the identical class distribution. This may lead to the model not exploiting the labeled data reliably and drastically degrade the performance of SSL methods, which could not be rescued by the traditional distribution alignment. In RDA, we enforce a reciprocal alignment on the distributions of the predictions from two classifiers predicting pseudo-labels and complementary labels on the unlabeled data. These two distributions, carrying complementary information, could be utilized to regularize each other without any prior of class distribution. Moreover, we theoretically show that RDA maximizes the input-output mutual information. Our approach achieves promising performance in SSL under a variety of scenarios of mismatched distributions, as well as the conventional matched SSL setting. The code is available at: https://github.com/NJUyued/RDA4RobustSSL.



Paperid:1258
Authors:Tarun Kalluri; Astuti Sharma; Manmohan Chandraker
Title: MemSAC: Memory Augmented Sample Consistency for Large Scale Domain Adaptation
Abstract:
Practical real world datasets with plentiful categories introduce new challenges for unsupervised domain adaptation like small inter-class discriminability, that existing approaches relying on domain invariance alone cannot handle sufficiently well. In this work we propose MemSAC, which exploits sample level similarity across source and target domains to achieve discriminative transfer, along with architectures that scale to a large number of categories. For this purpose, we first introduce a memory augmented approach to efficiently extract pairwise similarity relations between labeled source and unlabeled target domain instances, suited to handle an arbitrary number of classes. Next, we propose and theoretically justify a novel variant of the contrastive loss to promote local consistency among within-class cross domain samples while enforcing separation between classes, thus preserving discriminative transfer from source to target. We validate the advantages of MemSAC with significant improvements over previous state-of-the-art on multiple challenging transfer tasks designed for large-scale adaptation, such as DomainNet with 345 classes and fine-grained adaptation on Caltech-UCSD birds dataset with 200 classes. We also provide in-depth analysis and insights into the effectiveness of MemSAC.



Paperid:1259
Authors:Wenda Zhao; Fei Wei; You He; Huchuan Lu
Title: United Defocus Blur Detection and Deblurring via Adversarial Promoting Learning
Abstract:
Understanding blur from a single defocused image contains two tasks of defocus detection and deblurring. This paper makes the earliest effort to jointly learn both defocus detection and deblurring without using pixel-level defocus detection annotation and paired defocus deblurring ground truth. We build on the observation that these two tasks are supplementary to each other: Defocus detection can segment the focused area from the defocused image to guide the defocus deblurring; Conversely, to achieve better defocus deblurring, an accurate defocus detection as the guide is essential. Therefore, we implement an adversarial promoting learning framework to jointly handle defocus detection and defocus deblurring. Specifically, a defocus detection generator $G_{ws}$ is implemented to represent the defocused image as a layered composition of two elements: defocused image $I_{df}$ and a focused image $I_f$. Then, $I_{df}$ and $I_f$ are fed into a self-referenced defocus deblurring generator $G_{sr}$ to generate a deblurred image. Two generators of $G_{ws}$ and $G_{sr}$ are optimized alternately in an adversarial manner against a discriminator $D$ with unpaired realistic fully-clear images. Thus, $G_{sr}$ will produce a deblurred image to fool $D$, and $G_{ws}$ is forced to generate an accurate defocus detection map to effectively guide $G_{sr}$. Comprehensive experiments on two defocus detection datasets and one defocus deblurring dataset demonstrate the effectiveness of our framework. \textcolor[rgb]{0.00,0.50,0.50}{Code and model are available at: https://github.com/wdzhao123/APL.}"



Paperid:1260
Authors:Yun-Hao Cao; Peiqin Sun; Yechang Huang; Jianxin Wu; Shuchang Zhou
Title: Synergistic Self-Supervised and Quantization Learning
Abstract:
With the success of self-supervised learning (SSL), it has become a mainstream paradigm to fine-tune from self-supervised pretrained models to boost the performance on downstream tasks. However, we find that current SSL models suffer severe accuracy drops when performing low-bit quantization, prohibiting their deployment in resource-constrained applications. In this paper, we propose a method called synergistic self-supervised and quantization learning (SSQL) to pretrain quantization-friendly self-supervised models facilitating downstream deployment. SSQL contrasts the features of the quantized and full precision models in a self-supervised fashion, where the bit-width for the quantized model is randomly selected in each step. SSQL not only significantly improves the accuracy when quantized to lower bit-widths, but also boosts the accuracy of full precision models in most cases. By only training once, SSQL can then benefit various downstream tasks at different bit-widths simultaneously. Moreover, the bit-width flexibility is achieved without additional storage overhead, requiring only one copy of weights during training and inference. We theoretically analyze the optimization process of SSQL, and conduct exhaustive experiments on various benchmarks to further demonstrate the effectiveness of our method.



Paperid:1261
Authors:Zejia Weng; Xitong Yang; Ang Li; Zuxuan Wu; Yu-Gang Jiang
Title: Semi-Supervised Vision Transformers
Abstract:
We study the training of Vision Transformers for semi-supervised image classification. Transformers have recently demonstrated impressive performance on a multitude of supervised learning tasks. Surprisingly, we show Vision Transformers perform significantly worse than Convolutional Neural Networks when only a small set of labeled data is available. Inspired by this observation, we introduce a joint semi-supervised learning framework, Semiformer, which contains a transformer stream, a convolutional stream and a carefully designed fusion module for knowledge sharing between these streams. The convolutional stream is trained on limited labeled data and further used to generate pseudo labels to supervise the training of the transformer stream on unlabeled data. Extensive experiments on ImageNet demonstrate that Semiformer achieves 75.5% top-1 accuracy, outperforming the state-of-the-art by a clear margin. In addition, we show, among other things, Semiformer is a general framework that is compatible with most modern transformer and convolutional neural architectures. Code is available at https://github.com/wengzejia1/Semiformer.



Paperid:1262
Authors:Yun Xing; Dayan Guan; Jiaxing Huang; Shijian Lu
Title: Domain Adaptive Video Segmentation via Temporal Pseudo Supervision
Abstract:
Video semantic segmentation has achieved great progress under the supervision of large amounts of labelled training data. However, domain adaptive video segmentation, which can mitigate data labelling constraints by adapting from a labelled source domain toward an unlabelled target domain, is largely neglected. We design temporal pseudo supervision (TPS), a simple and effective method that explores the idea of consistency training for learning effective representations from unlabelled target videos. Unlike traditional consistency training that builds consistency in spatial space, we explore consistency training in spatiotemporal space by enforcing model consistency across augmented video frames which helps learn from more diverse target data. Specifically, we design cross-frame pseudo labelling to provide pseudo supervision from previous video frames while learning from the augmented current video frames. The cross-frame pseudo labelling encourages the network to produce high-certainty predictions, which facilitates consistency training with cross-frame augmentation effectively. Extensive experiments over multiple public datasets show that TPS is simpler to implement, much more stable to train, and achieves superior video segmentation accuracy as compared with the state-of-the-art. Code is available at https://github.com/xing0047/TPS.



Paperid:1263
Authors:Linfeng Li; Minyue Jiang; Yue Yu; Wei Zhang; Xiangru Lin; Yingying Li; Xiao Tan; Jingdong Wang; Errui Ding
Title: Diverse Learner: Exploring Diverse Supervision for Semi-Supervised Object Detection
Abstract:
Current state-of-the-art semi-supervised object detection methods (SSOD) typically adopt the teacher-student framework featured with pseudo labeling and Exponential Moving Average (EMA). Although the performance is desirable, many remaining issues still need to be resolved, for example: (1) the teacher updated by the student using EMA tends to lose its distinctiveness and hence generates similar predictions comparing with student and cause potential noise accumulation as the training proceeds; (2) the exploitation of pseudo labels still has much room for improvement. We present a diverse learner semi-supervised object detection framework to tackle these issues. Concretely, to maintain distinctiveness between teachers and students, our framework consists of two paired teacher-student models with diverse supervision strategy. In addition, we argue that the pseudo labels which are typically regarded as unreliable and obsoleted by many existing methods are of great value. A particular training strategy consisting of Multi-threshold Classification Loss (MTC) and Pseudo Label-Aware Erasing (PLAE) is hence designed to well explore the full set of all pseudo labels. Extensive experimental results show that our diverse teacher-student framework outperforms the previous state-of-the-art method on the MS-COCO dataset by 2.10%, 1.50% and 0.83% when training with only 1%, 5% and 10% labeled data, demonstrating the effectiveness of our proposed framework. Moreover, our approach also performs well with larger amount of data, e.g. using full COCO training set and 123K unlabeled images from COCO, reaching a new state-of-the-art performance of 44.86% mAP.



Paperid:1264
Authors:Lanxiao Li; Michael Heizmann
Title: A Closer Look at Invariances in Self-Supervised Pre-training for 3D Vision
Abstract:
Self-supervised pre-training for 3D vision has drawn increasing research interest in recent years. In order to learn informative representations, a lot of previous works exploit invariances of 3D data, e.g., perspective-invariance between views of the same scene, modality-invariance between depth and RGB images, format-invariance between point clouds and voxels. Although they have achieved promising results, previous researches lack a systematic and fair comparison of these invariances. To address this issue, our work, for the first time, introduces a unified framework, into which previous works fit. Based on the framework, we conduct extensive experiments and provide a closer look at the invariances in 3D pre-training. Also, we propose a simple but effective method that jointly pre-trains a 3D encoder and a depth map encoder using contrastive learning. Models pre-trained with our method gain significant performance boost in downstream tasks. For instance, a pre-trained VoteNet outperforms previous methods on SUN RGB-D and ScanNet object detection benchmarks with a clear margin.



Paperid:1265
Authors:Jiwon Kim; Youngjo Min; Daehwan Kim; Gyuseong Lee; Junyoung Seo; Kwangrok Ryoo; Seungryong Kim
Title: ConMatch: Semi-Supervised Learning with Confidence-Guided Consistency Regularization
Abstract:
We present a novel semi-supervised learning framework that intelligently leverages the consistency regularization between the model’s predictions from two strongly-augmented views of an image, weighted by a confidence of pseudo-label, dubbed ConMatch. While the latest semi-supervised learning methods use weakly- and strongly-augmented views of an image to define a directional consistency loss, how to define such direction for the consistency regularization between two strongly-augmented views remains unexplored. To account for this, we present novel confidence measures for pseudo-labels from strongly-augmented views by means of weakly-augmented view as an anchor in non-parametric and parametric approaches. Especially, in parametric approach, we present, for the first time, to learn the confidence of pseudo-label within the networks, which is learned with backbone model in an end-to-end manner. In addition, we also present a stage-wise training to boost the convergence of training. When incorporated in existing semi-supervised learners, ConMatch consistently boosts the performance. We conduct experiments to demonstrate the effectiveness of our ConMatch over the latest methods and provide extensive ablation studies. Code has been made publicly available at \url{https://github.com/JiwonCocoder/ConMatch"



Paperid:1266
Authors:Sungwon Han; Sungwon Park; Fangzhao Wu; Sundong Kim; Chuhan Wu; Xing Xie; Meeyoung Cha
Title: FedX: Unsupervised Federated Learning with Cross Knowledge Distillation
Abstract:
This paper presents FedX, an unsupervised federated learning framework. Our model learns unbiased representation from decentralized and heterogeneous local data. It employs a two-sided knowledge distillation with contrastive learning as a core component, allowing the federated system to function without requiring clients to share any data features. Furthermore, its adaptable architecture can be used as an add-on module for existing unsupervised algorithms in federated settings. Experiments show that our model improves performance significantly (1.58--5.52pp) on five unsupervised algorithms.



Paperid:1267
Authors:Zitong Huang; Yiping Bao; Bowen Dong; Erjin Zhou; Wangmeng Zuo
Title: W2N: Switching from Weak Supervision to Noisy Supervision for Object Detection
Abstract:
Weakly-supervised object detection (WSOD) aims to train an object detector only requiring the image-level annotations. Recently, some works have managed to select the accurate boxes generated from a well-trained WSOD network to supervise a semi-supervised detection framework for better performance. However, these approaches simply divide the training set into labeled and unlabeled sets according to the image-level criteria, such that sufficient mislabeled or wrongly localized box predictions are chosen as pseudo ground-truths, resulting in a sub-optimal solution of detection performance. To overcome this issue, we propose a novel WSOD framework with a new paradigm that switches from weak supervision to noisy supervision (W2N).Generally, with given pseudo ground-truths generated from the well-trained WSOD network, we propose a two-module iterative training algorithm to refine pseudo labels and supervise better object detector progressively. In the localization adaptation module, we propose a regularization loss to reduce the proportion of discriminative parts in original pseudo ground-truths, obtaining better pseudo ground-truths for further training. In the semi-supervised module, we propose a two tasks instance-level split method to select high-quality labels for training a semi-supervised detector. Experimental results on different benchmarks verify the effectiveness of W2N, and our W2N outperforms all existing pure WSOD methods and transfer learning methods.



Paperid:1268
Authors:Chaoning Zhang; Kang Zhang; Chenshuang Zhang; Axi Niu; Jiu Feng; Chang D. Yoo; In So Kweon
Title: Decoupled Adversarial Contrastive Learning for Self-Supervised Adversarial Robustness
Abstract:
\textit{Adversarial training} (AT) for robust representation learning and \textit{self-supervised learning} (SSL) for unsupervised representation learning are two active research fields. Integrating AT into SSL, multiple prior works have accomplished a highly significant yet challenging task: learning robust representation without labels. A widely used framework is adversarial contrastive learning which couples AT and SSL, and thus constitutes a very complex optimization problem. Inspired by the divide-and-conquer philosophy, we conjecture that it might be simplified as well as improved by solving two sub-problems: non-robust SSL and pseudo-supervised AT. This motivation shifts the focus of the task from seeking an optimal integrating strategy for a coupled problem to finding sub-solutions for sub-problems. With this said, this work discards prior practices of directly introducing AT to SSL frameworks and proposed a two-stage framework termed \underline{De}coupled \underline{A}dversarial \underline{C}ontrastive \underline{L}earning (DeACL). Extensive experimental results demonstrate that our DeACL achieves SOTA self-supervised adversarial robustness while significantly reducing the training time, which validates its effectiveness and efficiency. Moreover, our DeACL constitutes a more explainable solution, and its success also bridges the gap with semi-supervised AT for exploiting unlabeled samples for robust representation learning.



Paperid:1269
Authors:Huseyin Coskun; Alireza Zareian; Joshua L. Moore; Federico Tombari; Chen Wang
Title: GOCA: Guided Online Cluster Assignment for Self-Supervised Video Representation Learning
Abstract:
Clustering is a ubiquitous tool in unsupervised learning. Most of the existing self-supervised representation learning methods typically cluster samples based on visually dominant features. While this works well for image-based selfsupervision, it often fails for videos, which require understanding motion rather than focusing on background. Using optical flow as complementary information to RGB can alleviate this problem. However, we observe that a na¨ıve combination of the two modalities does not provide meaningful gains. In this paper, we propose a principled way to combine two modalities. Specifically, we propose a novel clustering strategy where we use the initial cluster assignment of each modality as prior to guide the final cluster assignment of the other modality. This idea will enforce similar cluster structures for both modalities, and the formed clusters will be semantically abstract and robust to noisy inputs coming from each individual modality. Additionally, we propose a novel regularization strategy to address the feature collapse problem, which is common in cluster-based self-supervised learning methods. Our extensive evaluation shows the effectiveness of our learned representations on downstream tasks, e.g., video retrieval and action recognition. Specifically, we outperform the state of the art by 7% on UCF and 4% on HMDB for video retrieval as well as 5% on UCF and 6% on HMDB for linear video classification.



Paperid:1270
Authors:K L Navaneet; Soroush Abbasi Koohpayegani; Ajinkya Tejankar; Kossar Pourahmadi; Akshayvarun Subramanya; Hamed Pirsiavash
Title: Constrained Mean Shift Using Distant Yet Related Neighbors for Representation Learning
Abstract:
We are interested in representation learning in self-supervised, supervised, and semi-supervised settings. Some recent self-supervised learning methods like mean-shift (MSF) cluster images by pulling the embedding of a query image to be closer to its nearest neighbors (NNs). Since most NNs are close to the query by design, the averaging may not affect the embedding of the query much. On the other hand, far away NNs may not be semantically related to the query. We generalize the mean-shift idea by constraining the search space of NNs using another source of knowledge so that NNs are far from the query while still being semantically related. We show that our method (1) outperforms MSF in SSL setting when the constraint utilizes a different augmentation of an image from the previous epoch, and (2) outperforms PAWS in semi-supervised setting with less training resources when the constraint ensures that the NNs have the same pseudo-label as the query.



Paperid:1271
Authors:Junqiang Huang; Xiangwen Kong; Xiangyu Zhang
Title: Revisiting the Critical Factors of Augmentation-Invariant Representation Learning
Abstract:
We focus on better understanding the critical factors of augmentation-invariant representation learning. We revisit MoCo v2 and BYOL and try to prove the authenticity of the following assumption: different frameworks bring about representations of different characteristics even with the same pretext task. We establish the first benchmark for fair comparisons between MoCo v2 and BYOL, and observe: (i) sophisticated model configurations enable better adaptation to pre-training dataset; (ii) mismatched optimization strategies of pre-training and fine-tuning hinder model from achieving competitive transfer performances. Given the fair benchmark, we make further investigation and find asymmetry of network structure endows contrastive frameworks to work well under the linear evaluation protocol, while may hurt the transfer performances on long-tailed classification tasks. Moreover, negative samples do not make models more sensible to the choice of data augmentations, nor does the asymmetric network structure. We believe our findings provide useful information for future work.



Paperid:1272
Authors:Lu Qi; Jason Kuen; Zhe Lin; Jiuxiang Gu; Fengyun Rao; Dian Li; Weidong Guo; Zhen Wen; Ming-Hsuan Yang; Jiaya Jia
Title: CA-SSL: Class-Agnostic Semi-Supervised Learning for Detection and Segmentation
Abstract:
To improve instance-level detection/segmentation performance, existing self-supervised and semi-supervised methods extract either very task-unrelated or very task-specific training signals from unlabeled data. We argue that these two approaches, at the two extreme ends of the task-specificity spectrum, are suboptimal for the task performance. Utilizing too little task-specific training signals causes underfitting to the ground-truth labels of downstream tasks, while the opposite causes overfitting to the ground-truth labels. To this end, we propose a novel Class-Agnostic Semi-Supervised Learning (CA-SSL) framework to achieve a more favorable task-specificity balance in extracting training signals from unlabeled data. CA-SSL has three training stages that act on either ground-truth labels (labeled data) or pseudo labels (unlabeled data). This decoupling strategy avoids the complicated scheme in traditional SSL methods that balances the contributions from both data types. Especially, we introduce a warmup training stage to achieve a more optimal balance in task specificity by ignoring class information in the pseudo labels, while preserving localization training signals. As a result, our warmup model can better avoid underfitting/overfitting when fine-tuned on the ground-truth labels in detection and segmentation tasks. Using 3.6M unlabeled data, we achieve a remarkable performance gain of 4.7% over ImageNet-pretrained baseline on FCOS object detection. Also, our warmup model demonstrates excellent transferability to other detection and segmentation frameworks.



Paperid:1273
Authors:Zhonghua Wu; Yicheng Wu; Guosheng Lin; Jianfei Cai; Chen Qian
Title: Dual Adaptive Transformations for Weakly Supervised Point Cloud Segmentation
Abstract:
Weakly supervised point cloud segmentation, i.e. semantically segmenting a point cloud with only a few labeled points in the whole 3D scene, is highly desirable due to the heavy burden of collecting abundant dense annotations for the model training. However, existing methods remain challenging to accurately segment 3D point clouds since limited annotated data may lead to insufficient guidance for label propagation to unlabeled data. Considering the smoothness-based methods have achieved promising progress, in this paper, we advocate applying the consistency constraint under various perturbations to effectively regularize unlabeled 3D points. Specifically, we propose a novel DAT (Dual Adaptive Transformations) model for weakly supervised point cloud segmentation, where the dual adaptive transformations are performed via an adversarial strategy at both point-level and region-level, aiming at enforcing the local and structural smoothness constraints on 3D point clouds. We evaluate our proposed DAT model with two popular backbones on the large-scale S3DIS and ScanNet-V2 datasets. Extensive experiments demonstrate that our model\footnote{The codes will be released publicly once acceptance.} can effectively leverage the unlabeled 3D points and achieve significant performance gains on both datasets, setting new state-of-the-art performance for weakly supervised point cloud segmentation.



Paperid:1274
Authors:Yingdong Hu; Renhao Wang; Kaifeng Zhang; Yang Gao
Title: Semantic-Aware Fine-Grained Correspondence
Abstract:
Establishing visual correspondence across images is a challenging and essential task. Recently, an influx of self-supervised methods have been proposed to better learn representations for visual correspondence. However, we find that these methods often fail to leverage semantic information and over-rely on the matching of low-level features. In contrast, human vision is capable of distinguishing between distinct objects as a pretext to tracking. Inspired by this paradigm, we propose to learn semantic-aware fine-grained correspondence. Firstly, we demonstrate that semantic correspondence is implicitly available through a rich set of image-level self-supervised methods. We further design a pixel-level self-supervised learning objective which specifically targets fine-grained correspondence. For downstream tasks, we fuse these two kinds of complementary correspondence representations together, demonstrating that they boost performance synergistically. Our method surpasses previous state-of-the-art self-supervised methods using convolutional networks on a variety of visual correspondence tasks, including video object segmentation, human pose tracking, and human part tracking.



Paperid:1275
Authors:Elad Amrani; Leonid Karlinsky; Alex Bronstein
Title: Self-Supervised Classification Network
Abstract:
We present Self-Classifier -- a novel self-supervised end-to-end classification learning approach. Self-Classifier learns labels and representations simultaneously in a single-stage end-to-end manner by optimizing for same-class prediction of two augmented views of the same sample. To guarantee non-degenerate solutions (i.e., solutions where all labels are assigned to the same class) we propose a mathematically motivated variant of the cross-entropy loss that has a uniform prior asserted on the predicted labels. In our theoretical analysis, we prove that degenerate solutions are not in the set of optimal solutions of our approach. Self-Classifier is simple to implement and scalable. Unlike other popular unsupervised classification and contrastive representation learning approaches, it does not require any form of pre-training, expectation-maximization, pseudo-labeling, external clustering, a second network, stop-gradient operation, or negative pairs. Despite its simplicity, our approach sets a new state of the art for unsupervised classification of ImageNet; and even achieves comparable to state-of-the-art results for unsupervised representation learning. Code is available at https://github.com/elad-amrani/self-classifier.



Paperid:1276
Authors:Lars Doorenbos; Raphael Sznitman; Pablo Márquez-Neila
Title: Data Invariants to Understand Unsupervised Out-of-Distribution Detection
Abstract:
Unsupervised out-of-distribution (U-OOD) detection has recently attracted much attention due to its importance in mission-critical systems and broader applicability over its supervised counterpart. Despite this increased attention, U-OOD methods suffer from important shortcomings. By performing a large-scale evaluation on different benchmarks and image modalities, we show in this work that most popular state-of-the-art methods are unable to consistently outperform a simple anomaly detector based on pre-trained features and the Mahalanobis distance (MahaAD). A key reason for the inconsistencies of these methods is the lack of a formal description of U-OOD. Motivated by a simple thought experiment, we propose a characterization of U-OOD based on the invariants of the training dataset. We show how this characterization is unknowingly embodied in the top-scoring MahaAD method, thereby explaining its quality. Furthermore, our approach can be used to interpret predictions of U-OOD detectors and provides insights into good practices for evaluating future U-OOD methods.



Paperid:1277
Authors:Haiyang Yang; Shixiang Tang; Meilin Chen; Yizhou Wang; Feng Zhu; Lei Bai; Rui Zhao; Wanli Ouyang
Title: Domain Invariant Masked Autoencoders for Self-Supervised Learning from Multi-Domains
Abstract:
Generalizing learned representations across significantly different visual domains is a fundamental yet crucial ability of the human visual system. While recent self-supervised learning methods have achieved good performances with evaluation set on the same domain as the training set, they will have an undesirable performance decrease when tested on a different domain. Therefore, the self-supervised learning from multiple domains task is proposed to learn domain-invariant features that are not only suitable for evaluation on the same domain as the training set, but also can be generalized to unseen domains. In this paper, we propose a Domain-invariant Masked AutoEncoder (DiMAE) for self-supervised learning from multi-domains, which designs a new pretext task, i.e., the cross-domain reconstruction task, to learn domain-invariant features. The core idea is to augment the input image with style noise from different domains and then reconstruct the image from the embedding of the augmented image, regularizing the encoder to learn domain-invariant features. To accomplish the idea, DiMAE contains two critical designs, 1) content-preserved style mix, which adds style information from other domains to input while persevering the content in a parameter-free manner, and 2) multiple domain-specific decoders, which recovers the corresponding domain style of input to the encoded domain-invariant features for reconstruction. Experiments on PACS and DomainNet illustrate that DiMAE achieves considerable gains compared with recent state-of-the-art methods.



Paperid:1278
Authors:Changrui Chen; Kurt Debattista; Jungong Han
Title: Semi-Supervised Object Detection via Virtual Category Learning
Abstract:
Due to the costliness of labelled data in real-world applications, semi-supervised object detectors, underpinned by pseudo labelling, are appealing. However, handling confusing samples is nontrivial: discarding valuable confusing samples would compromise the model generalisation while using them for training would exacerbate the confirmation bias issue caused by inevitable mislabelling. To solve this problem, this paper proposes to use confusing samples proactively without label correction. Specifically, a virtual category (VC) is assigned to each confusing sample such that they can safely contribute to the model optimisation even without a concrete label. It is attributed to specifying the embedding distance between the training sample and the virtual category as the lower bound of the inter-class distance. Moreover, we also modify the localisation loss to allow high-quality boundaries for location regression. Extensive experiments demonstrate that the proposed VC learning significantly surpasses the state-of-the-art, especially with small amounts of available labels.



Paperid:1279
Authors:Deepak Babu Sam; Abhinav Agarwalla; Jimmy Joseph; Vishwanath A. Sindagi; R. Venkatesh Babu; Vishal M. Patel
Title: Completely Self-Supervised Crowd Counting via Distribution Matching
Abstract:
Dense crowd counting is a challenging task that demands millions of head annotations for training models. Though existing self-supervised approaches could learn good representations, they require some labeled data to map these features to the end task of density estimation. We mitigate this issue with the proposed paradigm of complete self-supervision, which does not need even a single labeled image. The only input required to train, apart from a large set of unlabeled crowd images, is the approximate upper limit of the crowd count for the given dataset. Our method dwells on the idea that natural crowds follow a power law distribution, which could be leveraged to yield error signals for backpropagation. A density regressor is first pretrained with self-supervision and then the distribution of predictions is matched to the prior. Experiments show that this results in effective learning of crowd features and delivers significant counting performance.



Paperid:1280
Authors:Xiang Xiang; Yuwen Tan; Qian Wan; Jing Ma; Alan Yuille; Gregory D. Hager
Title: Coarse-to-Fine Incremental Few-Shot Learning
Abstract:
Different from fine-tuning models pre-trained on a large-scale dataset of preset classes, class-incremental learning (CIL) aims to recognize novel classes over time without forgetting pre-trained classes. However, a given model will be challenged by test images with finer-grained classes, e.g., a basenji is at most recognized as a dog. Such images form a new training set (i.e., support set) so that the incremental model is hoped to recognize a basenji (i.e., query) as a basenji next time. This paper formulates such a hybrid natural problem of coarse-to-fine few-shot (C2FS) recognition as a CIL problem named C2FSCIL, and proposes a simple, effective, and theoretically-sound strategy Knowe: to learn, normalize, and freeze a classifier’s weights from fine labels, once learning an embedding space contrastively from coarse labels. Besides, as CIL aims at a stability-plasticity balance, new overall performance metrics are proposed. In that sense, on CIFAR-100, BREEDS, and tieredImageNet, Knowe outperforms all recent relevant CIL or FSCIL methods.



Paperid:1281
Authors:Jian Hu; Haowen Zhong; Fei Yang; Shaogang Gong; Guile Wu; Junchi Yan
Title: Learning Unbiased Transferability for Domain Adaptation by Uncertainty Modeling
Abstract:
Domain adaptation (DA) aims to transfer knowledge learned from a labeled source domain to an unlabeled or a less labeled but related target domain. Ideally, the source and target distributions should be aligned to each other equally to achieve unbiased knowledge transfer. However, due to the significant imbalance between the amount of annotated data in the source and target domains, usually only the target distribution is aligned to the source domain, leading to adapting unnecessary source specific knowledge to the target domain, i.e., biased domain adaptation. To resolve this problem, in this work, we delve into the transferability estimation problem in domain adaptation, proposing a non-intrusive Unbiased Transferability Estimation Plug-in (UTEP) by modeling the uncertainty of a discriminator in adversarial-based DA methods to optimize unbiased transfer. We theoretically analyze the effectiveness of the proposed approach for unbiased transferability learning in DA. Furthermore, to alleviate the impact of imbalanced annotated data, we utilize the estimated uncertainty for pseudo label selection of unlabeled samples in the target domain, which helps achieve better marginal and conditional distribution alignments between domains. Extensive experimental results on a high variety of DA benchmark datasets show that the proposed approach can be readily incorporated into various adversarial-based DA methods, achieving compelling performance against the state-of-the-art methods.



Paperid:1282
Authors:Shreyank N Gowda; Marcus Rohrbach; Frank Keller; Laura Sevilla-Lara
Title: Learn2Augment: Learning to Composite Videos for Data Augmentation in Action Recognition
Abstract:
We address the problem of data augmentation for video action recognition. Standard augmentation strategies in video are hand designed and sample the space of possible augmented data points either at random, without knowing which augmented points will be better, or through heuristics. We propose to learn what makes a “good” video for action recognition and select only high-quality samples for augmentation. In particular, we choose video compositing of a foreground and a background video as the data augmentation process, which results in diverse and realistic new samples. We learn which pairs of videos to augment without having to actually composite them. This reduces the space of possible augmentations, which has two advantages: it saves computational cost and increases the accuracy of the final trained classifier, as the augmented pairs are of higher quality than average. We present experimental results on the entire spectrum of training settings: few-shot, semi-supervised and fully supervised. We observe consistent improvements across all of them over prior work and baselines on Kinetics, UCF101, HMDB51, and achieve a new state-of-the-art on settings with limited data. We see improvements of up to 8.6% in the semi-supervised setting.



Paperid:1283
Authors:Renhao Wang; Hang Zhao; Yang Gao
Title: CYBORGS: Contrastively Bootstrapping Object Representations by Grounding in Segmentation
Abstract:
Many recent approaches in contrastive learning have worked to close the gap between pretraining on iconic images like ImageNet and pretraining on complex scenes like COCO. This gap exists largely because commonly used random crop augmentations obtain semantically inconsistent content in crowded scene images of diverse objects. Previous works use preprocessing pipelines to localize salient objects for improved cropping, but an end-to-end solution is still elusive. In this work, we propose a framework which accomplishes this goal via joint learning of representations and segmentation. We leverage segmentation masks to train a model with a mask-dependent contrastive loss, and use the partially trained model to bootstrap better masks. By iterating between these two components, we ground the contrastive updates in segmentation information, and simultaneously improve segmentation throughout pretraining. Experiments show our representations transfer robustly to downstream tasks in classification, detection and segmentation.



Paperid:1284
Authors:Tianyue Cao; Yongxin Wang; Yifan Xing; Tianjun Xiao; Tong He; Zheng Zhang; Hao Zhou; Joseph Tighe
Title: PSS: Progressive Sample Selection for Open-World Visual Representation Learning
Abstract:
We propose a practical open-world representation learning setting where the objective is to learn the representations for unseen categories without prior knowledge or access to images associated with these novel categories during training. Existing open-world representation learning methods, however, assume many constraints, which are often violated in practice and thus fail to generalize to the proposed setting. We propose a novel progressive approach which, at each iteration, selects unlabeled samples that attain a high homogeneity while belonging to classes that are distant to the current set of known classes in the feature space. Then we use the high-quality pseudo-labels generated via clustering over these selected samples to improve the feature generalization iteratively. Experiments demonstrate that the proposed method consistently outperforms state-of-the-art open-world semi-supervised learning methods and novel class discovery methods over nature species image retrieval and face verification benchmarks.



Paperid:1285
Authors:Hao Liu; Mang Ye
Title: Improving Self-Supervised Lightweight Model Learning via Hard-Aware Metric Distillation
Abstract:
The performance of self-supervised learning (SSL) models is hindered by the scale of the network. Existing SSL methods suffer a precipitous drop in lightweight models, which is important for many mobile devices. To address this problem, we propose a method to improve the lightweight network (as student) via distilling the metric knowledge in a larger SSL model (as teacher). We exploit the relation between teacher and student to mine the positive and negative supervision from the unlabeled data, which captures more accurate supervision signals. To adaptively handle the uncertainty in positive and negative sample pairs, we incorporate a dynamic weighting strategy to the metric relation between embeddings. Different from previous self-supervised distillers, our solution directly optimizes the network from a metric transfer perspective by utilizing the relationships between samples and networks, without additional SSL constraints. Our method significantly boosts the performance of lightweight networks and outperforms existing distillers with fewer training epochs on the large-scale ImageNet. Interestingly, the SSL performance even beats the teacher network in several settings.



Paperid:1286
Authors:Jinhwan Seo; Wonho Bae; Danica J. Sutherland; Junhyug Noh; Daijin Kim
Title: Object Discovery via Contrastive Learning for Weakly Supervised Object Detection
Abstract:
Weakly Supervised Object Detection (WSOD) is a task that detects objects in an image using a model trained only on image-level annotations. Current state-of-the-art models benefit from self-supervised instance-level supervision, but since weak supervision does not include count or location information, the most common “argmax” labeling method often ignores many instances of objects. To alleviate this issue, we propose a novel multiple instance labeling method called object discovery. We further introduce a new contrastive loss under weak supervision where no instance-level information is available for sampling, called weakly supervised contrastive loss (WSCL). WSCL aims to construct a credible similarity threshold for object discovery by leveraging consistent features for embedding vectors in the same class. As a result, we achieve new state-of-the-art results on MS-COCO 2014 and 2017 as well as PASCAL VOC 2012, and competitive results on PASCAL VOC 2007. The code is available at https://github.com/jinhseo/OD-WSCL.



Paperid:1287
Authors:Hui Tang; Lin Sun; Kui Jia
Title: Stochastic Consensus: Enhancing Semi-Supervised Learning with Consistency of Stochastic Classifiers
Abstract:
Semi-supervised learning (SSL) has achieved new progress recently with the emerging framework of self-training deep networks, where the criteria for selection of unlabeled samples with pseudo labels play a key role in the empirical success. In this work, we propose such a new criterion based on consistency among multiple, stochastic classifiers, termed Stochastic Consensus (STOCO). Specifically, we model parameters of the classifiers as a Gaussian distribution whose mean and standard deviation are jointly optimized during training. Due to the scarcity of labels in SSL, modeling classifiers as a distribution itself provides additional regularization that mitigates overfitting to the labeled samples. We technically generate pseudo labels using a simple but flexible framework of deep discriminative clustering, which benefits from the overall structure of data distribution. We also provide theoretical analysis of our criterion by connecting with the theory of learning from noisy data. Our proposed criterion can be readily applied to self-training based SSL frameworks. By choosing the representative FixMatch as the baseline, our method with multiple stochastic classifiers achieves the state of the art on popular SSL benchmarks, especially in label-scarce cases.



Paperid:1288
Authors:Boah Kim; Inhwa Han; Jong Chul Ye
Title: DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model
Abstract:
Deformable image registration is one of the fundamental tasks in medical imaging. Classical registration algorithms usually require a high computational cost for iterative optimizations. Although deep-learning-based methods have been developed for fast image registration, it is still challenging to obtain realistic continuous deformations from a moving image to a fixed image with less topological folding problem. To address this, here we present a novel diffusion-model-based image registration method, called DiffuseMorph. DiffuseMorph not only generates synthetic deformed images through reverse diffusion but also allows image registration by deformation fields. Specifically, the deformation fields are generated by the conditional score function of the deformation between the moving and fixed images, so that the registration can be performed from continuous deformation by simply scaling the latent feature of the score. Experimental results on 2D facial and 3D medical image registration tasks demonstrate that our method provides flexible deformations with topology preservation capability.



Paperid:1289
Authors:Xinlei He; Hongbin Liu; Neil Zhenqiang Gong; Yang Zhang
Title: Semi-Leak: Membership Inference Attacks against Semi-Supervised Learning
Abstract:
Semi-supervised learning (SSL) leverages both labeled and unlabeled data to train machine learning (ML) models. State-of-the-art SSL methods can achieve comparable performance to supervised learning by leveraging much fewer labeled data. However, most existing works focus on improving the performance of SSL. In this work, we take a different angle by studying the training data privacy of SSL. Specifically, we propose the first data augmentation-based membership inference attacks against ML models trained by SSL. Given a data sample and the black-box access to a model, the goal of membership inference attack is to determine whether the data sample belongs to the training dataset of the model. Our evaluation shows that the proposed attack can consistently outperform existing membership inference attacks and achieves the best performance against the model trained by SSL. Moreover, we uncover that the reason for membership leakage in SSL is different from the commonly believed one in supervised learning, i.e., overfitting (the gap between training and testing accuracy). We observe that the SSL model is well generalized to the testing data (with almost 0 overfitting) but memorizes the training data by giving a more confident prediction regardless of its correctness. We also explore early stopping as a countermeasure to prevent membership inference attacks against SSL. The results show that early stopping can mitigate the membership inference attack, but with the cost of model’s utility degradation.



Paperid:1290
Authors:Mamshad Nayeem Rizve; Navid Kardan; Salman Khan; Fahad Shahbaz Khan; Mubarak Shah
Title: OpenLDN: Learning to Discover Novel Classes for Open-World Semi-Supervised Learning
Abstract:
Semi-supervised learning (SSL) is one of the dominant approaches to address the annotation bottleneck of supervised learning. Recent SSL methods can effectively leverage a large repository of unlabeled data to improve performance while relying on a small set of labeled data. One common assumption in most SSL methods is that the labeled and unlabeled data are from the same data distribution. However, this is hardly the case in many real-world scenarios, which limits their applicability. In this work, instead, we attempt to solve the challenging open-world SSL problem that does not make such an assumption. In the open-world SSL problem, the objective is to recognize samples of known classes, and simultaneously detect and cluster samples belonging to novel classes present in unlabeled data. This work introduces OpenLDN that utilizes a pairwise similarity loss to discover novel classes. Using a bi-level optimization rule this pairwise similarity loss exploits the information available in the labeled set to implicitly cluster novel class samples, while simultaneously recognizing samples from known classes. After discovering novel classes, OpenLDN transforms the open-world SSL problem into a standard SSL problem to achieve additional performance gains using existing SSL methods. Our extensive experiments demonstrate that OpenLDN outperforms the current state-of-the-art methods on multiple popular classification benchmarks while providing a better accuracy/training time trade-off.



Paperid:1291
Authors:Paul Albert; Eric Arazo; Noel E. O’Connor; Kevin McGuinness
Title: Embedding Contrastive Unsupervised Features to Cluster in- and Out-of-Distribution Noise in Corrupted Image Datasets
Abstract:
Using search engines for web image retrieval is a tempting alternative to manual curation when creating an image dataset, but their main drawback remains the proportion of incorrect (noisy) samples retrieved. These noisy samples have been evidenced by previous works to be a mixture of in-distribution (ID) samples, assigned to the incorrect category but presenting similar visual semantics to other classes in the dataset, and out-of-distribution (OOD) images, which share no semantic correlation with any category from the dataset. The latter are, in practice, the dominant type of noisy images retrieved. To tackle this noise duality, we propose a two stage algorithm starting with a detection step where we use unsupervised contrastive feature learning to represent images in a feature space. We find that the alignment and uniformity principles of contrastive learning allow OOD samples to be linearly separated from ID samples on the unit hypersphere. We then spectrally embed the unsupervised representations using a fixed neighborhood size and apply an outlier sensitive clustering at the class level to detect the clean and OOD clusters as well as ID noisy outliers. We finally train a noise robust neural network that corrects ID noise to the correct category and utilizes OOD samples in a guided contrastive objective, clustering them to improve low-level features. Our algorithm improves the state-of-the-art results on synthetic noise image datasets as well as real-world web-crawled data. Our work is fully reproducible github.com/PaulAlbert31/SNCF.



Paperid:1292
Authors:Shuo Li; Fang Liu; Zehua Hao; Kaibo Zhao; Licheng Jiao
Title: Unsupervised Few-Shot Image Classification by Learning Features into Clustering Space
Abstract:
Most few-shot image classification methods are trained based on tasks. Usually, tasks are built on base classes with a large number of labeled images, which consumes large effort. Unsupervised few-shot image classification methods do not need labeled images, because they require tasks to be built on unlabeled images. In order to efficiently build tasks with unlabeled images, we propose a novel single-stage clustering method: Learning Features into Clustering Space (LF2CS), which first set a separable clustering space by fixing the clustering centers and then use a learnable model to learn features into the clustering space. Based on our LF2CS, we put forward an image sampling and c-way k-shot task building method. With this, we propose a novel unsupervised few-shot image classification method, which jointly learns the learnable model, clustering and few-shot image classification. Experiments and visualization show that our LF2CS has a strong ability to generalize to the novel categories. From the perspective of image sampling, we implement four baselines according to how to build tasks. We conduct experiments on the Omniglot, miniImageNet, tieredImageNet and CIFARFS datasets based on the Conv-4 and ResNet-12 backbones. Experimental results show that ours outperform the state-of-the-art methods.



Paperid:1293
Authors:Mamshad Nayeem Rizve; Navid Kardan; Mubarak Shah
Title: Towards Realistic Semi-Supervised Learning
Abstract:
Deep learning is pushing the state-of-the-art in many computer vision applications. However, it relies on large annotated data repositories, and capturing the unconstrained nature of the real-world data is yet to be solved. Semi-supervised learning (SSL) complements the annotated training data with a large corpus of unlabeled data to reduce annotation cost. The standard SSL approach assumes unlabeled data are from the same distribution as annotated data. Recently, a more realistic SSL problem, called open-world SSL, is introduced, where the unannotated data might contain samples from unknown classes. In this paper, we propose a novel pseudo-label based approach to tackle SSL in open-world setting. At the core of our method, we utilize sample uncertainty and incorporate prior knowledge about class distribution to generate reliable class-distribution-aware pseudo-labels for unlabeled data belonging to both known and unknown classes. Our extensive experimentation showcases the effectiveness of our approach on several benchmark datasets, where it substantially outperforms the existing state-of-the-art on seven diverse datasets including CIFAR-100 ( 17%), ImageNet-100 ( 5%), and Tiny ImageNet ( 9%). We also highlight the flexibility of our approach in solving novel class discovery task, demonstrate its stability in dealing with imbalanced data, and complement our approach with a technique to estimate the number of novel classes.



Paperid:1294
Authors:Mahmoud Assran; Mathilde Caron; Ishan Misra; Piotr Bojanowski; Florian Bordes; Pascal Vincent; Armand Joulin; Michael Rabbat; Nicolas Ballas
Title: Masked Siamese Networks for Label-Efficient Learning
Abstract:
We propose Masked Siamese Networks (MSN), a self-supervised learning framework for learning image representations. Our approach matches the representation of an image view containing randomly masked patches to the representation of the original unmasked image. This self-supervised pre-training strategy is particularly scalable when applied to Vision Transformers since only the unmasked patches are processed by the network. As a result, MSNs improve the scalability of joint-embedding architectures, while producing representations of a high semantic level that perform competitively on low-shot image classification. For instance, on ImageNet-1K, with only 5,000 annotated images, our large MSN models achieves 72.1% top-1 accuracy, and with 1% of ImageNet-1K labels, we achieve 75.1% top-1 accuracy, setting a new state-of-the-art for self-supervised learning on this benchmark. Our code is publicly available at https://github.com/facebookresearch/msn.



Paperid:1295
Authors:Hannah M. Schlüter; Jeremy Tan; Benjamin Hou; Bernhard Kainz
Title: Natural Synthetic Anomalies for Self-Supervised Anomaly Detection and Localization
Abstract:
We introduce a simple and intuitive self-supervision task, Natural Synthetic Anomalies (NSA), for training an end-to-end model for anomaly detection and localization using only normal training data. NSA integrates Poisson image editing to seamlessly blend scaled patches of various sizes from separate images. This creates a wide range of synthetic anomalies which are more similar to natural sub-image irregularities than previous data-augmentation strategies for self-supervised anomaly detection. We evaluate the proposed method using natural and medical images. Our experiments with the MVTec AD dataset show that a model trained to localize NSA anomalies generalizes well to detecting real-world a priori unknown types of manufacturing defects. Our method achieves an overall detection AUROC of 97.2 outperforming all previous methods that learn without the use of additional datasets. Code available at https://github.com/hmsch/natural-synthetic-anomalies.



Paperid:1296
Authors:Alexander C. Li; Alexei A. Efros; Deepak Pathak
Title: Understanding Collapse in Non-Contrastive Siamese Representation Learning
Abstract:
Contrastive methods have led a recent surge in the performance of self-supervised representation learning (SSL). Recent methods like BYOL or SimSiam purportedly distill these contrastive methods down to their essence, removing bells and whistles, including the negative examples, that do not contribute to downstream performance. These “non-contrastive” methods surprisingly work well without using negatives even though the global minimum lies at trivial collapse. We empirically analyze these non-contrastive methods and find that SimSiam is extraordinarily sensitive to model size. In particular, SimSiam representations undergo partial dimensional collapse if the model is too small relative to the dataset size. We propose a metric to measure the degree of this collapse and show that it can be used to forecast the downstream task performance without any fine-tuning or labels. We further analyze architectural design choices and their effect on the downstream performance. Finally, we demonstrate that shifting to a continual learning setting acts as a regularizer and prevents collapse, and a hybrid between continual and multi-epoch training can improve linear probe accuracy by as many as 18 percentage points using ResNet-18 on ImageNet.



Paperid:1297
Authors:Yasar Abbas Ur Rehman; Yan Gao; Jiajun Shen; Pedro Porto Buarque de Gusmão; Nicholas Lane
Title: Federated Self-Supervised Learning for Video Understanding
Abstract:
The ubiquity of camera-enabled mobile devices has lead to large amounts of unlabelled video data being produced at the edge. Although various self-supervised learning (SSL) methods have been proposed to harvest their latent spatio-temporal representations for task-specific training, practical challenges including privacy concerns and communication costs prevent SSL from being deployed at large scales. To mitigate these issues, we propose the use of Federated Learning (FL) to the task of video SSL. In this work, we evaluate the performance of current state-of-the-art (SOTA) video-SSL techniques and identify their shortcomings when integrated into the large-scale FL setting simulated with kinetics-400 dataset. We follow by proposing a novel federated SSL framework for video, dubbed FedVSSL, that integrates different aggregation strategies and partial weight updating. Extensive experiments demonstrate the effectiveness and significance of FedVSSL as it outperforms the centralized SOTA for the downstream retrieval task by 6.66% on UCF-101 and 5.13% on HMDB-51.



Paperid:1298
Authors:Sravanti Addepalli; Kaushal Bhogale; Priyam Dey; R. Venkatesh Babu
Title: Towards Efficient and Effective Self-Supervised Learning of Visual Representations
Abstract:
Self-supervision has emerged as a propitious method for visual representation learning after the recent paradigm shift from handcrafted pretext tasks to instance-similarity based approaches. Most state-of-the-art methods enforce similarity between various augmentations of a given image, while some methods additionally use contrastive approaches to explicitly ensure diverse representations. While these approaches have indeed shown promising direction, they require a significantly larger number of training iterations when compared to the supervised counterparts. In this work, we explore reasons for the slow convergence of these methods, and further propose to strengthen them using well-posed auxiliary tasks that converge significantly faster, and are also useful for representation learning. The proposed method utilizes the task of rotation prediction to improve the efficiency of existing state-of-the-art methods. We demonstrate significant gains in performance using the proposed method on multiple datasets, specifically for lower training epochs.



Paperid:1299
Authors:Vitjan Zavrtanik; Matej Kristan; Danijel Skočaj
Title: DSR – A Dual Subspace Re-Projection Network for Surface Anomaly Detection
Abstract:
The state-of-the-art in discriminative unsupervised surface anomaly detection relies on external datasets for synthesizing anomaly-augmented training images. Such approaches are prone to failure on near-in-distribution anomalies since these are difficult to be synthesized realistically due to their similarity to anomaly-free regions. We propose an architecture based on quantized feature space representation with dual decoders, DSR, that avoids the image-level anomaly synthesis requirement. Without making any assumptions about the visual properties of anomalies, DSR generates the anomalies at the feature level by sampling the learned quantized feature space, which allows a controlled generation of near-in-distribution anomalies. DSR achieves state-of-the-art results on the KSDD2 and MVTec anomaly detection datasets. The experiments on the challenging real-world KSDD2 dataset show that DSR significantly outperforms other unsupervised surface anomaly detection methods, improving the previous top-performing methods by 10% AP in anomaly detection and 35% AP in anomaly localization.



Paperid:1300
Authors:Zhaoqi Leng; Shuyang Cheng; Benjamin Caine; Weiyue Wang; Xiao Zhang; Jonathon Shlens; Mingxing Tan; Dragomir Anguelov
Title: PseudoAugment: Learning to Use Unlabeled Data for Data Augmentation in Point Clouds
Abstract:
Data augmentation is an important technique to improve data efficiency and to save labeling cost for 3D detection in point clouds. Yet, existing augmentation policies have so far been designed to only utilize labeled data, which limits the data diversity. In this paper, we recognize that pseudo labeling and data augmentation are complementary, thus propose to leverage unlabeled data for data augmentation to enrich the training data. In particular, we design three novel pseudo-label based data augmentation policies (PseudoAugments) to fuse both labeled and pseudo-labeled scenes, including frames (PseudoFrame), objects (PseudoBBox), and background (PseudoBackground). PseudoAugments outperforms pseudo labeling by mitigating pseudo labeling errors and generating diverse fused training scenes. We demonstrate PseudoAugments generalize across point-based and voxel-based architectures, different model capacity and both KITTI and Waymo Open Dataset. To alleviate the cost of hyperparameter tuning and iterative pseudo labeling, we develop a population-based data augmentation framework for 3D detection, named AutoPseudoAugment. Unlike previous works that perform pseudo-labeling offline, our framework performs PseudoAugments and hyperparameter tuning in one shot to reduce computational cost. Experimental results on the large-scale Waymo Open Dataset show our method outperforms state-of-the-art auto data augmentation method (PPBA) and self-training method (pseudo labeling). In particular, AutoPseudoAugment is about 3X and 2X data efficient on vehicle and pedestrian tasks compared to prior arts. Notably, AutoPseudoAugment nearly matches the full dataset training results, with just 10% of the labeled run segments on the vehicle detection task.



Paperid:1301
Authors:Xiaofeng Wang; Zheng Zhu; Guan Huang; Fangbo Qin; Yun Ye; Yijia He; Xu Chi; Xingang Wang
Title: MVSTER: Epipolar Transformer for Efficient Multi-View Stereo
Abstract:
Learning-based Multi-View Stereo (MVS) methods warp source images into the reference camera frustum to form 3D volumes, which are fused as a cost volume to be regularized by subsequent networks. The fusing step plays a vital role in bridging 2D semantics and 3D spatial associations. However, previous methods utilize extra networks to learn 2D information as fusing cues, underusing 3D spatial correlations and bringing additional computation costs. Therefore, we present MVSTER, which leverages the proposed epipolar Transformer to learn both 2D semantics and 3D spatial associations efficiently. Specifically, the epipolar Transformer utilizes a detachable monocular depth estimator to enhance 2D semantics and uses cross-attention to construct data-dependent 3D associations along epipolar line. Additionally, MVSTER is built in a cascade structure, where entropy-regularized optimal transport is leveraged to propagate finer depth estimations in each stage. Extensive experiments show MVSTER achieves state-of-the-art reconstruction performance with significantly higher efficiency: Compared with MVSNet and CasMVSNet, our MVSTER achieves 34% and 14% relative improvements on the DTU benchmark, with 80% and 51% relative reductions in running time. MVSTER also ranks first on Tanks&Temples-Advanced among all published works. Code is available at https://github.com/JeffWang987/MVSTER.



Paperid:1302
Authors:Jason Y. Zhang; Deva Ramanan; Shubham Tulsiani
Title: RelPose: Predicting Probabilistic Relative Rotation for Single Objects in the Wild
Abstract:
We describe a data-driven method for inferring the camera viewpoints given multiple images of an arbitrary object. This task is a core component of classic geometric pipelines such as SfM and SLAM, and also serves as a vital pre-processing requirement for contemporary neural approaches (e.g. NeRF) to object reconstruction and view synthesis. In contrast to existing correspondence-driven methods that do not perform well given sparse views, we propose a top-down prediction based approach for estimating camera viewpoints. Our key technical insight is the use of an energy-based formulation for representing distributions over relative camera rotations, thus allowing us to explicitly represent multiple camera modes arising from object symmetries or views. Leveraging these relative predictions, we jointly estimate a consistent set of camera rotations from multiple images. We show that our approach outperforms state-of-the-art SfM and SLAM methods given sparse images on both seen and unseen categories. Further, our probabilistic approach significantly outperforms directly regressing relative poses, suggesting that modeling multimodality is important for coherent joint reconstruction. We demonstrate that our system can be a stepping stone toward in-the-wild reconstruction from multi-view datasets. The project page with code and videos can be found at https://jasonyzhang.com/relpose.



Paperid:1303
Authors:Huan Wang; Jian Ren; Zeng Huang; Kyle Olszewski; Menglei Chai; Yun Fu; Sergey Tulyakov
Title: R2L: Distilling Neural \textit{Radiance} Field to Neural \textit{Light} Field for Efficient Novel View Synthesis
Abstract:
Recent research explosion on Neural Radiance Field (NeRF) shows the encouraging potential to represent complex scenes with neural networks. One major drawback of NeRF is its prohibitive inference time: Rendering a single pixel requires querying the NeRF network hundreds of times. To resolve it, existing efforts mainly attempt to reduce the number of required sampled points. However, the problem of iterative sampling still exists. On the other hand, Neural \textit{Light} Field (NeLF) presents a more straightforward representation over NeRF in novel view synthesis -- the rendering of a pixel amounts to \textit{one single forward pass} without ray-marching. In this work, we present a \textit{deep residual MLP} network (88 layers) to effectively learn the light field. We show the key to successfully learning such a deep NeLF network is to have sufficient data, for which we transfer the knowledge from a pre-trained NeRF model via data distillation. Extensive experiments on both synthetic and real-world scenes show the merits of our method over other counterpart algorithms. On the synthetic scenes, we achieve $26\sim35\times$ FLOPs reduction (per camera ray) and $28\sim31\times$ runtime speedup, meanwhile delivering \textit{significantly better} ($1.4\sim2.8$ dB average PSNR improvement) rendering quality than NeRF without any customized parallelism requirement.



Paperid:1304
Authors:Yikang Ding; Qingtian Zhu; Xiangyue Liu; Wentao Yuan; Haotian Zhang; Chi Zhang
Title: KD-MVS: Knowledge Distillation Based Self-Supervised Learning for Multi-View Stereo
Abstract:
Supervised multi-view stereo (MVS) methods have achieved remarkable progress in terms of reconstruction quality, but suffer from the challenge of collecting large-scale ground-truth depth. In this paper, we propose a novel self-supervised training pipeline for MVS based on knowledge distillation, termed KD-MVS, which mainly consists of self-supervised teacher training and distillation-based student training. Specifically, the teacher model is trained in a self-supervised fashion using both photometric and featuremetric consistency. Then we distill the knowledge of the teacher model to the student model through probabilistic knowledge transferring. With the supervision of validated knowledge, the student model is able to outperform its teacher by a large margin. Extensive experiments performed on multiple datasets show our method can even outperform supervised methods. It ranks 1st among all submitted methods on Tanks and Temples benchmark.



Paperid:1305
Authors:John Lambert; Yuguang Li; Ivaylo Boyadzhiev; Lambert Wixson; Manjunath Narayana; Will Hutchcroft; James Hays; Frank Dellaert; Sing Bing Kang
Title: SALVe: Semantic Alignment Verification for Floorplan Reconstruction from Sparse Panoramas
Abstract:
We propose a new system for automatic 2D floorplan reconstruction that is enabled by SALVe, our novel pairwise learned alignment verifier. The inputs to our system are sparsely located 360 deg. panoramas, whose semantic features (windows, doors, and openings) are inferred and used to hypothesize pairwise room adjacency or overlap. SALVe initializes a pose graph, which is subsequently optimized using GTSAM. Once the room poses are computed, room layouts are inferred using HorizonNet, and the floorplan is constructed by stitching the most confident layout boundaries. We validate our system qualitatively and quantitatively as well as through ablation studies, showing that it outperforms state-of-the-art SfM systems in completeness by over 200%, without sacrificing accuracy. Our results point to the significance of our work: poses of 81% of panoramas are localized in the first 2 CCs connected components (CCs), and 89% in the first 3 CCs.



Paperid:1306
Authors:Di Chang; Aljaž Božič; Tong Zhang; Qingsong Yan; Yingcong Chen; Sabine Süsstrunk; Matthias Nießner
Title: RC-MVSNet: Unsupervised Multi-View Stereo with Neural Rendering
Abstract:
Finding accurate correspondences among different views is the Achilles’ heel of unsupervised Multi-View Stereo (MVS). Existing methods are built upon the assumption that corresponding pixels share similar photometric features. However, multi-view images in real scenarios observe non-Lambertian surfaces and experience occlusions. In this work, we propose a novel approach with neural rendering (RC-MVSNet) to solve such ambiguity issues of correspondences among views. Specifically, we impose a depth rendering consistency loss to constrain the geometry features close to the object surface to alleviate occlusions. Concurrently, we introduce a reference view synthesis loss to generate consistent supervision, even for non-Lambertian surfaces. Extensive experiments on DTU and Tanks&Temples benchmarks demonstrate that our RC-MVSNet approach achieves state-of-the-art performance over unsupervised MVS frameworks and competitive performance to many supervised methods. The code is released at https://github.com/Boese0601/RC-MVSNet.



Paperid:1307
Authors:Julian Chibane; Francis Engelmann; Tuan Anh Tran; Gerard Pons-Moll
Title: Box2Mask: Weakly Supervised 3D Semantic Instance Segmentation Using Bounding Boxes
Abstract:
Current 3D segmentation methods heavily rely on large-scale point-cloud datasets, which are notoriously laborious to annotate. Few attempts have been made to circumvent the need for dense per-point annotations. In this work, we look at weakly-supervised 3D semantic instance segmentation. The key idea is to leverage 3D bounding box labels which are easier and faster to annotate. Indeed, we show that it is possible to train dense segmentation models using only bounding box labels. At the core of our method, \name{}, lies a deep model, inspired by classical Hough voting, that directly votes for bounding box parameters, and a clustering method specifically tailored to bounding box votes. This goes beyond commonly used center votes, which would not fully exploit the bounding box annotations. On ScanNet test, our weakly supervised model attains leading performance among other weakly supervised approaches (+18 mAP@50). Remarkably, it also achieves 97% of the mAP@50 score of current fully supervised models. To further illustrate the practicality of our work, we train Box2Mask on the recently released ARKitScenes dataset which is annotated with 3D bounding boxes only, and show, for the first time, compelling 3D instance segmentation masks.



Paperid:1308
Authors:Yao Yao; Jingyang Zhang; Jingbo Liu; Yihang Qu; Tian Fang; David McKinnon; Yanghai Tsin; Long Quan
Title: NeILF: Neural Incident Light Field for Physically-Based Material Estimation
Abstract:
We present a differentiable rendering framework for material and lighting estimation from multi-view images and a reconstructed geometry. In the framework, we represent scene lightings as the Neural Incident Light Field (NeILF) and material properties as the surface BRDF modelled by multi-layer perceptrons. Compared with recent approaches that approximate scene lightings as the 2D environment map, NeILF is a fully 5D light field that is capable of modelling illuminations of any static scenes. In addition, occlusions and indirect lights can be handled naturally by the NeILF representation without requiring multiple bounces of ray tracing, making it possible to estimate material properties even for scenes with complex lightings and geometries. We also propose a smoothness regularization and a Lambertian assumption to reduce the material-lighting ambiguity during the optimization. Our method strictly follows the physically-based rendering equation, and jointly optimizes material and lighting through the differentiable rendering process. We have intensively evaluated the proposed method on our in-house synthetic dataset, the DTU MVS dataset, and real-world BlendedMVS scenes. Our method outperforms previous methods by a significant margin in terms of novel view rendering quality, setting a new state-of-the-art for image-based material and lighting estimation.



Paperid:1309
Authors:Kai Zhang; Nick Kolkin; Sai Bi; Fujun Luan; Zexiang Xu; Eli Shechtman; Noah Snavely
Title: ARF: Artistic Radiance Fields
Abstract:
We present a method for transferring the artistic features of an arbitrary style image to a 3D scene. Previous methods that perform 3D stylization on point clouds or meshes are sensitive to geometric reconstruction errors for complex real-world scenes. Instead, we propose to stylize the more robust radiance field representation. We find that the commonly used Gram matrix-based loss tends to produce blurry results lacking in faithful style detail. We instead utilize a nearest neighbor-based loss that is highly effective at capturing style details while maintaining multi-view consistency. We also propose a novel deferred back-propagation method to enable optimization of memory-intensive radiance fields using style losses defined on full-resolution rendered images. Our evaluation demonstrates that, compared to baselines, our method transfers artistic appearance in a way that more closely resembles the style image. Please see our project webpage for video results and an open-source implementation: https://www.cs.cornell.edu/projects/arf/.



Paperid:1310
Authors:Zeyu Ma; Zachary Teed; Jia Deng
Title: Multiview Stereo with Cascaded Epipolar RAFT
Abstract:
We address multiview stereo (MVS), an important 3D vision task that reconstructs a 3D model such as a dense point cloud from multiple calibrated images. We propose CER-MVS (Cascaded Epipolar RAFT Multiview Stereo), a new approach based on the RAFT (Recurrent All-Pairs Field Transforms) architecture developed for optical flow. CER-MVS introduces five new changes to RAFT: epipolar cost volumes, cost volume cascading, multiview fusion of cost volumes, dynamic supervision, and multiresolution fusion of depth maps. CER-MVS is significantly different from prior work in multiview stereo. Unlike prior work, which operates by updating a 3D cost volume, CER-MVS operates by updating a disparity field. Furthermore, we propose an adaptive thresholding method to balance the completeness and accuracy of the reconstructed point clouds. Experiments show that our approach achieves state-of-the-art performance on the DTU and Tanks-and-Temples benchmarks (both intermediate and advanced set). Code is available at https://github.com/princeton-vl/CER-MVS.



Paperid:1311
Authors:Shaofei Wang; Katja Schwarz; Andreas Geiger; Siyu Tang
Title: ARAH: Animatable Volume Rendering of Articulated Human SDFs
Abstract:
Combining human body models with differentiable rendering has recently enabled animatable avatars of clothed humans from sparse sets of multi-view RGB videos. While state-of-the-art approaches achieve a realistic appearance with neural radiance fields (NeRF), the inferred geometry often lacks detail due to missing geometric constraints. Further, animating avatars in out-of-distribution poses is not yet possible because the mapping from observation space to canonical space does not generalize faithfully to unseen poses. In this work, we address these shortcomings and propose a model to create animatable clothed human avatars with detailed geometry that generalize well to out-of-distribution poses. To achieve detailed geometry, we combine an articulated implicit surface representation with volume rendering. For generalization, we propose a novel joint root-finding algorithm for simultaneous ray-surface intersection search and correspondence search. Our algorithm enables efficient point sampling and accurate point canonicalization while generalizing well to unseen poses. We demonstrate that our proposed pipeline can generate clothed avatars with high-quality pose-dependent geometry and appearance from a sparse set of multi-view RGB videos. Our method achieves state-of-the-art performance on geometry and appearance reconstruction while creating animatable avatars that generalize well to out-of-distribution poses beyond the small number of training poses.



Paperid:1312
Authors:Hongkai Chen; Zixin Luo; Lei Zhou; Yurun Tian; Mingmin Zhen; Tian Fang; David McKinnon; Yanghai Tsin; Long Quan
Title: ASpanFormer: Detector-Free Image Matching with Adaptive Span Transformer
Abstract:
Generating robust and reliable correspondences across images is a fundamental task for a diversity of applications. To capture context at both global and local granularity, we propose ASpanFormer, a Transformer-based detector-free matcher that is built on hierarchical attention structure, adopting a novel attention operation which is capable of adjusting attention span in a self-adaptive manner. To achieve this goal, first, flow maps are regressed in each cross attention phase to locate the center of search region. Next, a sampling grid is generated around the center, whose size, instead of being empirically configured as fixed, is adaptively computed from a pixel uncertainty estimated along with the flow map. Finally, attention is computed across two images within derived regions, referred to as attention span. By these means, we are able to not only maintain long-range dependencies, but also enable fine-grained attention among pixels of high relevance that compensates essential locality and piece-wise smoothness in matching tasks. State-of-the-art accuracy on a wide range of evaluation benchmarks validates the strong matching capability of our method.



Paperid:1313
Authors:Ruiqi Zhang; Jie Chen
Title: NDF: Neural Deformable Fields for Dynamic Human Modelling
Abstract:
We propose Neural Deformable Fields (NDF), a new representation for dynamic human digitization from a multi-view video. Recent works proposed to represent a dynamic human body with shared canonical neural radiance fields which links to the observation space with deformation fields estimations. However, the learned canonical representation is static and the current design of the deformation fields is not able to represent large movements or detailed geometry changes. In this paper, we propose to learn a neural deformable field wrapped around a fitted parametric body model to represent the dynamic human. The NDF is spatially aligned by the underlying reference surface. A neural network is then learned to map pose to the dynamics of NDF. The proposed NDF representation can synthesize the digitized performer with novel views and novel poses with a detailed and reasonable dynamic appearance. Experiments show that our method significantly outperforms recent human synthesis methods.



Paperid:1314
Authors:Itsuki Ueda; Yoshihiro Fukuhara; Hirokatsu Kataoka; Hiroaki Aizawa; Hidehiko Shishido; Itaru Kitahara
Title: Neural Density-Distance Fields
Abstract:
The success of neural fields for 3D vision tasks is now indisputable. Following this trend, several methods aiming for visual localization (e.g., SLAM) have been proposed to estimate distance or density fields using neural fields. However, it is difficult to achieve high localization performance by only density fields-based methods such as Neural Radiance Field (NeRF) since they do not provide density gradient in most empty regions. On the other hand, distance field-based methods such as Neural Implicit Surface (NeuS) have limitations in objects’ surface shapes. This paper proposes Neural Distance-Density Field (NeDDF), a novel 3D representation that reciprocally constrains the distance and density fields. We extend distance field formulation to shapes with no explicit boundary surface, such as fur or smoke, which enable explicit conversion from distance field to density field. Consistent distance and density fields realized by explicit conversion enable both robustness to initial values and high-quality registration. Furthermore, the consistency between fields allows fast convergence from sparse point clouds. Experiments show that NeDDF can achieve high localization performance while providing comparable results to NeRF on novel view synthesis. The code is available at https://github.com/ueda0319/neddf.



Paperid:1315
Authors:Yunxiao Wang; Yanjie Li; Peidong Liu; Tao Dai; Shu-Tao Xia
Title: NeXT: Towards High Quality Neural Radiance Fields via Multi-Skip Transformer
Abstract:
Neural Radiance Fields (NeRF) methods show impressive performance for novel view synthesis by representing a scene via a neural network. However, most existing NeRF based methods, including its variants, treat each sample point individually as input, while ignoring the inherent relationships between adjacent sample points from the corresponding rays, thus hindering the reconstruction performance. To address this issue, we explore a brand new scheme, namely NeXT, introducing a multi-skip transformer to capture the rich relationships between various sample points in a ray-level query. Specifically, ray tokenization is proposed to represent each ray as a sequence of point embeddings which is taken as input of our proposed NeXT. In this way, relationships between sample points are captured via the built-in self-attention mechanism to promote the reconstruction. Besides, our proposed NeXT can be easily combined with other NeRF based methods to improve their rendering quality. Extensive experiments conducted on three datasets demonstrate that NeXT significantly outperforms all previous state-of- the-art work by a large margin. In particular, the proposed NeXT surpasses the strong NeRF baseline by 2.74 dB of PSNR on Blender dataset. The code is available at https://github.com/Crishawy/NeXT.



Paperid:1316
Authors:Erik Sandström; Martin R. Oswald; Suryansh Kumar; Silvan Weder; Fisher Yu; Cristian Sminchisescu; Luc Van Gool
Title: Learning Online Multi-sensor Depth Fusion
Abstract:
Many hand-held or mixed reality devices are used with a single sensor for 3D reconstruction, although they often comprise multiple sensors. Multi-sensor depth fusion is able to substantially improve the robustness and accuracy of 3D reconstruction methods, but existing techniques are not robust enough to handle sensors which operate with diverse value ranges as well as noise and outlier statistics. To this end, we introduce SenFuNet, a depth fusion approach that learns sensor-specific noise and outlier statistics and combines the data streams of depth frames from different sensors in an online fashion. Our method fuses multi-sensor depth streams regardless of time synchronization and calibration and generalizes well with little training data. We conduct experiments with various sensor combinations on the real-world CoRBS and Scene3D datasets, as well as the Replica dataset. Experiments demonstrate that our fusion strategy outperforms traditional and recent online depth fusion approaches. In addition, the combination of multiple sensors yields more robust outlier handling and more precise surface reconstruction than the use of a single sensor. The source code and data are available at https://github.com/tfy14esa/SenFuNet.



Paperid:1317
Authors:Yuanbo Xiangli; Linning Xu; Xingang Pan; Nanxuan Zhao; Anyi Rao; Christian Theobalt; Bo Dai; Dahua Lin
Title: BungeeNeRF: Progressive Neural Radiance Field for Extreme Multi-Scale Scene Rendering
Abstract:
Neural Radiance Field (NeRF) has achieved outstanding performance in modeling 3D objects and controlled scenes, usually under a single scale. In this work, we focus on multi-scale cases where large changes in imagery are observed at drastically different scales. This scenario vastly exists in the real world, such as city scenes, with views ranging from satellite level that captures the overview of a city, to ground level imagery showing complex details of an architecture; and can also be commonly identified in landscape and delicate minecraft 3D models. The wide span of viewing points within these scenes yields multiscale renderings with very different levels of detail, which poses great challenges to neural radiance field and biases it towards compromised results. To address these issues, we introduce BungeeNeRF, a progressive neural radiance field that achieves level-of-detail rendering across drastically varied scales. Starting from fitting distant views with a shallow base block, as training progresses, new blocks are appended to accommodate the emerging details in the increasingly closer views. The strategy progressively activates high-frequency channels in NeRF’s positional encoding inputs to unfold more complex details as the training proceeds. We demonstrate the superiority of BungeeNeRF in modeling diverse multi-scale scenes with drastically varying views on multiple data sources (e.g., city models, synthetic, and drone captured data), and its support for high-quality rendering in different levels of detail.



Paperid:1318
Authors:Huizong Yang; Anthony Yezzi
Title: Decomposing the Tangent of Occluding Boundaries according to Curvatures and Torsions
Abstract:
This paper develops new insight into the local structure of occluding boundaries on 3D surfaces. Prior literature has addressed the relationship between 3D occluding boundaries and their 2D image projections by radial curvature, planar curvature, and Gaussian curvature. Occluding boundaries have also been studied implicitly as intersections of level surfaces, avoiding their explicit description in terms of local surface geometry. In contrast, this work studies and characterizes the local structure of occluding curves explicitly in terms of the local geometry of the surface. We show how the first order structure of the occluding curve (its tangent) can be extracted from the second order structure of the surface purely along the viewing direction, without the need to consider curvatures or torsions in other directions. We derive a theorem to show that the tangent vector of the occluding boundary exhibits a strikingly elegant decomposition along the viewing direction and its orthogonal tangent, where the decomposition weights precisely match the geodesic torsion and the normal curvature of the surface respectively only along the line-of-sight! Though the focus of this paper is an enhanced theoretical understanding of the occluding curve in the continuum, we nevertheless demonstrate its potential numerical utility in a straight-forward marching method to explicitly trace out the occluding curve. We also present mathematical analysis to show the relevance of this theory to computer vision and how it might be leveraged in more accurate future algorithms for 2D/3D registration and/or multiview stereo reconstruction.



Paperid:1319
Authors:Jiepeng Wang; Peng Wang; Xiaoxiao Long; Christian Theobalt; Taku Komura; Lingjie Liu; Wenping Wang
Title: NeuRIS: Neural Reconstruction of Indoor Scenes Using Normal Priors
Abstract:
Reconstructing 3D indoor scenes from 2D images is an important task in many computer vision and graphics applications. A main challenge in this task is that large texture-less areas in typical indoor scenes make existing methods struggle to produce satisfactory reconstruction results. We propose a new method, dubbed NeuRIS, for high-quality reconstruction of indoor scenes. The key idea of NeuRIS is to integrate estimated normal vectors of indoor scenes as a prior in a neural rendering framework for reconstructing large texture-less shapes and, importantly, does so in an adaptive manner to also enable the reconstruction of irregular shapes with fine details. Specifically, we evaluate the faithfulness of the normal priors on-the-fly by checking the multi-view consistency of reconstruction during the optimization process. Only the normal priors accepted as faithful will be utilized for 3D reconstruction, which typically happens in regions of smooth shapes possibly with weak texture. However, for those regions with small objects or thin structure, for which the normal priors are usually unreliable, we will only rely on visual features of the input images, since such regions typically contain relatively rich visual features (e.g., shade changes and boundary contours). Extensive experiments show that NeuRIS significantly outperforms the state-of-the-art methods in terms of reconstruction quality.



Paperid:1320
Authors:Mohammed Suhail; Carlos Esteves; Leonid Sigal; Ameesh Makadia
Title: Generalizable Patch-Based Neural Rendering
Abstract:
Neural rendering has received tremendous attention since the advent of Neural Radiance Fields (NeRF), and has pushed the state-of-the-art on novel-view synthesis considerably. The recent focus has been on models that overfit to a single scene, and the few attempts to learn models that can synthesize novel views of unseen scenes mostly consist of combining deep convolutional features with a NeRF-like model. We propose a different paradigm, where no deep features and no NeRF-like volume rendering are needed. Our method is capable of predicting the color of a target ray in a novel scene directly, just from a collection of patches sampled from the scene. We first leverage epipolar geometry to extract patches along the epipolar lines of each reference view. Each patch is linearly projected into a 1D feature vector and a sequence of transformers process the collection. For positional encoding, we parameterize rays as in a light field representation, with the crucial difference that the coordinates are canonicalized with respect to the target ray, which makes our method independent of the reference frame and improves generalization. We show that our approach outperforms the state-of-the-art on novel view synthesis of unseen scenes even when being trained with considerably less data than prior work.



Paperid:1321
Authors:Ziming Wang; Xiaoliang Huo; Zhenghao Chen; Jing Zhang; Lu Sheng; Dong Xu
Title: Improving RGB-D Point Cloud Registration by Learning Multi-Scale Local Linear Transformation
Abstract:
Point cloud registration aims at estimating the geometric transformation between two point cloud scans, in which accurate correspondence estimation is the key to its success. In addition to previous methods that seek correspondences by hand-crafted or learnt geometric features, recent point cloud registration methods have tried to apply RGB-D data to achieve more accurate correspondence. However, it is not trivial to effectively fuse the geometric and visual information from these two distinctive modalities, especially for the registration problem. In this work, we propose a new Geometry-Aware Visual Feature Extractor (GAVE) that employs multi-scale local linear transformation to progressively fuse these two modalities, where the geometric features from the depth data act as the geometry-dependent convolution kernels to transform the visual features from the RGB data. The resultant visual-geometric features are in canonical feature spaces with alleviated visual dissimilarity caused by geometric changes, by which more reliable correspondence can be achieved. The proposed GAVE module can be readily plugged into recent RGB-D point cloud registration framework. Extensive experiments on 3D Match and ScanNet demonstrate that our method outperforms the state-of-the-art point cloud registration methods even without correspondence or pose supervision.



Paperid:1322
Authors:Hao Ouyang; Bo Zhang; Pan Zhang; Hao Yang; Jiaolong Yang; Dong Chen; Qifeng Chen; Fang Wen
Title: Real-Time Neural Character Rendering with Pose-Guided Multiplane Images
Abstract:
We propose pose-guided multiplane image (MPI) synthesis which can render an animatable character in real scenes with photorealistic quality. We use a portable camera rig to capture the multi-view images along with the driving signal for the moving subject. Our method generalizes the image-to-image translation paradigm, which translates the human pose to a 3D scene representation -- MPIs that can be rendered in free viewpoints, using the multi-views captures as supervision. To fully cultivate the potential of MPI, we propose depth-adaptive MPI which can be learned using variable exposure images while being robust to inaccurate camera registration. Our method demonstrates advantageous novel-view synthesis quality over the state-of-the-art approaches for characters with challenging motions. Moreover, the proposed method is generalizable to novel combinations of training poses and can be explicitly controlled. Our method achieves such expressive and animatable character rendering all in real-time, serving as a promising solution for practical applications.



Paperid:1323
Authors:Xiaoxiao Long; Cheng Lin; Peng Wang; Taku Komura; Wenping Wang
Title: SparseNeuS: Fast Generalizable Neural Surface Reconstruction from Sparse Views
Abstract:
We introduce SparseNeuS, a novel neural rendering based method for the task of surface reconstruction from multi-view images. This task becomes more difficult when only sparse images are provided as input, a scenario where existing neural reconstruction approaches usually produce incomplete or distorted results. Moreover, their inability of generalizing to unseen new scenes impedes their application in practice. Contrarily, SparseNeuS can generalize to new scenes and work well with sparse images (as few as 2 or 3). SparseNeuS adopts signed distance function (SDF) as the surface representation, and learns generalizable priors from image features by introducing geometry encoding volumes for generic surface prediction. Moreover, several strategies are introduced to effectively leverage sparse views for high-quality reconstruction, including 1) a multi-level geometry reasoning framework to recover the surfaces in a coarse-to-fine manner; 2) a multi-scale color blending scheme for more reliable color prediction; 3) a consistency-aware fine-tuning scheme to control the inconsistent regions caused by occlusion and noise. Extensive experiments demonstrate that our approach not only outperforms the state-of-the-art methods, but also exhibits good efficiency, generalizability, and flexibility.



Paperid:1324
Authors:Ziyue Feng; Liang Yang; Longlong Jing; Haiyan Wang; YingLi Tian; Bing Li
Title: Disentangling Object Motion and Occlusion for Unsupervised Multi-Frame Monocular Depth
Abstract:
Conventional self-supervised monocular depth prediction methods are based on a static environment assumption, which leads to accuracy degradation in dynamic scenes due to the mismatch and occlusion problems introduced by object motions. Existing dynamic-object-focused methods only partially solved the mismatch problem at the training loss level. In this paper, we accordingly propose a novel multi-frame monocular depth prediction method to solve these problems at both the prediction and supervision loss levels. Our method, called DynamicDepth, is a new framework trained via a self-supervised cycle consistent learning scheme. A Dynamic Object Motion Disentanglement (DOMD) module is proposed to disentangle object motions to solve the mismatch problem. Moreover, novel occlusion-aware Cost Volume and Re-projection Loss are designed to alleviate the occlusion effects of object motions. Extensive analyses and experiments on the Cityscapes and KITTI datasets show that our method significantly outperforms the state-of-the-art monocular depth prediction methods, especially in the areas of dynamic objects. Code is available at https://github.com/AutoAILab/DynamicDepth"



Paperid:1325
Authors:Vitor Guizilini; Igor Vasiljevic; Jiading Fang; Rareș Ambruș; Greg Shakhnarovich; Matthew R. Walter; Adrien Gaidon
Title: Depth Field Networks for Generalizable Multi-View Scene Representation
Abstract:
Modern 3D computer vision leverages learning to boost geometric reasoning, mapping image data to classical structures such as cost volumes or epipolar constraints to improve matching. These architectures are specialized according to the particular problem, and thus require significant task-specific tuning, often leading to poor domain generalization performance. Recently, generalist Transformer architectures have achieved impressive results in tasks such as optical flow and depth estimation by encoding geometric priors as inputs rather than as enforced constraints. In this paper, we extend this idea and propose to learn an implicit, multi-view consistent scene representation, introducing a series of 3D data augmentation techniques as a geometric inductive prior to increase view diversity. We also show that introducing view synthesis as an auxiliary task further improves depth estimation. Our Depth Field Networks (DeFiNe) achieve state-of-the-art results in stereo and video depth estimation without explicit geometric constraints, and improve on zero-shot domain generalization by a wide margin.



Paperid:1326
Authors:Weiyu Guo; Zhaoshuo Li; Yongkui Yang; Zheng Wang; Russell H. Taylor; Mathias Unberath; Alan Yuille; Yingwei Li
Title: Context-Enhanced Stereo Transformer
Abstract:
Stereo depth estimation is of great interest for computer vision research. However, existing methods struggles to generalize and predict reliably in hazardous regions, such as large uniform regions. To overcome these limitations, we propose Context Enhanced Path (CEP). CEP improves the generalization and robustness against common failure cases in existing solutions by capturing the long-range global information. We construct our stereo depth estimation model, Context Enhanced Stereo Transformer (CEST), by plugging CEP into the state-of-the-art stereo depth estimation method Stereo Transformer. CEST is examined on distinct public datasets, such as Scene Flow, Middlebury-2014, KITTI-2015, and MPI-Sintel. We find CEST outperforms prior approaches by a large margin. For example, in the zero-shot synthetic-to-real setting, CEST outperforms the best competing approaches on Middlebury-2014 dataset by 11%. Our extensive experiments demonstrate that the long-range information is critical for stereo matching task and CEP successfully captures such information.



Paperid:1327
Authors:Zhelun Shen; Yuchao Dai; Xibin Song; Zhibo Rao; Dingfu Zhou; Liangjun Zhang
Title: PCW-Net: Pyramid Combination and Warping Cost Volume for Stereo Matching
Abstract:
Existing deep learning based stereo matching methods either focus on achieving optimal performances on the target dataset while with poor generalization for other datasets or focus on handling the cross-domain generalization by suppressing the domain sensitive features which results in a significant sacrifice on the performance. To tackle these problems, we propose PCW-Net, a Pyramid Combination and Warping cost volume-based network to achieve good performance on both cross-domain generalization and stereo matching accuracy on various benchmarks. In particular, our PCW-Net is designed for two purposes. First, we construct combination volumes on the upper levels of the pyramid and develop a cost volume fusion module to integrate them for initial disparity estimation. Multi-scale receptive fields can be covered by fusing multi-scale combination volumes, thus, domain-invariant features can be extracted. Second, we construct the warping volume at the last level of the pyramid for disparity refinement. The proposed warping volume can narrow down the residue searching range from the initial disparity searching range to a fine-grained one, which can dramatically alleviate the difficulty of the network to find the correct residue in an unconstrained residue searching space. When training on synthetic datasets and generalizing to unseen real datasets, our method shows strong cross-domain generalization and outperforms existing state-of-the-arts with a large margin. After fine-tuning on the real datasets, our method ranks first on KITTI 2012, second on KITTI 2015, and first on the Argoverse among all published methods as of 6, March 2022.



Paperid:1328
Authors:Yuan Liu; Yilin Wen; Sida Peng; Cheng Lin; Xiaoxiao Long; Taku Komura; Wenping Wang
Title: Gen6D: Generalizable Model-Free 6-DoF Object Pose Estimation from RGB Images
Abstract:
In this paper, we present a generalizable model-free 6-DoF object pose estimator called Gen6D. Existing generalizable pose estimators either need the high-quality object models or require additional depth maps or object masks in test time, which significantly limits their application scope. In contrast, our pose estimator only requires some posed images of the unseen object and is able to accurately predict poses of the object in arbitrary environments. Gen6D consists of an object detector, a viewpoint selector and a pose refiner, all of which do not require the 3D object model and can generalize to unseen objects. Experiments show that Gen6D achieves state-of-the-art results on two model-free datasets: the MOPED dataset and a new GenMOP dataset. In addition, on the LINEMOD dataset, Gen6D achieves competitive results compared with instance-specific pose estimators.



Paperid:1329
Authors:Zixing Lei; Shunli Ren; Yue Hu; Wenjun Zhang; Siheng Chen
Title: Latency-Aware Collaborative Perception
Abstract:
Collaborative perception has recently shown great potential to improve perception capabilities over single-agent perception. Existing collaborative perception methods usually consider an ideal communication environment. However, in practice, the communication system inevitably suffers from latency issues, causing potential performance degradation and high risks in safety-critical applications, such as autonomous driving. To mitigate the effect caused by the inevitable latency, from a machine learning perspective, we present the first latency-aware collaborative perception system, which actively adapts asynchronous perceptual features from multiple agents to the same time stamp, promoting the robustness and effectiveness of collaboration. To achieve such a featurelevel synchronization, we propose a novel latency compensation module, called SyncNet, which leverages feature-attention symbiotic estimation and time modulation techniques. Experiments results show that the proposed latency aware collaborative perception system with SyncNet can outperforms the state-of-the-art collaborative perception method by 15.6% in the communication latency scenario and keep collaborative perception being superior to single agent perception under severe latency.



Paperid:1330
Authors:Anpei Chen; Zexiang Xu; Andreas Geiger; Jingyi Yu; Hao Su
Title: TensoRF: Tensorial Radiance Fields
Abstract:
We present TensoRF, a novel approach to model and reconstruct radiance fields. Unlike NeRF that purely uses MLPs, we model the radiance field of a scene as a 4D tensor, which represents a 3D voxel grid with per-voxel multi-channel features. Our central idea is to factorize the 4D scene tensor into multiple compact low-rank tensor components. We demonstrate that applying traditional CANDECOMP/PARAFAC (CP) decomposition -- that factorizes tensors into rank-one components with compact vectors -- in our framework leads to improvements over vanilla NeRF. To further boost performance, we introduce a novel vector-matrix (VM) decomposition that relaxes the low-rank constraints for two modes of a tensor and factorizes tensors into compact vector and matrix factors. Beyond superior rendering quality, our models with CP and VM decompositions lead to a significantly lower memory footprint in comparison to previous and concurrent works that directly optimize per-voxel features. Experimentally, we demonstrate that TensoRF with CP decomposition achieves fast reconstruction (<30 min) with better rendering quality and even a smaller model size (<4 MB) compared to NeRF. Moreover, TensoRF with VM decomposition further boosts rendering quality and outperforms previous state-of-the-art methods, while reducing the reconstruction time (<10 min) and retaining a compact model size (<75 MB).



Paperid:1331
Authors:Luca Cavalli; Marc Pollefeys; Daniel Barath
Title: NeFSAC: Neurally Filtered Minimal Samples
Abstract:
Since RANSAC, a great deal of research has been devoted to improving both its accuracy and run-time. Still, only a few methods aim at recognizing invalid minimal samples early, before the often expensive model estimation and quality calculation are done. To this end, we propose NeFSAC, an efficient algorithm for neural filtering of motion-inconsistent and poorly-conditioned minimal samples. We train NeFSAC to predict the probability of a minimal sample leading to an accurate relative pose, only based on the pixel coordinates of the image correspondences. Our neural filtering model learns typical motion patterns of samples which lead to unstable poses, and regularities in the possible motions to favour well-conditioned and likely-correct samples. The novel lightweight architecture implements the main invariants of minimal samples for pose estimation, and a novel training scheme addresses the problem of extreme class imbalance. NeFSAC can be plugged into any existing RANSAC-based pipeline. We integrate it into USAC and show that it consistently provides strong speed-ups even under extreme train-test domain gaps -- for example, the model trained for the autonomous driving scenario works on PhotoTourism too. We tested NeFSAC on more than 100k image pairs from three publicly available real-world datasets and found that it leads to one order of magnitude speed-up, while often finding more accurate results than USAC alone. The source code is available at https://github.com/cavalli1234/NeFSAC.



Paperid:1332
Authors:Eldar Insafutdinov; Dylan Campbell; João F. Henriques; Andrea Vedaldi
Title: SNeS: Learning Probably Symmetric Neural Surfaces from Incomplete Data
Abstract:
We present a method for the accurate 3D reconstruction of partly-symmetric objects. We build on the strengths of recent advances in neural reconstruction and rendering such as Neural Radiance Fields (NeRF). A major shortcomings of such approaches is that they fail to reconstruct any part of the object which is not clearly visible in the training image, which is often the case for in-the-wild images and videos. When evidence is lacking, structural priors such as symmetry can be used to complete the missing information. However, exploiting such priors in neural rendering is highly non-trivial: while geometry and non-reflective materials may be symmetric, shadows and reflections from the ambient scene are not symmetric in general. To address this, we apply a soft symmetry constraint to the 3D geometry and material properties, having factored appearance into lighting, albedo colour and reflectivity. We evaluate our method on the recently introduced CO3D dataset, focusing on the car category due to the challenge of reconstructing highly-reflective materials. We show that it can reconstruct unobserved regions with high fidelity and render high-quality novel view images.



Paperid:1333
Authors:Kim Jun-Seong; Kim Yu-Ji; Moon Ye-Bin; Tae-Hyun Oh
Title: HDR-Plenoxels: Self-Calibrating High Dynamic Range Radiance Fields
Abstract:
We propose high dynamic range radiance (HDR) fields, HDR-Plenoxels, that learns a plenoptic function of 3D HDR radiance fields, geometry information, and varying camera settings inherent in 2D low dynamic range (LDR) images. Our voxel-based volume rendering pipeline reconstructs HDR radiance fields with only multi-view LDR images taken from varying camera settings in an end-to-end manner and has a fast convergence speed. To deal with various cameras in real-world scenario, we introduce a tone mapping module that models the digital in camera imaging pipeline (ISP) and disentangles radiometric settings. Our tone mapping module allows us to render by controlling the radiometric settings of each novel view. Finally, we build a multi-view dataset with varying camera conditions, which fits our problem setting. Our experiments show that HDR-Plenoxels can express detail and high-quality HDR novel views from only LDR images with various cameras.



Paperid:1334
Authors:Wei Jiang; Kwang Moo Yi; Golnoosh Samei; Oncel Tuzel; Anurag Ranjan
Title: NeuMan: Neural Human Radiance Field from a Single Video
Abstract:
Photorealistic rendering and reposing of humans is important for enabling augmented reality experiences. We propose a novel framework to reconstruct the human and the scene that can be rendered with novel human poses and views from just a single in-the-wild video. Given a video captured by a moving camera, we train two NeRF models: a human NeRF model and a scene NeRF model. To train these models, we rely on existing methods to estimate the rough geometry of the human and the scene. Those rough geometry estimates allow us to create a warping field from the observation space to the canonical pose-independent space, where we train the human model in. Our method is able to learn subject specific details, including cloth wrinkles and accessories, from just a 10 seconds video clip, and to provide high quality renderings of the human under novel poses, from novel views, together with the background.



Paperid:1335
Authors:Ruilong Li; Julian Tanke; Minh Vo; Michael Zollhöfer; Jürgen Gall; Angjoo Kanazawa; Christoph Lassner
Title: TAVA: Template-Free Animatable Volumetric Actors
Abstract:
Coordinate-based volumetric representations have the potential to generate photo-realistic virtual avatars from images. However, virtual avatars need to be controllable and be rendered in novel poses that may not have been observed. Traditional techniques, such as LBS, provide such a controlling function; yet it usually requires a hand-designed body template, 3D scan data, and surface-based appearance models. On the other hand, neural representations have been shown to be powerful in representing visual details, but are under-explored in dynamic and articulated settings. In this paper, we propose TAVA, a method to create Template-free Animatable Volumetric Actors, based on neural representations. We rely solely on multi-view data and a tracked skeleton to create a volumetric model of an actor, which can be animated at test time given novel poses. Since TAVA does not require a body template, it is applicable to humans as well as other creatures such as animals. Furthermore, TAVA is designed such that it can recover accurate dense correspondences, making it amenable to content-creation and editing tasks. Through extensive experiments, we demonstrate that he proposed method generalizes well to novel poses as well as unseen views and showcase basic editing capabilities. The code is available at https://github.com/facebookresearch/tava"



Paperid:1336
Authors:Qiang Wang; Shaohuai Shi; Kaiyong Zhao; Xiaowen Chu
Title: EASNet: Searching Elastic and Accurate Network Architecture for Stereo Matching
Abstract:
Recent advanced studies have spent considerable human efforts on optimizing network architectures for stereo matching but hardly achieved both high accuracy and fast inference speed. To ease the workload in network design, neural architecture search (NAS) has been applied with great success to various sparse prediction tasks, such as image classification and object detection. Recent advanced studies have spent considerable human efforts on optimizing network architectures for stereo matching but hardly achieved both high accuracy and fast inference speed. To ease the workload in network design, neural architecture search (NAS) has been applied with great success to various sparse prediction tasks, such as image classification and object detection. However, existing NAS studies on the dense prediction task, especially stereo matching, still cannot be efficiently and effectively deployed on devices of different computing capability. To this end, we propose to train an \underline{e}lastic and \underline{a}ccurate network for \underline{s}tereo matching (EASNet) that supports various 3D architectural settings on devices with different compute capability. Given the deployment latency constraint on the target device, we can quickly extract a sub-network from the full EASNet without additional training while the accuracy of the sub-network can still be maintained. Extensive experiments show that our EASNet outperforms both state-of-the-art human-designed and NAS-based architectures on Scene Flow and MPI Sintel datasets in terms of model accuracy and inference speed. Particularly, deployed on an inference GPU, EASNet achieves a new SOTA 0.73 EPE on the Scene Flow dataset with 100 ms, which is 4.5x faster than LEAStereo with a better quality model. The codes of EASNet are available at: https://github.com/HKBU-HPML/EASNet.git"



Paperid:1337
Authors:Daniel Barath; Zuzana Kukelova
Title: Relative Pose from SIFT Features
Abstract:
This paper proposes the geometric relationship of epipolar geometry and orientation- and scale-covariant, e.g., SIFT, features. We derive a new linear constraint relating the unknown elements of the fundamental matrix and the orientation and scale. This equation can be used together with the well-known epipolar constraint to, e.g., estimate the fundamental matrix from four SIFT correspondences, essential matrix from three, and to solve the semi-calibrated case from three correspondences. Requiring fewer correspondences than the well-known point-based approaches (e.g., 5PT, 6PT and 7PT solvers) for epipolar geometry estimation makes RANSAC-like randomized robust estimation significantly faster. The proposed constraint is tested on a number of problems in a synthetic environment and on publicly available real-world datasets on more than 80000 image pairs. It is superior to the state-of-the-art in terms of processing time while often leading more accurate results.



Paperid:1338
Authors:Hoonhee Cho; Kuk-Jin Yoon
Title: Selection and Cross Similarity for Event-Image Deep Stereo
Abstract:
Standard frame-based cameras have shortcomings of low dynamic range and motion blur in real applications. On the other hand, event cameras, which are bio-inspired sensors, asynchronously output the polarity values of pixel-level log intensity changes and report continuous stream data even under fast motion with a high dynamic range. Therefore, event cameras are effective in stereo depth estimation under challenging illumination conditions and/or fast motion. To estimate the disparity map with events, existing state-of-the-art event-based stereo models use the image together with past events occurred up to the current image acquisition time. However, not all events equally contribute to the disparity estimation of the current frame, since past events occur at different times under different movements with different disparity values. Therefore, events need to be carefully selected for accurate event-guided disparity estimation. In this paper, we aim to effectively deal with events that continuously occur with different disparity in the scene depending on the camera’s movement. To this end, we first propose the differentiable event selection network to select the most relevant events for current depth estimation. Furthermore, we effectively use feature-like events triggered around the boundary of objects, leading them to serve as ideal guides in disparity estimation. To that end, we propose a neighbor cross similarity feature (NCSF) that considers the similarity between different modalities. Our experiments on various datasets demonstrate the superiority of our method to estimate the depth using images and event data together. Our project code is available at: https://github.com/Chohoonhee/SCSNet.



Paperid:1339
Authors:Zhenyu Chen; Qirui Wu; Matthias Nießner; Angel X. Chang
Title: D3Net: A Unified Speaker-Listener Architecture for 3D Dense Captioning and Visual Grounding
Abstract:
Recent work on dense captioning and visual grounding in 3D have achieved impressive results. Despite developments in both areas, the limited amount of available 3D vision-language data causes overfitting issues for 3D visual grounding and 3D dense captioning methods. Also, how to discriminatively describe objects in complex 3D environments is not fully studied yet. To address these challenges, we present D3Net, an end-to-end neural speaker-listener architecture that can detect, describe and discriminate. Our D3Net unifies dense captioning and visual grounding in 3D in a self-critical manner. This self-critical property of D3Net encourages generation of discriminative object captions and enables semi-supervised training on scan data with partially annotated descriptions. Our method outperforms SOTA methods in both tasks on the ScanRefer dataset, surpassing the SOTA 3D dense captioning method by a significant margin.



Paperid:1340
Authors:Hao-Xiang Chen; Jiahui Huang; Tai-Jiang Mu; Shi-Min Hu
Title: CIRCLE: Convolutional Implicit Reconstruction and Completion for Large-Scale Indoor Scene
Abstract:
We present CIRCLE, a framework for large-scale scene completion and geometric refinement based on local implicit signed distance functions. It is based on an end-to-end sparse convolutional network, CircNet, which jointly models local geometric details and global scene structural contexts, allowing it to preserve fine-grained object detail while recovering missing regions commonly arising in traditional 3D scene data. A novel differentiable rendering module further enables a test-time refinement for better reconstruction quality. Extensive experiments on both real-world and synthetic datasets show that our concise framework is effective, achieving better reconstruction quality while being significantly faster.



Paperid:1341
Authors:Wang Zhao; Shaohui Liu; Hengkai Guo; Wenping Wang; Yong-Jin Liu
Title: ParticleSfM: Exploiting Dense Point Trajectories for Localizing Moving Cameras in the Wild
Abstract:
Estimating the pose of a moving camera from monocular video is a challenging problem, especially due to the presence of moving objects in dynamic environments, where the performance of existing camera pose estimation methods are susceptible to pixels that are not geometrically consistent. To tackle this challenge, we present a robust dense indirect structure-from-motion method for videos that is based on dense correspondence initialized from pairwise optical flow. Our key idea is to optimize long range video correspondence as dense point trajectories and use it to learn robust estimation of motion segmentation. A novel neural network architecture is proposed for processing irregular point trajectory data. Camera poses are then estimated and optimized with global bundle adjustment over the portion of long-range point trajectories that are classified as static. Experiments on MPI Sintel dataset show that our system produces significantly more accurate camera trajectories compared to existing state-of-the-art methods. In addition, our method is able to retain reasonable accuracy of camera poses on fully static scenes, which consistently outperforms strong state-of-the-art dense correspondence based methods with end-to-end deep learning, demonstrating the potential of dense indirect methods based on optical flow and point trajectories. As the point trajectory representation is general, we further present results and comparisons on in-the-wild monocular videos with complex motion of dynamic objects. Code is available at https://github.com/bytedance/particle-sfm.



Paperid:1342
Authors:Yujin Chen; Matthias Nießner; Angela Dai
Title: 4DContrast: Contrastive Learning with Dynamic Correspondences for 3D Scene Understanding
Abstract:
We present a new approach to instill 4D dynamic object priors into learned 3D representations by unsupervised pre-training. We observe that dynamic movement of an object through an environment provides important cues about its objectness, and thus propose to imbue learned 3D representations with such dynamic understanding, that can then be effectively transferred to improved performance in downstream 3D semantic scene understanding tasks. We propose a new data augmentation scheme leveraging synthetic 3D shapes moving in static 3D environments, and employ contrastive learning under 3D-4D constraints that encode 4D invariances into the learned 3D representations. Experiments demonstrate that our unsupervised representation learning results in improvement in downstream 3D semantic segmentation, object detection, and instance segmentation tasks, and moreover, notably improves performance in data-scarce scenarios. Our results show that our 4D pre-training method improves downstream tasks such as object detection mAP@0.5 by 5.5%/6.5% over training from scratch on ScanNet/SUN RGB-D while involving no additional run-time overhead at test time.



Paperid:1343
Authors:Amine Ouasfi; Adnane Boukhayma
Title: Few `\textit{Zero Level Set}'-Shot Learning of Shape Signed Distance Functions in Feature Space
Abstract:
We explore a new idea for learning based shape reconstruction from a point cloud, based on the recently popularized implicit neural shape representations. We cast the problem as a few-shot learning of implicit neural signed distance functions in feature space, that we approach using gradient based meta-learning. We use a convolutional encoder to build a feature space given the input point cloud. An implicit decoder learns to predict signed distance values given points represented in this feature space. Setting the input point cloud, i.e. samples from the target shape function’s zero level set, as the support (i.e. context) in few-shot learning terms, we train the decoder such that it can adapt its weights to the underlying shape of this context with a few (5) tuning steps. We thus combine two types of implicit neural network conditioning mechanisms simultaneously for the first time, namely feature encoding and meta-learning. Our numerical and qualitative evaluation shows that in the context of implicit reconstruction from a sparse point cloud, our proposed strategy, i.e. meta-learning in feature space, outperforms existing alternatives, namely standard supervised learning in feature space, and meta-learning in euclidean space, while still providing fast inference.



Paperid:1344
Authors:Gaku Nakano
Title: Solution Space Analysis of Essential Matrix Based on Algebraic Error Minimization
Abstract:
This paper reports on a solution space analysis of the essential matrix based on algebraic error minimization. Although it has been known since 1988 that an essential matrix has at most 10 real solutions for five-point pairs, the number of solutions in the least-squares case has not been explored. We first derive that the Karush-Kuhn-Tucker conditions of algebraic errors satisfying the Demazure constraints can be represented by a system of polynomial equations without Lagrange multipliers. Then, using computer algebra software, we reveal that the simultaneous equation has at most 220 real solutions, which can be obtained by the Gauss-Newton method, Groebner basis, and homotopy continuation. Through experiments on synthetic and real data, we quantitatively evaluate the convergence of the proposed and the existing methods to globally optimal solutions. Finally, we visualize a spatial distribution of the global and local minima in 3D space.



Paperid:1345
Authors:Leonid Keselman; Martial Hebert
Title: Approximate Differentiable Rendering with Algebraic Surfaces
Abstract:
Differentiable renderers provide a direct mathematical link between an object’s 3D representation and images of that object. In this work, we develop an approximate differentiable renderer for a compact, interpretable representation, which we call Fuzzy Metaballs. Our approximate renderer focuses on rendering shapes via depth maps and silhouettes. It sacrifices fidelity for utility, producing fast runtimes and high-quality gradient information that can be used to solve vision tasks. Compared to mesh-based differentiable renderers, our method has forward passes that are 5x faster and backwards passes that are 30x faster. The depth maps and silhouette images generated by our method are smooth and defined everywhere. In our evaluation of differentiable renderers for pose estimation, we show that our method is the only one comparable to classic techniques. In shape from silhouette, our method performs well using only gradient descent and a per-pixel loss, without any surrogate losses or regularization. These reconstructions work well even on natural video sequences with segmentation artifacts.



Paperid:1346
Authors:Will Hutchcroft; Yuguang Li; Ivaylo Boyadzhiev; Zhiqiang Wan; Haiyan Wang; Sing Bing Kang
Title: CoVisPose: Co-Visibility Pose Transformer for Wide-Baseline Relative Pose Estimation in 360° Indoor Panoramas
Abstract:
We present CoVisPose, a new end-to-end supervised learning method for relative camera pose estimation in wide baseline 360 indoor panoramas. To address the challenges of occlusion, perspective changes, and textureless or repetitive regions, we generate rich representations for direct pose regression by jointly learning dense bidirectional visual overlap, correspondence, and layout geometry. We estimate three image column-wise quantities: co-visibility (the probability that a given column’s image content is seen in the other panorama), angular correspondence (angular matching of columns across panoramas), and floor layout (the vertical floor-wall boundary angle). We learn these dense outputs by applying a transformer over the image-column feature sequences, which cover the full 360 field-of-view (FoV) from both panoramas. The resultant rich representation supports learning robust relative poses with an efficient 1D convolutional decoder. In addition to learned direct pose regression with scale, our network also supports pose estimation through a RANSAC-based rigid registration of the predicted corresponding layout boundary points. Our method is robust to extremely wide baselines with very low visual overlap, as well as significant occlusions. We improve upon the SOTA by a large margin, as demonstrated on a large-scale dataset of real homes, ZInD.



Paperid:1347
Authors:Banglei Guan; Ji Zhao
Title: Affine Correspondences between Multi-Camera Systems for 6DOF Relative Pose Estimation
Abstract:
We present a novel method to compute the 6DOF relative pose of multi-camera systems using two affine correspondences (ACs). Existing solutions to the multi-camera relative pose estimation are either restricted to special cases of motion, have too high computational complexity, or require too many point correspondences (PCs). Thus, these solvers impede an efficient or accurate relative pose estimation when applying RANSAC as a robust estimator. This paper shows that the relative pose estimation problem using ACs permits a feasible minimal solution, when exploiting the geometric constraints between ACs and multi-camera systems using a special parameterization. We present a problem formulation based on two ACs that encompass two common types of ACs across two views, i.e., inter-camera and intra-camera. Experiments on both virtual and real multi-camera systems prove that the proposed solvers are more efficient than the state-of-the-art algorithms, while resulting in a better relative pose accuracy. Source code is available at https://github.com/jizhaox/relpose-mcs-depth.



Paperid:1348
Authors:Keqiang Li; Mingyang Zhao; Huaiyu Wu; Dong-Ming Yan; Zhen Shen; Fei-Yue Wang; Gang Xiong
Title: GraphFit: Learning Multi-Scale Graph-Convolutional Representation for Point Cloud Normal Estimation
Abstract:
We propose a precise and efficient normal estimation method that can deal with noise and nonuniform density for unstructured 3D point clouds. Unlike existing approaches that directly take patches and ignore the local neighborhood relationships, which make them susceptible to challenging regions such as sharp edges, we propose to learn graph convolutional feature representation for normal estimation, which emphasizes more local neighborhood geometry and effectively encodes intrinsic relationships. Additionally, we design a novel adaptive module based on the attention mechanism to integrate point features with their neighboring features, hence further enhancing the robustness of the proposed normal estimator against point density variations. To make it more distinguishable, we introduce a multi-scale architecture in the graph block to learn richer geometric features. Our method outperforms competitors with the state-of-the-art accuracy on various benchmark datasets, and is quite robust against noise, outliers, as well as the density variations.



Paperid:1349
Authors:Likang Wang; Yue Gong; Xinjun Ma; Qirui Wang; Kaixuan Zhou; Lei Chen
Title: IS-MVSNet: Importance Sampling-Based MVSNet
Abstract:
This paper presents a novel coarse-to-fine multi-view stereo (MVS) algorithm called importance-sampling-based MVSNet (IS-MVSNet) to address a crucial problem of limited depth resolution adopted by current learning-based MVS methods. We proposed an importance-sampling module for sampling candidate depth, effectively achieving higher depth resolution and yielding better point-cloud results while introducing no additional cost. Furthermore, we proposed an unsupervised error distribution estimation method for adjusting the density variation of the importance-sampling module. Notably, the proposed sampling module does not require any additional training and works reasonably well with the pre-trained weights of the baseline model. Our proposed method leads to up to 20x promotion on the most refined depth resolution, thus significantly benefiting most scenarios and excellently superior on fine details. As a result, IS-MVSNet outperforms all the published papers on TNT’s intermediate benchmark with an F-score of 62.82%. Code is available at github.com/NoOneUST/IS-MVSNet.



Paperid:1350
Authors:Jiaxiang Tang; Xiaokang Chen; Jingbo Wang; Gang Zeng
Title: Point Scene Understanding via Disentangled Instance Mesh Reconstruction
Abstract:
Semantic scene reconstruction from point cloud is an essential and challenging task for 3D scene understanding. This task requires not only to recognize each instance in the scene, but also to recover their geometries based on the partial observed point cloud. Existing methods usually attempt to directly predict occupancy values of the complete object based on incomplete point cloud proposals from a detection-based backbone. However, this framework always fails to reconstruct high fidelity mesh due to the obstruction of various detected false positive object proposals and the ambiguity of incomplete point observations for learning occupancy values of complete objects. To circumvent the hurdle, we propose a Disentangled Instance Mesh Reconstruction (DIMR) framework for effective point scene understanding. A segmentation-based backbone is applied to reduce false positive object proposals, which further benefits our exploration on the relationship between recognition and reconstruction. Based on the accurate proposals, we leverage a mesh-aware latent code space to disentangle the processes of shape completion and mesh generation, relieving the ambiguity caused by the incomplete point observations. Furthermore, with access to the CAD model pool at test time, our model can also be used to improve the reconstruction quality by performing mesh retrieval without extra training. We thoroughly evaluate the reconstructed mesh quality with multiple metrics, and demonstrate the superiority of our method on the challenging ScanNet dataset. Code is available at https://github.com/ashawkey/dimr.



Paperid:1351
Authors:Ruizhi Shao; Zerong Zheng; Hongwen Zhang; Jingxiang Sun; Yebin Liu
Title: DiffuStereo: High Quality Human Reconstruction via Diffusion-Based Stereo Using Sparse Cameras
Abstract:
We propose DiffuStereo, a novel system using only sparse cameras (8 in this work) for high-quality 3D human reconstruction. At its core is a novel diffusion-based stereo module, which introduces diffusion models, a type of powerful generative models, into the iterative stereo matching network. To this end, we design a new diffusion kernel and additional stereo constraints to facilitate stereo matching and depth estimation in the network. We further present a multi-level stereo network architecture to handle high-resolution (up to 4k) inputs without requiring unaffordable memory footprint. Given a set of sparse-view color images of a human, the proposed multi-level diffusion-based stereo network can produce highly accurate depth maps, which are then converted into a high-quality 3D human model through an efficient multi-view fusion strategy. Overall, our method enables automatic reconstruction of human models with quality on par to high-end dense-view camera rigs, and this is achieved using a much more light-weight hardware setup. Experiments show that our method outperforms state-of-the-art methods by a large margin both qualitatively and quantitatively.



Paperid:1352
Authors:Daniel Barath; Gábor Valasek
Title: Space-Partitioning RANSAC
Abstract:
A new algorithm is proposed to accelerate the RANSAC model quality calculations. The method is based on partitioning the joint correspondence space, e.g., 2D-2D point correspondences, into a pair of regular grids. The grid cells are mapped by minimal sample models, estimated within RANSAC, to reject correspondences that are inconsistent with the model parameters early. The proposed technique is general. It works with arbitrary transformations even if a point is mapped to a point set, e.g., as a fundamental matrix maps to epipolar lines. The method is tested on thousands of image pairs from publicly available datasets on fundamental and essential matrix, homography and radially distorted homography estimation. On average, the proposed space partitioning algorithm reduces the RANSAC run-time by 41% with provably no deterioration in the accuracy. When combined with SPRT, the run-time drops to its 30%.It can be straightforwardly plugged into any state-of-the-art RANSAC framework.



Paperid:1353
Authors:Mohamed Sayed; John Gibson; Jamie Watson; Victor Prisacariu; Michael Firman; Clément Godard
Title: SimpleRecon: 3D Reconstruction without 3D Convolutions
Abstract:
Traditionally, 3D indoor scene reconstruction from posed images happens in two phases: per-image depth estimation, followed by depth merging and surface reconstruction. Recently, a family of methods have emerged that perform reconstruction directly in final 3D volumetric feature space. While these methods have shown impressive reconstruction results, they rely on expensive 3D convolutional layers, limiting their application in resource-constrained environments. In this work, we instead go back to the traditional route, and show how focusing on high quality multi-view depth prediction leads to highly accurate 3D reconstructions using simple off-the-shelf depth fusion. We propose a simple state-of-the-art multi-view depth estimator with two main contributions: 1) a carefully-designed 2D CNN which utilizes strong image priors alongside a plane-sweep feature volume and geometric losses, combined with 2) the integration of keyframe and geometric metadata into the cost volume which allows informed depth plane scoring. Our method achieves a significant lead over the current state-of-the-art for depth estimation and close or better for 3D reconstruction on ScanNet and 7-Scenes, yet still allows for online real-time low-memory reconstruction. Code, models and results are available at https://nianticlabs.github.io/simplerecon.



Paperid:1354
Authors:Zhoutong Zhang; Forrester Cole; Zhengqi Li; Noah Snavely; Michael Rubinstein; William T. Freeman
Title: Structure and Motion from Casual Videos
Abstract:
Casual videos, such as those captured in daily life using a hand-held cell phone, pose problems for conventional structure-from-motion (SfM) techniques: the camera is often roughly stationary (not much parallax), and a large portion of the video may contain moving objects. Under such conditions, state-of-the-art SfM methods tend to produce erroneous results, often failing entirely. To address these issues, we propose CasualSAM, a method to estimate camera poses and dense depth maps from a monocular, casually-captured video. Like conventional SfM, our method performs a joint optimization over 3D structure and camera poses, but uses a pretrained depth prediction network to represent 3D structure rather than sparse keypoints. In contrast to previous approaches, our method does not assume motion is rigid or determined by semantic segmentation, instead optimizing for a per-pixel motion map based on reprojection error. Our method sets a new state-of-the-art for pose and depth estimation on the Sintel dataset, and produces high-quality results for the DAVIS dataset where most prior methods fail to produce usable camera poses.



Paperid:1355
Authors:Guangming Wang; Yunzhe Hu; Zhe Liu; Yiyang Zhou; Masayoshi Tomizuka; Wei Zhan; Hesheng Wang
Title: What Matters for 3D Scene Flow Network
Abstract:
3D scene flow estimation from point clouds is a low-level 3D motion perception task in computer vision. Flow embedding is a commonly used technique in scene flow estimation, and it encodes the point motion between two consecutive frames. Thus, it is critical for the flow embeddings to capture the correct overall direction of the motion. However, previous works only search locally to determine a soft correspondence, ignoring the distant points that turn out to be the actual matching ones. In addition, the estimated correspondence is usually from the forward direction of the adjacent point clouds, and may not be consistent with the estimated correspondence acquired from the backward direction. To tackle these problems, we propose a novel all-to-all flow embedding layer with backward reliability validation during the initial scene flow estimation. Besides, we investigate and compare several design choices in key components of the 3D scene flow network, including the point similarity calculation, input elements of predictor, and predictor & refinement level design. After carefully choosing the most effective designs, we are able to present a model that achieves the state-of-the-art performance on FlyingThings3D and KITTI Scene Flow datasets. Our proposed model surpasses all existing methods by at least 38.2% on FlyingThings3D dataset and 24.7% on KITTI Scene Flow dataset for EPE3D metric. We release our codes at https://github.com/IRMVLab/3DFlow.



Paperid:1356
Authors:Lalit Manam; Venu Madhav Govindu
Title: Correspondence Reweighted Translation Averaging
Abstract:
Translation averaging methods use the consistency of input translation directions to solve for camera translations. However, translation directions obtained using epipolar geometry are error-prone. This paper argues that the improved accuracy of translation averaging should be leveraged to mitigate the errors in the input translation direction estimates. To this end, we introduce weights for individual correspondences which are iteratively refined to yield improved translation directions. In turn, these refined translation directions are averaged to obtain camera translations. This results in an alternating approach to translation averaging. The modularity of our framework allows us to use existing translation averaging methods and improve their results. The efficacy of the scheme is demonstrated by comparing performance with state-of-the-art methods on a number of real-world datasets. We also show that our approach yields reasonably good 3D reconstructions with straightforward triangulation, i.e. without any bundle adjustment iterations.



Paperid:1357
Authors:Radu Alexandru Rosu; Shunsuke Saito; Ziyan Wang; Chenglei Wu; Sven Behnke; Giljoo Nam
Title: Neural Strands: Learning Hair Geometry and Appearance from Multi-View Images
Abstract:
We present Neural Strands, a novel learning framework for modeling accurate hair geometry and appearance from multi-view image inputs. The learned hair model can be rendered in real-time from any viewpoint with high-fidelity view-dependent effects. Our model achieves intuitive shape and style control unlike volumetric counterparts. To enable these properties, we propose a novel hair representation based on a neural scalp texture that encodes the geometry and appearance of individual strands at each texel location. Furthermore, we introduce a novel neural rendering framework based on rasterization of the learned hair strands. Our neural rendering is strand-accurate and anti-aliased, making the rendering view-consistent and photorealistic. Combining appearance with a multi-view geometric prior, we enable, for the first time, the joint learning of appearance and explicit hair geometry from a multi-view setup. We demonstrate the efficacy of our approach in terms of fidelity and efficiency for various hairstyles.



Paperid:1358
Authors:Xin Liu; Xiaofei Shao; Bo Wang; Yali Li; Shengjin Wang
Title: GraphCSPN: Geometry-Aware Depth Completion via Dynamic GCNs
Abstract:
Image guided depth completion aims to recover per-pixel dense depth maps from sparse depth measurements with the help of aligned color images, which has a wide range of applications from robotics to autonomous driving. However, the 3D nature of sparse-to-dense depth completion has not been fully explored by previous methods. In this work, we propose a Graph Convolution based Spatial Propagation Network (GraphCSPN) as a general approach for depth completion. First, unlike previous methods, we leverage convolution neural networks as well as graph neural networks in a complementary way for geometric representation learning. In addition, the proposed networks explicitly incorporate learnable geometric constraints to regularize the propagation process performed in three-dimensional space rather than in two-dimensional plane. Furthermore, we construct the graph utilizing sequences of feature patches, and update it dynamically with an edge attention module during propagation, so as to better capture both the local neighboring features and global relationships over long distance. Extensive experiments on both indoor NYU-Depth-v2 and outdoor KITTI datasets demonstrate that our method achieves the state-of-the-art performance, especially when compared in the case of using only a few propagation steps. Code and models are available at the project page.



Paperid:1359
Authors:Aikaterini Adam; Torsten Sattler; Konstantinos Karantzalos; Tomas Pajdla
Title: Objects Can Move: 3D Change Detection by Geometric Transformation Consistency
Abstract:
AR/VR applications and robots need to know when the scene has changed. An example is when objects are moved, added, or removed from the scene. We propose a 3D object discovery method that is based only on scene changes. Our method does not need to encode any assumptions about what is an object, but rather discovers objects by exploiting their coherent move. Changes are initially detected as differences in the depth maps and segmented as objects if they undergo rigid motions. A graph cut optimization propagates the changing labels to geometrically consistent regions. Experiments show that our method achieves state-of-the-art performance on the 3RScan dataset against competitive baselines. The source code of our method can be found at https://github.com/katadam/ObjectsCanMove.



Paperid:1360
Authors:Dávid Rozenberszki; Or Litany; Angela Dai
Title: Language-Grounded Indoor 3D Semantic Segmentation in the Wild
Abstract:
Recent advances in 3D semantic segmentation with deep neural networks have shown remarkable success, with rapid performance increase on available datasets. However, current 3D semantic segmentation benchmarks contain only a small number of categories -- less than 30 for ScanNet and SemanticKITTI, for instance, which are not enough to reflect the diversity of real environments (e.g., semantic image understanding covers hundreds to thousands of classes). Thus, we propose to study a larger vocabulary for 3D semantic segmentation with a new extended benchmark on ScanNet data with 200 class categories, an order of magnitude more than previously studied. This large number of class categories also induces a large natural class imbalance, both of which are challenging for existing 3D semantic segmentation methods. To learn more robust 3D features in this context, we propose a language-driven pre-training method to encourage learned 3D features that might have limited training examples to lie close to their pre-trained text embeddings. Extensive experiments show that our approach consistently outperforms state-of-the-art 3D pre-training for 3D semantic segmentation on our proposed benchmark (+9% relative mIoU), including limited-data scenarios with +25% relative mIoU using only 5% annotations.



Paperid:1361
Authors:Sameera Ramasinghe; Simon Lucey
Title: Beyond Periodicity: Towards a Unifying Framework for Activations in Coordinate-MLPs
Abstract:
Coordinate-MLPs are emerging as an effective tool for modeling multidimensional continuous signals, overcoming many drawbacks associated with discrete grid-based approximations. However, coordinate-MLPs with ReLU activations, in their rudimentary form, demonstrate poor performance in representing signals with high fidelity, promoting the need for positional embedding layers. Recently, Sitzmann et al. proposed a sinusoidal activation function that has the capacity to omit positional embedding from coordinate-MLPs while still preserving high signal fidelity. Despite its potential, ReLUs are still dominating the space of coordinate-MLPs; we speculate that this is due to the hyper-sensitivity of networks -- that employ such sinusoidal activations -- to the initialization schemes. In this paper, we attempt to broaden the current understanding of the effect of activations in coordinate-MLPs, and show that there exists a broader class of activations that are suitable for encoding signals. We affirm that sinusoidal activations are only a single example in this class, and propose several non-periodic functions that empirically demonstrate more robust performance against random initializations than sinusoids. Finally, we advocate for a shift towards coordinate-MLPs that employ these non-traditional activation functions due to their high performance and simplicity.



Paperid:1362
Authors:Tianhan Xu; Tatsuya Harada
Title: Deforming Radiance Fields with Cages
Abstract:
Recent advances in radiance fields enable photorealistic rendering of static or dynamic 3D scenes, but still do not support explicit deformation that is used for scene manipulation or animation. In this paper, we propose a method that enables a new type of deformation of the radiance field: free-form radiance field deformation. We use a triangular mesh that encloses the foreground object called cage as an interface, and by manipulating the cage vertices, our approach enables the free-form deformation of the radiance field. The core of our approach is cage-based deformation which is commonly used in mesh deformation. We propose a novel formulation to extend it to the radiance field, which maps the position and the view direction of the sampling points from the deformed space to the canonical space, thus enabling the rendering of the deformed scene. The deformation results of the synthetic datasets and the real-world datasets demonstrate the effectiveness of our approach.



Paperid:1363
Authors:Brian Gordon; Sigal Raab; Guy Azov; Raja Giryes; Daniel Cohen-Or
Title: FLEX: Extrinsic Parameters-Free Multi-View 3D Human Motion Reconstruction
Abstract:
The increasing availability of video recordings made by multiple cameras has offered new means for mitigating occlusion and depth ambiguities in pose and motion reconstruction methods. Yet, multi-view algorithms strongly depend on camera parameters, particularly on relative transformations between the cameras. Such a dependency becomes a hurdle once shifting to dynamic capture in uncontrolled settings. We introduce FLEX (Free muLti-view rEconstruXion), an end-to-end extrinsic parameter-free multi-view model. FLEX is extrinsic parameter-free (dubbed ep-free) in the sense that it does not require extrinsic camera parameters. Our key idea is that the 3D angles between skeletal parts, as well as bone lengths, are invariant to the camera position. Hence, learning 3D rotations and bone lengths rather than locations allows for predicting common values for all camera views. Our network takes multiple video streams, learns fused deep features through a novel multi-view fusion layer, and reconstructs a single consistent skeleton with temporally coherent joint rotations. We demonstrate quantitative and qualitative results on three public data sets, and on multi-person synthetic video streams captured by dynamic cameras. We compare our model to state-of-the-art methods that are not ep-free and show that in the absence of camera parameters, we outperform them by a large margin while obtaining comparable results when camera parameters are available. Code, trained models, and other materials are available on https://briang13.github.io/FLEX.



Paperid:1364
Authors:Ming Li; Xueqian Jin; Xuejiao Hu; Jingzhao Dai; Sidan Du; Yang Li
Title: MODE: Multi-View Omnidirectional Depth Estimation with 360° Cameras
Abstract:
In this paper, we propose a two-stage omnidirectional depth estimation framework with multi-view 360-degree cameras. The framework first estimates the depth maps from different camera pairs via omnidirectional stereo matching and then fuses the depth maps to achieve robustness against mud spots, water drops on camera lenses, and glare caused by intense light. We adopt spherical feature learning to address the distortion of panoramas. In addition, a synthetic 360-degree dataset consisting of 12K road scene panoramas and 3K ground truth depth maps is presented to train and evaluate 360-degree depth estimation algorithms. Our dataset takes soiled camera lenses and glare into consideration, which is more consistent with the real-world environment. Experimental results show that the proposed framework generates reliable results in both synthetic and real-world environments, and it achieves state-of-the-art performance on different datasets. The code and data are available at https://github.com/nju-ee/MODE-2022"



Paperid:1365
Authors:Simon Schreiberhuber; Jean-Baptiste Weibel; Timothy Patten; Markus Vincze
Title: GigaDepth: Learning Depth from Structured Light with Branching Neural Networks
Abstract:
Structured light-based depth sensors provide accurate depth information independently of the scene appearance by extracting pattern positions from the captured pixel intensities. Spatial neighborhood encoding, in particular, is a popular structured light approach for off-the-shelf hardware. However, it suffers from the distortion and fragmentation of the projected pattern by the scene’s geometry in the vicinity of a pixel. This forces algorithms to find a delicate balance between depth prediction accuracy and robustness to pattern fragmentation or appearance change. While stereo matching provides more robustness at the expense of accuracy, we show that learning to regress a pixel’s position within the projected pattern is not only more accurate when combined with classification but can be made equally robust. We propose to split the regression problem into smaller classification sub-problems in a coarse-to-fine manner with the use of a weight-adaptive layer that efficiently implements branching per-pixel Multilayer Perceptrons applied to features extracted by a Convolutional Neural Network. As our approach requires full supervision, we train our algorithm on a rendered dataset sufficiently close to the real-world domain. On a separately captured real-world dataset, we show that our network outperforms state-of-the-art and is significantly more robust than other regression-based approaches.



Paperid:1366
Authors:Xuran Pan; Zihang Lai; Shiji Song; Gao Huang
Title: ActiveNeRF: Learning Where to See with Uncertainty Estimation
Abstract:
Recently, Neural Radiance Fields (NeRF) has shown promising performances on reconstructing 3D scenes and synthesizing novel views from a sparse set of 2D images. Albeit effective, the performance of NeRF is highly influenced by the quality of training samples. With limited posed images from the scene, NeRF fails to generalize well to novel views and may collapse to trivial solutions in unobserved regions. This makes NeRF impractical under resource-constrained scenarios. In this paper, we present a novel learning framework, \textit{ActiveNeRF}, aiming to model a 3D scene with a constrained input budget. Specifically, we first incorporate uncertainty estimation into a NeRF model, which ensures robustness under few observations and provides an interpretation of how NeRF understands the scene. On this basis, we propose to supplement the existing training set with newly captured samples based on an active learning scheme. By evaluating the reduction of uncertainty given new inputs, we select the samples that bring the most information gain. In this way, the quality of novel view synthesis can be improved with minimal additional resources. Extensive experiments validate the performance of our model on both realistic and synthetic scenes, especially with scarcer training data.



Paperid:1367
Authors:Matteo Taiana; Matteo Toso; Stuart James; Alessio Del Bue
Title: PoserNet: Refining Relative Camera Poses Exploiting Object Detections
Abstract:
The estimation of the camera poses associated with a set of images commonly relies on feature matches between the images. In contrast, we are the first to address this challenge by using objectness regions to guide the pose estimation problem rather than explicit semantic object detections. We propose Pose Refiner Network (PoserNet) a light-weight Graph Neural Network to refine the approximate pair-wise relative camera poses. PoserNet exploits associations between the objectness regions - concisely expressed as bounding boxes - across multiple views to globally refine sparsely connected view graphs. We evaluate on the 7-Scenes dataset across varied sizes of graphs and show how this process can be beneficial to optimization-based motion averaging algorithms improving the median error on the rotation by 62 degrees with respect to the initial estimates obtained based on bounding boxes. Code and data are available at https://github.com/IIT-PAVIS/PoserNet.



Paperid:1368
Authors:Shin-Fang Chng; Sameera Ramasinghe; Jamie Sherrah; Simon Lucey
Title: Gaussian Activated Neural Radiance Fields for High Fidelity Reconstruction \& Pose Estimation
Abstract:
Despite Neural Radiance Fields (NeRF) showing compelling results in photorealistic novel views synthesis of real-world scenes, most existing approaches require accurate prior camera poses. Although approaches for jointly recovering the radiance field and camera pose exist, they rely on a cumbersome coarse-to-fine auxiliary positional embedding to ensure good performance. We present Gaussian Activated Neural Radiance Fields (GARF), a new positional embedding-free neural radiance field architecture -- employing Gaussian activations -- that is competitive with the current state-of-the-art in terms of high fidelity reconstruction and pose estimation.



Paperid:1369
Authors:Jan U. Müller; Michael Weinmann; Reinhard Klein
Title: Unbiased Gradient Estimation for Differentiable Surface Splatting via Poisson Sampling
Abstract:
We propose an efficient and GPU-accelerated sampling framework which enables unbiased gradient approximation for differentiable point cloud rendering based on surface splatting. Our framework models the contribution of a point to the rendered image as a probability distribution. We derive an unbiased approximative gradient for the rendering function within this model. To efficiently evaluate the proposed sample estimate, we introduce a tree-based data-structure which employs multi-pole methods to draw samples in near linear time. Our gradient estimator allows us to avoid regularization required by previous methods, leading to a more faithful shape recovery from images. Furthermore, we validate that these improvements are applicable to real-world applications by refining the camera poses and point cloud obtained from a real-time SLAM system. Finally, employing our framework in a neural rendering setting optimizes both the point cloud and network parameters, highlighting the framework’s ability to enhance data driven approaches.



Paperid:1370
Authors:Kushagra Tiwary; Tzofi Klinghoffer; Ramesh Raskar
Title: Towards Learning Neural Representations from Shadows
Abstract:
We present a method that learns neural shadow fields, which are neural scene representations that are only learnt from the shadows present in the scene. While traditional shape-from-shadow (SfS) algorithms reconstruct geometry from shadows, they assume a fixed scanning setup and fail to generalize to complex scenes. Neural rendering algorithms, on the other hand, rely on photometric consistency between RGB images, but largely ignore physical cues such as shadows, which have been shown to provide valuable information about the scene. We observe that shadows are a powerful cue that can constrain neural scene representations to learn SfS, and even outperform NeRF to reconstruct otherwise hidden geometry. We propose a graphics-inspired differentiable approach to render accurate shadows with volumetric rendering, predicting a shadow map that can be compared to the ground truth shadow. Even with just binary shadow maps, we show that neural rendering can localize the object and estimate coarse geometry. Our approach reveals that sparse cues in images can be used to estimate geometry using differentiable volumetric rendering. Moreover, our framework is highly generalizable and can work alongside existing 3D reconstruction techniques that otherwise only use photometric consistency.



Paperid:1371
Authors:Subhankar Roy; Mingxuan Liu; Zhun Zhong; Nicu Sebe; Elisa Ricci
Title: Class-Incremental Novel Class Discovery
Abstract:
We study the new task of class-incremental Novel Class Discovery (class-iNCD), which refers to the problem of discovering novel categories in an unlabelled data set by leveraging a pre-trained model that has been trained on a labelled data set containing disjoint yet related categories. Apart from discovering novel classes, we also aim at preserving the ability of the model to recognize previously seen base categories. Inspired by rehearsal-based incremental learning methods, in this paper we propose a novel approach for class-iNCD which prevents forgetting of past information about the base classes by jointly exploiting base class feature prototypes and feature-level knowledge distillation. We also propose a self-training clustering strategy that simultaneously clusters novel categories and trains a joint classifier for both the base and novel classes. This makes our method able to operate in a class-incremental setting. Our experiments, conducted on three common benchmarks, demonstrate that our method significantly outperforms state-of-the-art approaches. Code is available at \url{https://github.com/OatmealLiu/class-iNCD}.



Paperid:1372
Authors:Jie Liu; Xiaoqing Guo; Yixuan Yuan
Title: Unknown-Oriented Learning for Open Set Domain Adaptation
Abstract:
Open set domain adaptation (OSDA) aims to tackle the distribution shift of partially shared categories between the source and target domains, meanwhile identifying target samples non-appeared in source domain. The key issue behind this problem is to classify these various unseen samples as unknown category with the absent of relevant knowledge from the source domain. Though impressing performance, existing works neglect the complex semantic information and huge intra-category variation of unknown category, incapable of representing the complicated distribution. To overcome this, we propose a novel Unknown-Oriented Learning (UOL) framework for OSDA, and it is composed of three stages: true unknown excavation, false unknown suppression and known alignment. Specifically, to excavate the diverse semantic information in unknown category, the multi-unknown detector (MUD) equipped with weight discrepancy constraint is proposed in true unknown excavation. During false unknown suppression, Source-to-Target grAdient Graph (S2TAG) is constructed to select reliable target samples with the proposed super confidence criteria. Then, Target-to-Target grAdient Graph (T2TAG) exploits the geometric structure in gradient manifold to obtain confident pseudo labels for target data. At the last stage, known alignment, the known samples in the target domain are aligned with the source domain to alleviate the domain gap. Extensive experiments demonstrate the superiority of our method compared with state-of-the-art methods on three benchmarks.



Paperid:1373
Authors:Hongbin Lin; Yifan Zhang; Zhen Qiu; Shuaicheng Niu; Chuang Gan; Yanxia Liu; Mingkui Tan
Title: Prototype-Guided Continual Adaptation for Class-Incremental Unsupervised Domain Adaptation
Abstract:
This paper studies a new, practical but challenging problem, called Class-Incremental Unsupervised Domain Adaptation (CI-UDA), where the labeled source domain contains all classes, but the classes in the unlabeled target domain increase sequentially. This problem is challenging due to two difficulties. First, source and target label sets are inconsistent at each time step, which makes it difficult to conduct accurate domain alignment. Second, previous target classes are unavailable in the current step, resulting in the forgetting of previous knowledge. To address this problem, we propose a novel Prototype-guided Continual Adaptation (ProCA) method, consisting of two solution strategies. 1) Label prototype identification: we identify target label prototypes by detecting shared classes with cumulative prediction probabilities of target samples. 2) Prototype-based alignment and replay: based on the identified label prototypes, we align both domains and enforce the model to retain previous knowledge. With these two strategies, ProCA is able to adapt the source model to a class-incremental unlabeled target domain effectively. Extensive experiments demonstrate the effectiveness and superiority of ProCA in resolving CI-UDA. The source code is available at https://github.com/Hongbin98/ProCA.git.



Paperid:1374
Authors:Xin Lai; Zhuotao Tian; Xiaogang Xu; Yingcong Chen; Shu Liu; Hengshuang Zhao; Liwei Wang; Jiaya Jia
Title: DecoupleNet: Decoupled Network for Domain Adaptive Semantic Segmentation
Abstract:
Unsupervised domain adaptation in semantic segmentation alleviates the reliance on expensive pixel-wise annotation. It uses a labeled source domain dataset as well as unlabeled target domain images to learn a segmentation network. In this paper, we observe two main issues of existing domain-invariant learning framework. (1) Being distracted by the feature distribution alignment, the network cannot focus on the segmentation task. (2) Fitting source domain data well would compromise the target domain performance. To address these issues, we propose DecoupleNet to alleviate source domain overfitting and let the final model focus more on the segmentation task. Also, we put forward Self-Discrimination (SD) and introduce an auxiliary classifier to learn more discriminative target domain features with pseudo labels. Finally, we propose Online Enhanced Self-Training (OEST) to contextually enhance the quality of pseudo labels in an online manner. Experiments show our method outperforms existing state-of-the-art methods. Extensive ablation studies verify the effectiveness of each component. Code is available at https://github.com/dvlab-research/DecoupleNet.



Paperid:1375
Authors:Shenjian Gong; Shanshan Zhang; Jian Yang; Dengxin Dai; Bernt Schiele
Title: Class-Agnostic Object Counting Robust to Intraclass Diversity
Abstract:
Most previous works on object counting are limited to pre-defined categories. In this paper, we focus on classagnostic counting, i.e., counting object instances in an image by simply specifying a few exemplar boxes of interest. We start with an analysis on intraclass diversity and point out three factors: color, shape and scale diversity seriously hurts counting performance. Motivated by this analysis, we propose a new counter robust to high intraclass diversity, for which we propose two effective modules: Exemplar Feature Augmentation (EFA) and Edge Matching (EM). Aiming to handle diversity from all aspects, EFA generates a large variety of exemplars in the feature space based on the provided exemplars. Additionally, the edge matching branch focuses on the more reliable cue of shape, making our counter more robust to color variations. Experimental results on standard benchmarks show that our Robust Class-Agnostic Counter (RCAC) achieves state-of-the-art performance. The code is publicly available at https://github.com/Yankeegsj/RCAC.



Paperid:1376
Authors:Luyu Yang; Mingfei Gao; Zeyuan Chen; Ran Xu; Abhinav Shrivastava; Chetan Ramaiah
Title: Burn after Reading: Online Adaptation for Cross-Domain Streaming Data
Abstract:
In the context of online privacy, many methods propose complex security preserving measures to protect sensitive data. In this paper, we note that: not storing any sensitive data is the best form of security. We propose an online framework called ""Burn After Reading"", i.e. each online sample is permanently deleted after it is processed. Our framework utilizes the labels from the public data and predicts on the unlabeled sensitive private data. To tackle the inevitable distribution shift from the public data to the private data, we propose a novel domain adaptation algorithm that directly aims at the fundamental challenge of this online setting--the lack of diverse source-target data pairs. We design a Cross-Domain Bootstrapping approach, named CroDoBo, to increase the combined data diversity across domains. To fully exploit the valuable discrepancies among the diverse combinations, we employ the training strategy of multiple learners with co-supervision. CroDoBo achieves state-of-the-art online performance on four domain adaptation benchmarks. Code is provided.



Paperid:1377
Authors:Guodong Xu; Yuenan Hou; Ziwei Liu; Chen Change Loy
Title: Mind the Gap in Distilling StyleGANs
Abstract:
StyleGAN family is one of the most popular Generative adversarial networks (GANs) for unconditional generation. Despite its impressive performance, its high demand on storage and computation impedes their deployment on resource-constrained devices. This paper provides a comprehensive study of distilling from the popular StyleGAN-like architecture. Our key insight is that the main challenge of StyleGAN distillation lies in the output discrepancy issue, where the teacher and student model yield different outputs given the same input latent code. Standard knowledge distillation losses typically fail under this heterogeneous distillation scenario. We conduct thorough analysis about the reasons and effects of this discrepancy issue, and identify that the style module plays a vital role in determining semantic information of generated images. Based on this finding, we propose a novel initialization strategy for the student model, which can ensure the output consistency to the maximum extent. To further enhance the semantic consistency between the teacher and student model, we present a latent-direction-based distillation loss that preserves the semantic relations in latent space. Extensive experiments demonstrate the effectiveness of our approach in distilling StyleGAN2 and StyleGAN3, outperforming existing GAN distillation methods by a large margin. Code and models will be released.



Paperid:1378
Authors:Sungha Choi; Seunghan Yang; Seokeon Choi; Sungrack Yun
Title: Improving Test-Time Adaptation via Shift-Agnostic Weight Regularization and Nearest Source Prototypes
Abstract:
This paper proposes a novel test-time adaptation strategy that adjusts the model pre-trained on the source domain using only unlabeled online data from the target domain to alleviate the performance degradation due to the distribution shift between the source and target domains. Adapting the entire model parameters using the unlabeled online data may be detrimental due to the erroneous signals from an unsupervised objective. To mitigate this problem, we propose a shift-agnostic weight regularization that encourages largely updating the model parameters sensitive to distribution shift while slightly updating those insensitive to the shift, during test-time adaptation. This regularization enables the model to quickly adapt to the target domain without performance degradation by utilizing the benefit of a high learning rate. In addition, we present an auxiliary task based on nearest source prototypes to align the source and target features, which helps reduce the distribution shift and leads to further performance improvement. We show that our method exhibits state-of-the-art performance on various standard benchmarks and even outperforms its supervised counterpart.



Paperid:1379
Authors:Yuliang Zou; Zizhao Zhang; Chun-Liang Li; Han Zhang; Tomas Pfister; Jia-Bin Huang
Title: Learning Instance-Specific Adaptation for Cross-Domain Segmentation
Abstract:
We propose a test-time adaptation method for cross-domain image segmentation. Our method is simple: Given a new unseen instance at the test time, we adapt a pre-trained model by conducting instance-specific BatchNorm (statistics) calibration. Our approach has two core components. First, we replace the manually designed BatchNorm calibration rule with a learnable module. Second, we leverage strong data augmentation to simulate random domain shifts for learning the calibration rule. In contrast to existing domain adaptation methods, our method does not require accessing the target domain data at training time or conducting computationally expensive test-time model training/optimization. Equipping our method with models trained by standard recipes achieves significant improvement, comparing favorably with several state-of-the-art domain generalization and one-shot unsupervised domain adaptation approaches. Combining our method with the domain generalization methods further improves the performance, reaching a new state of the art.



Paperid:1380
Authors:Yufei Xu; Qiming Zhang; Jing Zhang; Dacheng Tao
Title: RegionCL: Exploring Contrastive Region Pairs for Self-Supervised Representation Learning
Abstract:
Self-supervised methods (SSL) have achieved significant success via maximizing the mutual information between two augmented views, where cropping is a popular augmentation technique. Cropped regions are widely used to construct positive pairs, while the remained regions after cropping have rarely been explored in existing methods, although they together constitute the same image instance and both contribute to the description of the category. In this paper, we make the first attempt to demonstrate the importance of both regions in cropping from a complete perspective and the effectiveness of using both regions via designing a simple yet effective pretext task called Region Contrastive Learning (RegionCL). Technically, to construct the two kinds of regions, we randomly crop a region (called the paste view) from each input image with the same size and swap them between different images to compose new images together with the remained regions (called the canvas view). Then, instead of taking the new images as a whole for positive or negative samples, contrastive pairs are efficiently constructed from the regional perceptive based on the following simple criteria, i.e., each view is (1) positive with views augmented from the same original image and (2) negative with views augmented from other images. With minor modifications to popular SSL methods, RegionCL exploits those abundant pairs and helps the model distinguish the regions features from both canvas and paste views, therefore learning better visual representations. Experiments on ImageNet, MS COCO, and Cityscapes demonstrate that RegionCL improves MoCov2, DenseCL, and SimSiam by large margins and achieves state-of-the-art performance on classification, detection, and segmentation tasks.



Paperid:1381
Authors:Xialei Liu; Yu-Song Hu; Xu-Sheng Cao; Andrew D. Bagdanov; Ke Li; Ming-Ming Cheng
Title: Long-Tailed Class Incremental Learning
Abstract:
In class incremental learning (CIL) a model must learn new classes in a sequential manner without forgetting old ones. However, conventional CIL methods consider a balanced distribution for each new task, which ignores the prevalence of long-tailed distributions in the real world. In this work we propose two long-tailed CIL scenarios, which we term Ordered and Shuffled LT-CIL. Ordered LT-CIL considers the scenario where we learn from head classes collected with more samples than tail classes which have few. Shuffled LT-CIL, on the other hand, assumes a completely random long-tailed distribution for each task. We systematically evaluate existing methods in both LT-CIL scenarios and demonstrate very different behaviors compared to conventional CIL scenarios. Additionally, we propose a two-stage learning baseline with a learnable weight scaling layer for reducing the bias caused by long-tailed distribution in LT-CIL and which in turn also improves the performance of conventional CIL due to the limited exemplars. Our results demonstrate the superior performance (up to 6.44 points in average incremental accuracy) of our approach on CIFAR-100 and ImageNet-Subset. The code is available at https://github.com/xialeiliu/Long-Tailed-CIL.



Paperid:1382
Authors:Hyounguk Shon; Janghyeon Lee; Seung Hwan Kim; Junmo Kim
Title: DLCFT: Deep Linear Continual Fine-Tuning for General Incremental Learning
Abstract:
Pre-trained representation is one of the key elements in the success of modern deep learning. However, existing works on continual learning methods have mostly focused on learning models incrementally from scratch. In this paper, we explore an alternative framework to incremental learning where we continually fine-tune the model from a pre-trained representation. Our method takes advantage of linearization technique of a pre-trained neural network for simple and effective continual learning. We show that this allows us to design a linear model where quadratic parameter regularization method is placed as the optimal continual learning policy, and at the same time enjoying the high performance of neural networks. We also show that the proposed algorithm enables parameter regularization methods to be applied to class-incremental problems. Additionally, we provide a theoretical reason why the existing parameter-space regularization algorithms such as EWC underperform on neural networks trained with cross-entropy loss. We show that the proposed method can prevent forgetting while achieving high continual fine-tuning performance on image classification tasks. To show that our method can be applied to general continual learning settings, we evaluate our method in data-incremental, task-incremental, and class-incremental learning problems.



Paperid:1383
Authors:Kun-Yu Lin; Jiaming Zhou; Yukun Qiu; Wei-Shi Zheng
Title: Adversarial Partial Domain Adaptation by Cycle Inconsistency
Abstract:
Unsupervised partial domain adaptation (PDA) is a unsupervised domain adaptation problem which assumes that the source label space subsumes the target label space. A critical challenge of PDA is the negative transfer problem, which is triggered by learning to match the whole source and target domains. To mitigate negative transfer, we note a fact that, it is impossible for a source sample of outlier classes to find a target sample of the same category due to the absence of outlier classes in the target domain, while it is possible for a source sample of shared classes. Inspired by this fact, we exploit the cycle inconsistency, i.e., category discrepancy between the original features and features after cycle transformations, to distinguish outlier classes apart from shared classes in the source domain. Accordingly, we propose to filter out source samples of outlier classes by weight suppression and align the distributions of shared classes between the source and target domains by adversarial learning. To learn accurate weight assignment for filtering out outlier classes, we design cycle transformations based on domain prototypes and soft nearest neighbor, where center losses are introduced in individual domains to reduce the intra-class variation. Experiment results on three benchmark datasets demonstrate the effectiveness of our proposed method.



Paperid:1384
Authors:Sehyun Hwang; Sohyun Lee; Sungyeon Kim; Jungseul Ok; Suha Kwak
Title: Combating Label Distribution Shift for Active Domain Adaptation
Abstract:
We consider the problem of active domain adaptation (ADA) to unlabeled target data, of which subset is actively selected and labeled given a budget constraint. Inspired by recent analysis on a critical issue from label distribution mismatch between source and target in domain adaptation, we devise a method that addresses the issue for the first time in ADA. At its heart lies a novel sampling strategy, which seeks target data that best approximate the entire target distribution as well as being representative, diverse, and uncertain. The sampled target data are then used not only for supervised learning but also for matching label distributions of source and target domains, leading to remarkable performance improvement. On four public benchmarks, our method substantially outperforms existing methods in every adaptation scenario.



Paperid:1385
Authors:Cristiano Saltori; Evgeny Krivosheev; Stéphane Lathuilière; Nicu Sebe; Fabio Galasso; Giuseppe Fiameni; Elisa Ricci; Fabio Poiesi
Title: GIPSO: Geometrically Informed Propagation for Online Adaptation in 3D LiDAR Segmentation
Abstract:
3D point cloud semantic segmentation is fundamental for autonomous driving. Most approaches in the literature neglect an important aspect, i.e., how to deal with domain shift when handling dynamic scenes. This can significantly hinder the navigation capabilities of self-driving vehicles. This paper advances the state of the art in this research field. Our first contribution consists in analysing a new unexplored scenario in point cloud segmentation, namely Source-Free Online Unsupervised Domain Adaptation (SF-OUDA). We experimentally show that state-of-the-art methods have a rather limited ability to adapt pre-trained deep network models to unseen domains in an online manner. Our second contribution is an approach that relies on adaptive self-training and geometric-feature propagation to adapt a pre-trained source model online without requiring either source data or target labels. Our third contribution is to study SF-OUDA in a challenging setup where source data is synthetic and target data is point clouds captured in the real world. We use the recent SynLiDAR dataset as a synthetic source and introduce two new synthetic (source) datasets, which can stimulate future synthetic-to-real autonomous driving research. Our experiments show the effectiveness of our segmentation approach on thousands of real-world point clouds.



Paperid:1386
Authors:Cristiano Saltori; Fabio Galasso; Giuseppe Fiameni; Nicu Sebe; Elisa Ricci; Fabio Poiesi
Title: CoSMix: Compositional Semantic Mix for Domain Adaptation in 3D LiDAR Segmentation
Abstract:
3D LiDAR semantic segmentation is fundamental for autonomous driving. Several Unsupervised Domain Adaptation (UDA) methods for point cloud data have been recently proposed to improve model generalization for different sensors and environments. Researchers working on UDA problems in the image domain have shown that sample mixing can mitigate domain shift. We propose a new approach of sample mixing for point cloud UDA, namely Compositional Semantic Mix (CoSMix), the first UDA approach for point cloud segmentation based on sample mixing. CoSMix consists of a two-branch symmetric network that can process labelled synthetic data (source) and real-world unlabelled point clouds (target) concurrently. Each branch operates on one domain by mixing selected pieces of data from the other one, and by using the semantic information derived from source labels and target pseudo-labels. We evaluate CoSMix on two large-scale datasets, showing that it outperforms state-of-the-art methods by a large margin.



Paperid:1387
Authors:Donghyun Kim; Kaihong Wang; Kate Saenko; Margrit Betke; Stan Sclaroff
Title: A Unified Framework for Domain Adaptive Pose Estimation
Abstract:
While pose estimation is an important computer vision task, it requires expensive annotation and suffers from domain shift. In this paper, we investigate the problem of domain adaptive 2D pose estimation that transfers knowledge learned on a synthetic source domain to a target domain without supervision. While several domain adaptive pose estimation models have been proposed recently, they are not generic but only focus on either human pose or animal pose estimation, and thus their effectiveness is somewhat limited to specific scenarios. In this work, we propose a unified framework that generalizes well on various domain adaptive pose estimation problems. We propose to align representations using both input-level and output-level cues (pixels and pose labels, respectively), which facilitates the knowledge transfer from the source domain to the unlabeled target domain. Our experiments show that our method achieves state-of-the-art performance under various domain shifts. Our method outperforms existing baselines on human pose estimation by up to 4.5 percent points (pp), hand pose estimation by up to 7.4 pp, and animal pose estimation by up to 4.8 pp for dogs and 3.3 pp for sheep. These results suggest that our method is able to mitigate domain shift on diverse tasks and even unseen domains and objects (e.g., trained on horse and tested on dog). Our code will be publicly available at: https://github.com/VisionLearningGroup/UDA_PoseEstimation.



Paperid:1388
Authors:Donghyun Kim; Kaihong Wang; Stan Sclaroff; Kate Saenko
Title: A Broad Study of Pre-training for Domain Generalization and Adaptation
Abstract:
Deep models must learn robust and transferable representations in order to perform well on new domains. While domain transfer methods (\eg, domain adaptation, domain generalization) have been proposed to learn transferable representations across domains, they are typically applied to ResNet backbones pre-trained on ImageNet. Thus, existing works pay little attention to the effects of pre-training on domain transfer tasks. In this paper, we provide a broad study and in-depth analysis of pre-training for domain adaptation and generalization, namely: network architectures, size, pre-training loss, and datasets. We observe that simply using a state-of-the-art backbone outperforms existing state-of-the-art domain adaptation baselines and set new baselines on Office-Home and DomainNet improving by 10.7% and 5.5%. We hope that this work can provide more insights for future domain transfer research.



Paperid:1389
Authors:Tao Sun; Cheng Lu; Haibin Ling
Title: Prior Knowledge Guided Unsupervised Domain Adaptation
Abstract:
The waive of labels in the target domain makes Unsupervised Domain Adaptation (UDA) an attractive technique in many real-world applications, though it also brings great challenges as model adaptation becomes harder without labeled target data. In this paper, we address this issue by seeking compensation from target domain prior knowledge, which is often (partially) available in practice, e.g., from human expertise. This leads to a novel yet practical setting where in addition to the training data, some prior knowledge about the target class distribution are available. We term the setting as Knowledge-guided Unsupervised Domain Adaptation (KUDA). In particular, we consider two specific types of prior knowledge about the class distribution in the target domain: Unary Bound that describes the lower and upper bounds of individual class probabilities, and Binary Relationship that describes the relations between two class probabilities. We propose a general rectification module that uses such prior knowledge to refine model generated pseudo labels. The module is formulated as a Zero-One Programming problem derived from the prior knowledge and a smooth regularizer. It can be easily plugged into self-training based UDA methods, and we combine it with two state-of-the-art methods, SHOT and DINE. Empirical results on four benchmarks confirm that the rectification module clearly improves the quality of pseudo labels, which in turn benefits the self-training stage. With the guidance from prior knowledge, the performances of both methods are substantially boosted. We expect our work to inspire further investigations in integrating prior knowledge in UDA. Code is available at https://github.com/tsun/KUDA.



Paperid:1390
Authors:Gilhyun Nam; Gyeongjae Choi; Kyungmin Lee
Title: GCISG: Guided Causal Invariant Learning for Improved Syn-to-Real Generalization
Abstract:
Training a deep learning model with artificially generated data can be an alternative when training data are scarce, yet it suffers from poor generalization performance due to a large domain gap. In this paper, we characterize the domain gap by using a causal framework for data generation. We assume that the real and synthetic data have common content variables but different style variables. Thus, a model trained on synthetic dataset might have poor generalization as the model learns the nuisance style variables. To that end, we propose causal invariance learning which encourages the model to learn a style-invariant representation that enhances the syn-to-real generalization. Furthermore, we propose a simple yet effective feature distillation method that prevents catastrophic forgetting of semantic knowledge of the real domain. In sum, we refer to our method as Guided Causal Invariant Syn-to-real Generalization that effectively improves the performance of syn-to-real generalization. We empirically verify the validity of proposed methods, and especially, our method achieves state-of-the-art on visual syn-to-real domain generalization tasks such as image classification and semantic segmentation.



Paperid:1391
Authors:Yipeng Gao; Lingxiao Yang; Yunmu Huang; Song Xie; Shiyong Li; Wei-Shi Zheng
Title: AcroFOD: An Adaptive Method for Cross-Domain Few-Shot Object Detection
Abstract:
Under the domain shift, cross-domain few-shot object detection aims to adapt object detectors in the target domain with a few annotated target data. There exists two significant challenges: (1) Highly insufficient target domain data; (2) Potential over-adaptation and misleading caused by inappropriately amplified target samples without any restriction. To address these challenges, we propose an adaptive method consisting of two parts. First, we propose an adaptive optimization strategy to select augmented data similar to target samples rather than blindly increasing the amount. Specifically, we filter the augmented candidates which significantly deviate from the target feature distribution in the very beginning. Second, to further relieve the data limitation, we propose the multi-level domain-aware data augmentation to increase the diversity and rationality of augmented data, which exploits the cross-image foreground-background mixture. Experiments show that the proposed method achieves state-of-the-art performance on multiple benchmarks.



Paperid:1392
Authors:Jayeon Yoo; Inseop Chung; Nojun Kwak
Title: Unsupervised Domain Adaptation for One-Stage Object Detector Using Offsets to Bounding Box
Abstract:
Most existing domain adaptive object detection methods exploit adversarial feature alignment to adapt the model to a new domain. Recent advances in adversarial feature alignment strives to reduce the negative effect of alignment, or negative transfer, that occurs because the distribution of features varies depending on the category of objects. However, by analyzing the features of the anchor-free one-stage detector, in this paper, we find that negative transfer may occur because the feature distribution varies depending on the regression value for the offset to the bounding box as well as the category. To obtain domain invariance by addressing this issue, we align the feature conditioned on the offset value, considering the modality of the feature distribution. With a very simple and effective conditioning method, we propose OADA (Offset-Aware Domain Adaptive object detector) that achieves state-of-the-art performances in various experimental settings. In addition, by analyzing through singular value decomposition, we find that our model enhances both discriminability and transferability.



Paperid:1393
Authors:Menglin Jia; Luming Tang; Bor-Chun Chen; Claire Cardie; Serge Belongie; Bharath Hariharan; Ser-Nam Lim
Title: Visual Prompt Tuning
Abstract:
The current modus operandi in adapting pre-trained models involves updating all the backbone parameters, i.e. full fine-tuning. This paper introduces Visual Prompt Tuning (VPT) as an efficient and effective alternative to full fine-tuning for large-scale Transformer models in vision. Taking inspiration from recent advances in efficiently tuning large language models, VPT introduces only a small amount (less than 1% of model parameters) of trainable parameters in the input space while keeping the model backbone frozen. Via extensive experiments on a wide variety of downstream recognition tasks, we show that VPT achieves significant performance gains compared to other parameter-efficient tuning protocols. Most importantly, VPT even outperforms full fine-tuning in many cases across model capacities and training scales, while reducing per-task storage cost. Code is available at https://github.com/kmnp/vpt.



Paperid:1394
Authors:Yongwei Chen; Zihao Wang; Longkun Zou; Ke Chen; Kui Jia
Title: Quasi-Balanced Self-Training on Noise-Aware Synthesis of Object Point Clouds for Closing Domain Gap
Abstract:
Semantic analyses of object point clouds are largely driven by releasing of benchmarking datasets, including synthetic ones whose instances are sampled from object CAD models. However, learning from synthetic data may not generalize to practical scenarios, where point clouds are typically incomplete, non-uniformly distributed, and noisy. Such a challenge of Simulation-to-Reality (Sim2Real) domain gap could be mitigated via learning algorithms of domain adaptation; however, we argue that generation of synthetic point clouds via more physically realistic rendering is a powerful alternative, as systematic non-uniform noise patterns can be captured. To this end, we propose an integrated scheme consisting of physically realistic synthesis of object point clouds via rendering stereo images via projection of speckle patterns onto CAD models and a novel quasi-balanced self-training designed for more balanced data distribution by sparsity-driven selection of pseudo labeled samples for long tailed classes. Experiment results can verify the effectiveness of our method as well as both of its modules for unsupervised domain adaptation on point cloud classification, achieving the state-of-the-art performance. Source codes and the SpeckleNet synthetic dataset are available at https://github.com/Gorilla-Lab-SCUT/QS3.



Paperid:1395
Authors:Xinhao Li; Jingjing Li; Zhekai Du; Lei Zhu; Wen Li
Title: Interpretable Open-Set Domain Adaptation via Angular Margin Separation
Abstract:
Open-set Domain Adaptation (OSDA) aims to recognize classes in the target domain that are seen in the source domain while rejecting other unseen target-exclusive classes into an unknown class, which ignores the diversity of the latter and is therefore incapable of their interpretation. The recently-proposed Semantic Recovery OSDA (SR-OSDA) brings in semantic attributes and attacks the challenge via partial alignment and visual-semantic projection, marking the first step towards interpretable OSDA. Following that line, in this work, we propose a representation learning framework termed Angular Margin Separation (AMS) that unveils the power of discriminative and robust representation for both open-set domain adaptation and cross-domain semantic recovery. Our core idea is to exploit an additive angular margin with regularization for both robust feature fine-tuning and discriminative joint feature alignment, which turns out advantageous to learning an accurate and less biased visual-semantic projection. Further, we propose a post-training re-projection that boosts the performance of seen classes interpretation without deterioration on unseen classes. Verified by extensive experiments, AMS achieves a notable improvement over the existing SR-OSDA baseline, with an average 7.6% increment in semantic recovery accuracy of unseen classes in multiple transfer tasks. Our code is available at https://github.com/LeoXinhaoLee/AMS.



Paperid:1396
Authors:Rui Gong; Martin Danelljan; Dengxin Dai; Danda Pani Paudel; Ajad Chhatkuli; Fisher Yu; Luc Van Gool
Title: TACS: Taxonomy Adaptive Cross-Domain Semantic Segmentation
Abstract:
Traditional domain adaptive semantic segmentation addresses the task of adapting a model to a novel target domain under limited or no additional supervision. While tackling the input domain gap, the standard domain adaptation settings assume no domain change in the output space. In semantic prediction tasks, different datasets are often labeled according to different semantic taxonomies. In many real-world settings, the target domain task requires a different taxonomy than the one imposed by the source domain. We therefore introduce the more general taxonomy adaptive cross-domain semantic segmentation (TACS) problem, allowing for inconsistent taxonomies between the two domains. We further propose an approach that jointly addresses the image-level and label-level domain adaptation. On the label-level, we employ a bilateral mixed sampling strategy to augment the target domain, and a relabelling method to unify and align the label spaces. We address the image-level domain gap by proposing an uncertainty-rectified contrastive learning method, leading to more domain-invariant and class-discriminative features. We extensively evaluate the effectiveness of our framework under different TACS settings: open taxonomy, coarse-to-fine taxonomy, and implicitly-overlapping taxonomy. Our approach outperforms the previous state-of-the-art by a large margin, while being capable of adapting to target taxonomies. Our implementation is publicly available at https://github.com/ETHRuiGong/TADA.



Paperid:1397
Authors:Zhengkai Jiang; Yuxi Li; Ceyuan Yang; Peng Gao; Yabiao Wang; Ying Tai; Chengjie Wang
Title: Prototypical Contrast Adaptation for Domain Adaptive Semantic Segmentation
Abstract:
Unsupervised Domain Adaptation (UDA) aims to adapt the model trained on the labeled source domain to an unlabeled target domain. In this paper, we present Prototypical Contrast Adaptation (ProCA), a simple and efficient contrastive learning method for unsupervised domain adaptive semantic segmentation. Previous domain adaptation methods merely consider the alignment of the intra-class representational distributions across various domains, while the inter-class structural relationship is insufficiently explored, resulting in the aligned representations on the target domain might not be as easily discriminated as done on the source domain anymore. Instead, ProCA incorporates inter-class information into class-wise prototypes, and adopts the class-centered distribution alignment for adaptation. By considering the same class prototypes as positives and other class prototypes as negatives to achieve class-centered distribution alignment, ProCA achieves state-of-the-art performance on classical domain adaptation tasks, {\em i.e., GTA5 $\to$ Cityscapes \text{and} SYNTHIA $\to$ Cityscapes}. Code will be made available.



Paperid:1398
Authors:Hanbin Zhao; Fengyu Yang; Xinghe Fu; Xi Li
Title: RBC: Rectifying the Biased Context in Continual Semantic Segmentation
Abstract:
Recent years have witnessed a great development of Convolutional Neural Networks in semantic segmentation, where all classes of training images are simultaneously available. In practice, new images are usually made available in a consecutive manner, leading to a problem called Continual Semantic Segmentation (CSS). Typically, CSS faces the forgetting problem since previous training images are unavailable, and the semantic shift problem of the background class. Considering the semantic segmentation as a context-dependent pixel-level classification task, we explore CSS from a new perspective of context analysis in this paper. We observe that the context of old-class pixels in the new images is much more biased on new classes than that in the old images, which can sharply aggravate the old-class forgetting and new-class overfitting. To tackle the obstacle, we propose a biased-context-rectified CSS framework with a context-rectified image-duplet learning scheme and a biased-context-insensitive consistency loss. Furthermore, we propose an adaptive re-weighting class-balanced learning strategy for the biased class distribution. Our approach outperforms state-of-the-art methods by a large margin in existing CSS scenarios.



Paperid:1399
Authors:Xingyi Yang; Jingwen Ye; Xinchao Wang
Title: Factorizing Knowledge in Neural Networks
Abstract:
In this paper, we explore a novel and ambitious knowledge-transfer task, termed Knowledge Factorization (KF). The core idea of KF lies in the modularization and assemblability of knowledge: given a pretrained network model as input, KF aims to decompose it into several factor networks, each of which handles only a dedicated task and maintains task-specific knowledge factorized from the source network. Such factor networks are task-wise disentangled and can be directly assembled, without any fine-tuning, to produce the more competent combined-task networks. In other words, the factor networks serve as Lego-brick-like building blocks, allowing us to construct customized networks in a plug-and-play manner. Specifically, each factor network comprises two modules, a common-knowledge module that is task-agnostic and shared by all factor networks, alongside with a task-specific module dedicated to the factor network itself. We introduce an information-theoretic objective, InfoMax-Bottleneck (IMB), to carry out KF by optimizing the mutual information between the learned representations and input. Experiments across various benchmarks demonstrate that, the derived factor networks yield gratifying performances on not only the dedicated tasks but also disentanglement, while enjoying much better interpretability and modularity. Moreover, the learned common-knowledge representations give rise to impressive results on transfer learning. Our code is available at https://github.com/Adamdad/KnowledgeFactor.



Paperid:1400
Authors:Jaemin Na; Dongyoon Han; Hyung Jin Chang; Wonjun Hwang
Title: Contrastive Vicinal Space for Unsupervised Domain Adaptation
Abstract:
Recent unsupervised domain adaptation methods have utilized vicinal space between the source and target domains. However, the equilibrium collapse of labels, a problem where the source labels are dominant over the target labels in the predictions of vicinal instances, has never been addressed. In this paper, we propose an instance-wise minimax strategy that minimizes the entropy of high uncertainty instances in the vicinal space to tackle the stated problem. We divide the vicinal space into two subspaces through the solution of the minimax problem: contrastive space and consensus space. In the contrastive space, inter-domain discrepancy is mitigated by constraining instances to have contrastive views and labels, and the consensus space reduces the confusion between intra-domain categories. The effectiveness of our method is demonstrated on public benchmarks, including Office-31, Office-Home, and VisDA-C, achieving state-of-the-art performances. We further show that our method outperforms the current state-of-the-art methods on PACS, which indicates that our instance-wise approach works well for multi-source domain adaptation as well. Code is available at https://github.com/NaJaeMin92/CoVi.



Paperid:1401
Authors:Sk Miraj Ahmed; Suhas Lohit; Kuan-Chuan Peng; Michael J. Jones; Amit K. Roy-Chowdhury
Title: Cross-Modal Knowledge Transfer without Task-Relevant Source Data
Abstract:
Cost-effective depth and infrared sensors as alternatives to usual RGB sensors are now a reality, and have some advantages over RGB in domains like autonomous navigation and remote sensing. As such, building computer vision and deep learning systems for depth and infrared data are crucial. However, large labeled datasets for these modalities are still lacking. In such cases, transferring knowledge from a neural network trained on a well-labeled large dataset in the source modality (RGB) to a neural network that works on a target modality (depth, infrared, etc.) is of great value. For reasons like memory and privacy, it may not be possible to access the source data, and knowledge transfer needs to work with only the source models. We describe an effective solution, SOCKET: SOurce-free Cross-modal KnowledgE Transfer for this challenging task of transferring knowledge from one source modality to a different target modality without access to task-relevant source data. The framework reduces the modality gap using paired task-irrelevant data, as well as by matching the mean and variance of the target features with the batch-norm statistics that are present in the source models. We show through extensive experiments that our method significantly outperforms existing source-free methods for classification tasks which do not account for the modality gap.



Paperid:1402
Authors:Theodoros Panagiotakopoulos; Pier Luigi Dovesi; Linus Härenstam-Nielsen; Matteo Poggi
Title: Online Domain Adaptation for Semantic Segmentation in Ever-Changing Conditions
Abstract:
Unsupervised Domain Adaptation (UDA) aims at reducing the domain gap between training and testing data and is, in most cases, carried out in offline manner. However, domain changes may occur continuously and unpredictably during deployment (e.g. sudden weather changes). In such conditions, deep neural networks witness dramatic drops in accuracy and offline adaptation may not be enough to contrast it. In this paper, we tackle Online Domain Adaptation (OnDA) for semantic segmentation. We design a pipeline that is robust to continuous domain shifts, either gradual or sudden, and we evaluate it in the case of rainy and foggy scenarios. Our experiments show that our framework can effectively adapt to new domains during deployment, while not being affected by catastrophic forgetting of the previous domains.



Paperid:1403
Authors:Yuecong Xu; Jianfei Yang; Haozhi Cao; Keyu Wu; Min Wu; Zhenghua Chen
Title: Source-Free Video Domain Adaptation by Learning Temporal Consistency for Action Recognition
Abstract:
Video-based Unsupervised Domain Adaptation (VUDA) methods improve the robustness of video models, enabling them to be applied to action recognition tasks across different environments. However, these methods require constant access to source data during the adaptation process. Yet in many real-world applications, subjects and scenes in the source video domain should be irrelevant to those in the target video domain. With the increasing emphasis on data privacy, such methods that require source data access would raise serious privacy issues. Therefore, to cope with such concern, a more practical domain adaptation scenario is formulated as the Source-Free Video-based Domain Adaptation (SFVDA). Though there are a few methods for Source-Free Domain Adaptation (SFDA) on image data, these methods yield degenerating performance in SFVDA due to the multi-modality nature of videos, with the existence of additional temporal features. In this paper, we propose a novel Attentive Temporal Consistent Network (ATCoN) to address SFVDA by learning temporal consistency, guaranteed by two novel consistency objectives, namely feature consistency and source prediction consistency, performed across local temporal features. ATCoN further constructs effective overall temporal features by attending to local temporal features based on prediction confidence. Empirical results demonstrate the state-of-the-art performance of ATCoN across various cross-domain action recognition benchmarks.



Paperid:1404
Authors:Sanqing Qu; Guang Chen; Jing Zhang; Zhijun Li; Wei He; Dacheng Tao
Title: BMD: A General Class-Balanced Multicentric Dynamic Prototype Strategy for Source-Free Domain Adaptation
Abstract:
Source-free Domain Adaptation (SFDA) aims to adapt a pre-trained source model to the unlabeled target domain without accessing the well-labeled source data, which is a much more practical setting due to the data privacy, security, and transmission issues. To make up for the absence of source data, most existing methods introduced feature prototype based pseudo-labeling strategies to realize self-training model adaptation. However, feature prototypes are obtained by instance-level predictions based feature clustering, which is category-biased and tends to result in noisy labels since the visual domain gaps between source and target are usually different between categories. In addition, we found that a monocentric feature prototype may be ineffective to represent each category and introduce negative transfer, especially for those hard-transfer data. To address these issues, we propose a general class-Balanced Multicentric Dynamic prototype (BMD) strategy for the SFDA task. Specifically, for each target category, we first introduce a global inter-class balanced sampling strategy to aggregate potential representative target samples. Then, we design an intra-class multicentric clustering strategy to achieve more robust and representative prototypes generation. In contrast to existing strategies that update the pseudo label at a fixed training period, we further introduce a dynamic pseudo labeling strategy to incorporate network update information during model adaptation. Extensive experiments show that the proposed model-agnostic BMD strategy significantly improves representative SFDA methods to yield new state-of-the-art results. The code is available at https://github.com/ispc-lab/BMD.



Paperid:1405
Authors:Yawen Huang; Feng Zheng; Xu Sun; Yuexiang Li; Ling Shao; Yefeng Zheng
Title: Generalized Brain Image Synthesis with Transferable Convolutional Sparse Coding Networks
Abstract:
High inter-equipment variability and expensive examination costs of brain imaging remain key challenges in leveraging the heterogeneous scans effectively. Despite rapid growth in image-to-image translation with deep learning models, the target brain data may not always be achievable due to the specific attributes of brain imaging. In this paper, we present a novel generalized brain image synthesis method, powered by our transferable convolutional sparse coding networks, to address the lack of interpretable cross-modal medical image representation learning. The proposed approach masters the ability to imitate the machine-like anatomically meaningful imaging by translating features directly under a series of mathematical processings, leading to the reduced domain discrepancy while enhancing model transferability. Specifically, we first embed the globally normalized features into a domain discrepancy metric to learn the domain-invariant representations, then optimally preserve domain-specific geometrical property to reflect the intrinsic graph structures, and further penalize their subspace mismatching to reduce the generalization error. The overall framework is cast in a minimax setting, and the extensive experiments show that the proposed method yields state-of-the-art results on multiple datasets.



Paperid:1406
Authors:Haifeng Xia; Pu Wang; Zhengming Ding
Title: Incomplete Multi-View Domain Adaptation via Channel Enhancement and Knowledge Transfer
Abstract:
Unsupervised domain adaptation (UDA) borrows well-labeled source knowledge to solve the specific task on unlabeled target domain with the assumption that both domains are from a single sensor, e.g., RGB or depth images. To boost model performance, multiple sensors are deployed on new-produced devices like autonomous vehicles to benefit from enriched information. However, the model trained with multi-view data difficultly becomes compatible with conventional devices only with a single sensor. This scenario is defined as incomplete multi-view domain adaptation (IMVDA), which considers that the source domain consists of multi-view data while the target domain only includes single-view instances. To overcome this practical demand, this paper proposes a novel Channel Enhancement and Knowledge Transfer (CEKT) framework with two modules. Concretely, the source channel enhancement module distinguishes view-common from view-specific channels and explores channel similarity to magnify the representation of important channels. Moreover, the adaptive knowledge transfer module attempts to enhance target representation towards multi-view semantic through implicit missing view recovery and adaptive cross-domain alignment. Extensive experimental results illustrate the effectiveness of our method in solving the IMVDA challenge.



Paperid:1407
Authors:Xueqing Deng; Dawei Sun; Shawn Newsam; Peng Wang
Title: DistPro: Searching a Fast Knowledge Distillation Process via Meta Optimization
Abstract:
Recent Knowledge distillation (KD) studies show that different manually designed schemes impact the learned results significantly. Yet, in KD, automatically searching an optimal distillation scheme has not yet been well explored. In this paper, we propose DistPro, a novel framework which searches for an optimal KD process via differentiable meta-learning. Specifically, given a pair of student and teacher networks, DistPro first sets up a rich set of KD pathways from the transmitting layers of the teacher to the receiving layers of the student, and in the meanwhile, various transforms are also proposed for comparing feature maps along a pathway for the distillation. Then, each combination of a pathway and a transform choice is associated with a stochastic weighting process which indicates its importance at every step during the distillation. In the searching stage, the process can be effectively learned through our proposed bi-level meta-optimization strategy. In the distillation stage, DistPro adopts the learned processes for knowledge distillation, which significantly improves the student accuracy especially when faster training is required. Lastly, we find the learned processes can be generalized between similar tasks and networks. In our experiments, DistPro produces state-of-the-art (SoTA) accuracy under varying number of learning epochs on popular datasets, i.e. CIFAR100 and ImageNet, which demonstrate the effectiveness of our framework.



Paperid:1408
Authors:Fei Pan; Sungsu Hur; Seokju Lee; Junsik Kim; In So Kweon
Title: ML-BPM: Multi-Teacher Learning with Bidirectional Photometric Mixing for Open Compound Domain Adaptation in Semantic Segmentation
Abstract:
Open compound domain adaptation (OCDA) considers the target domain as the compound of multiple unknown homogeneous subdomains. The goal of OCDA is to minimize the domain gap between the source domain and the compound target domain, which brings the benefit of the model generalization to the unseen domains. Current OCDA for semantic segmentation methods adopt manual domain separation and employ a single model to adapt to all the target subdomains simultaneously. However, adapting to a target subdomain hinders the model from adapting to other dissimilar target subdomains, which leads to limited performance. In this work, we introduce a multi-teacher framework with bidirectional photometric mixing to adapt to every target subdomain separately. First, we present an automatic domain separation to find the optimal number of subdomains. On this basis, we propose a multi-teacher framework in which each teacher model uses the bidirectional photometric mixing to adapt to one target subdomain. Furthermore, we conduct an adaptive distillation to learn a student model and apply consistency regularization to improve the student generalization. Experimental results on benchmark datasets demonstrate the effectiveness of our approach for both the compound domain and the open domains against existing state-of-the-art approaches.



Paperid:1409
Authors:Nan Ding; Xi Chen; Tomer Levinboim; Soravit Changpinyo; Radu Soricut
Title: PACTran: PAC-Bayesian Metrics for Estimating the Transferability of Pretrained Models to Classification Tasks
Abstract:
With the increasing abundance of pretrained models in recent years, the problem of selecting the best pretrained checkpoint for a particular downstream classification task has been gaining increased attention. Although several methods have recently been proposed to tackle the selection problem (e.g. LEEP, H-score), these methods resort to applying heuristics that are not well motivated by learning theory. In this paper we present PACTran, a theoretically grounded family of metrics for pretrained model selection and transferability measurement. We first show how to derive PACTran metrics from the optimal PAC-Bayesian bound under the transfer learning setting. We then empirically evaluate three metric instantiations of PACTran on a number of vision tasks (VTAB) as well as a language-and-vision (OKVQA) task. An analysis of the results shows PACTran is a more consistent and effective transferability measure compared to existing selection methods.



Paperid:1410
Authors:Xiang Deng; Jian Zheng; Zhongfei Zhang
Title: Personalized Education: Blind Knowledge Distillation
Abstract:
Knowledge distillation compresses a large model (teacher) to a smaller one by letting the student imitate the outputs of the teacher. An interesting question is why the student still typically underperforms the teacher after the imitation. The existing literature usually attributes this to model capacity differences between them. However, capacity differences are unavoidable in model compression, and even large capacity differences are desired for achieving high compression rates. By designing exploratory experiments with theoretical analysis, we find that model capacity differences are not necessarily the root reason; instead the distillation data matter when the student capacity is greater than a threshold. In light of this, we propose personalized education (PE) to first help each student adaptively find its own blind knowledge region (BKR) where the student has not captured the knowledge from the teacher, and then teach the student on this region. Extensive experiments on several benchmark datasets demonstrate that PE substantially reduces the performance gap between students and teachers, even enables small students to outperform large teachers, and also beats the state-of-the-art approaches. Code link: https://github.com/Xiang-Deng-DL/PEBKD"



Paperid:1411
Authors:Wenqi Shao; Xun Zhao; Yixiao Ge; Zhaoyang Zhang; Lei Yang; Xiaogang Wang; Ying Shan; Ping Luo
Title: Not All Models Are Equal: Predicting Model Transferability in a Self-Challenging Fisher Space
Abstract:
This paper addresses an important problem of ranking the pre-trained deep neural networks and screening the most transferable ones for downstream tasks. It is challenging because the ground-truth model ranking for each task can only be generated by fine-tuning the pre-trained models on the target dataset, which is brute-force and computationally expensive. Recent advanced methods proposed several lightweight transferability metrics to predict the fine-tuning results. However, these approaches only capture static representations but neglect the fine-tuning dynamics. To this end, this paper proposes a new transferability metric, called \textbf{S}elf-challenging \textbf{F}isher \textbf{D}iscriminant \textbf{A}nalysis (\textbf{SFDA}), which has many appealing benefits that existing works do not have. First, SFDA can embed the static features into a Fisher space and refine them for better separability between classes. Second, SFDA uses a self-challenging mechanism to encourage different pre-trained models to differentiate on hard examples. Third, SFDA can easily select multiple pre-trained models for the model ensemble. Extensive experiments on $33$ pre-trained models of $11$ downstream tasks show that SFDA is efficient, effective, and robust when measuring the transferability of pre-trained models. For instance, compared with the state-of-the-art method NLEEP, SFDA demonstrates an average of $59.1$\% gain while bringing $22.5$x speedup in wall-clock time. The code will be available at \url{https://github.com/TencentARC/SFDA}.



Paperid:1412
Authors:Andrea Agostinelli; Michal Pándy; Jasper Uijlings; Thomas Mensink; Vittorio Ferrari
Title: How Stable Are Transferability Metrics Evaluations?
Abstract:
Transferability metrics is a maturing field with increasing interest, which aims at providing heuristics for selecting the most suitable source models to transfer to a given target dataset, without fine-tuning them all. However, existing works rely on custom experimental setups which differ across papers, leading to inconsistent conclusions about which transferability metrics work best. In this paper we conduct a large-scale study by systematically constructing a broad range of 715k experimental setup variations. We discover that even small variations to an experimental setup lead to different conclusions about the superiority of a transferability metric over another. Then we propose better evaluations by aggregating across many experiments, enabling to reach more stable conclusions. As a result, we reveal the superiority of LogME at selecting good source datasets to transfer from in a semantic segmentation scenario, NLEEP at selecting good source architectures in an image classification scenario, and GBC at determining which target task benefits most from a given source model. Yet, no single transferability metric works best in all scenarios.



Paperid:1413
Authors:Rang Meng; Xianfeng Li; Weijie Chen; Shicai Yang; Jie Song; Xinchao Wang; Lei Zhang; Mingli Song; Di Xie; Shiliang Pu
Title: Attention Diversification for Domain Generalization
Abstract:
Convolutional neural networks (CNNs) have demonstrated gratifying results at learning discriminative features. However, when applied to unseen domains, state-of-the-art models are usually prone to errors due to domain shift. After investigating this issue from the perspective of shortcut learning, we find the devils lie in the fact that models trained on different domains merely bias to different domain-specific features yet overlook diverse task-related features. Under this guidance, a novel Attention Diversification framework is proposed, in which Intra-Model and Inter-Model Attention Diversification Regularization are collaborated to reassign appropriate attention to diverse task-related features. Briefly, Intra-Model Attention Diversification Regularization is equipped on the high-level feature maps to achieve in-channel discrimination and cross-channel diversification via forcing different channels to pay their most salient attention to different spatial locations. Besides, Inter-Model Attention Diversification Regularization is proposed to further provide task-related attention diversification and domain-related attention suppression, which is a paradigm of ""simulate, divide and assemble"": simulate domain shift via exploiting multiple domain-specific models, divide attention maps into task-related and domain-related groups, and assemble them within each group respectively to execute regularization. Extensive experiments and analyses are conducted on various benchmarks to demonstrate that our method achieves state-of-the-art performance over other competing methods. Code is available at https://github.com/hikvision-research/DomainGeneralization.



Paperid:1414
Authors:Zhaoning Sun; Nico Messikommer; Daniel Gehrig; Davide Scaramuzza
Title: ESS: Learning Event-Based Semantic Segmentation from Still Images
Abstract:
Retrieving accurate semantic information in challenging high dynamic range (HDR) and high-speed conditions remains an open challenge for image-based algorithms due to severe image degradations. Event cameras promise to address these challenges since they feature a much higher dynamic range and are resilient to motion blur. Nonetheless, semantic segmentation with event cameras is still in its infancy which is chiefly due to the lack of high-quality, labeled datasets. In this work, we introduce ESS (Event-based Semantic Segmentation), which tackles this problem by directly transferring the semantic segmentation task from existing labeled image datasets to unlabeled events via unsupervised domain adaptation (UDA). Compared to existing UDA methods, our approach aligns recurrent, motion-invariant event embeddings with image embeddings. For this reason, our method neither requires video data nor per-pixel alignment between images and events and, crucially, does not need to hallucinate motion from still images. Additionally, we introduce DSEC-Semantic, the first large-scale event-based dataset with fine-grained labels. We show that using image labels alone, ESS outperforms existing UDA approaches, and when combined with event labels, it even outperforms state-of-the-art supervised approaches on both DDD17 and DSEC-Semantic. Finally, ESS is general-purpose, which unlocks the vast amount of existing labeled image datasets and paves the way for new and exciting research directions in new fields previously inaccessible for event cameras.



Paperid:1415
Authors:Yuetian Weng; Zizheng Pan; Mingfei Han; Xiaojun Chang; Bohan Zhuang
Title: An Efficient Spatio-Temporal Pyramid Transformer for Action Detection
Abstract:
The task of action detection aims at deducing both the action category and localization of the start and end moment for each action instance in a long, untrimmed video. While vision Transformers have driven the recent advances in video understanding, it is non-trivial to design an efficient architecture for action detection due to the prohibitively expensive self-attentions over a long sequence of video clips. To this end, we present an efficient hierarchical Spatio-Temporal Pyramid Transformer (STPT) for action detection, building upon the fact that the early self-attention layers in Transformers still focus on local patterns. Specifically, we propose to use local window attention to encode rich local spatio-temporal representations in the early stages while applying global attention modules to capture long-term space-time dependencies in the later stages. In this way, our STPT can encode both locality and dependency with largely reduced redundancy, delivering a promising trade-off between accuracy and efficiency. For example, with only RGB input, the proposed STPT achieves 53.6% mAP on THUMOS14, surpassing I3D+AFSD RGB model by over 10% and performing favorably against state-of-the-art AFSD that uses additional flow features with 31% fewer GFLOPs, which serves as an effective and efficient end-to-end Transformer-based framework for action detection.



Paperid:1416
Authors:Jiangbei Yue; Dinesh Manocha; He Wang
Title: Human Trajectory Prediction via Neural Social Physics
Abstract:
Trajectory prediction has been widely pursued in many fields, and many model-based and model-free methods have been explored. The former include rule-based, geometric or optimization-based models, and the latter are mainly comprised of deep learning approaches. In this paper, we propose a new method combining both methodologies based on a new Neural Differential Equation model. Our new model (Neural Social Physics or NSP) is a deep neural network within which we use an explicit physics model with learnable parameters. The explicit physics model serves as a strong inductive bias in modeling pedestrian behaviors, while the rest of the network provides a strong data-fitting capability in terms of system parameter estimation and dynamics stochasticity modeling. We compare NSP with 15 recent deep learning methods on 6 datasets and improve the state-of-the-art performance by 5.56%-70%. Besides, we show that NSP has better generalizability in predicting plausible trajectories in drastically different scenarios where the density is 2-5 times as high as the testing data. Finally, we show that the physics model in NSP can provide plausible explanations for pedestrian behaviors, as opposed to black-box deep learning. Code is available: https://github.com/realcrane/Human- Trajectory-Prediction-via-Neural-Social-Physics.



Paperid:1417
Authors:Yuansheng Zhu; Wentao Bao; Qi Yu
Title: Towards Open Set Video Anomaly Detection
Abstract:
Open Set Video Anomaly Detection (OpenVAD) aims to identify abnormal events from video data where both known anomalies and novel ones exist in testing. Unsupervised models learned solely from normal videos are applicable to any testing anomalies but suffer from a high false positive rate. In contrast, weakly supervised methods are effective in detecting known anomalies but could fail in an open world. We develop a novel weakly supervised method for the OpenVAD problem by integrating evidential deep learning (EDL) and normalizing flows (NFs) into a multiple instance learning (MIL) framework. Specifically, we propose to use graph neural networks and triplet loss to learn discriminative features for training the EDL classifier, where the EDL is capable of identifying the unknown anomalies by quantifying the uncertainty. Moreover, we develop an uncertainty-aware selection strategy to obtain clean anomaly instances and a NFs module to generate the pseudo anomalies. Our method is superior to existing approaches by inheriting the advantages of both the unsupervised NFs and the weakly-supervised MIL framework. Experimental results on multiple real-world video datasets show the effectiveness of our method.



Paperid:1418
Authors:Yan-Bo Lin; Jie Lei; Mohit Bansal; Gedas Bertasius
Title: ECLIPSE: Efficient Long-Range Video Retrieval Using Sight and Sound
Abstract:
We introduce an audiovisual method for long-range text-to-video retrieval. Unlike previous approaches designed for short video retrieval (e.g., 5-15 seconds in duration), our approach aims to retrieve minute-long videos that capture complex human actions. One challenge of standard video-only approaches is the large computational cost associated with processing hundreds of densely extracted frames from such long videos. To address this issue, we propose to replace parts of the video with compact audio cues that succinctly summarize dynamic audio events and are cheap to process. Our method, named ECLIPSE (Efficient CLIP with Sound Encoding), adapts the popular CLIP model to an audiovisual video setting, by adding a unified audiovisual transformer block that captures complementary cues from the video and audio streams. In addition to being 2.92× faster and 2.34× memory-efficient than long-range video-only approaches, our method also achieves better text-to-video retrieval accuracy on several diverse long-range video datasets such as ActivityNet, QVHighlights, YouCook2, DiDeMo, and Charades.



Paperid:1419
Authors:Haoyue Cheng; Zhaoyang Liu; Hang Zhou; Chen Qian; Wayne Wu; Limin Wang
Title: Joint-Modal Label Denoising for Weakly-Supervised Audio-Visual Video Parsing
Abstract:
This paper focuses on the weakly-supervised audio-visual video parsing task, which aims to recognize all events belonging to each modality and localize their temporal boundaries. This task is challenging because only overall labels indicating the video events are provided for training. However, an event might be labeled but not appear in one of the modalities, which results in a modality-specific noisy label problem. In this work, we propose a training strategy to identify and remove modality-specific noisy labels dynamically. It is motivated by two key observations: 1) networks tend to learn clean samples first; and 2) a labeled event would appear in at least one modality. Specifically, we sort the losses of all instances within a mini-batch individually in each modality, and then select noisy samples according to the relationships between intra-modal and inter-modal losses. Besides, we also propose a simple but valid noise ratio estimation method by calculating the proportion of instances whose confidence is below a preset threshold. Our method makes large improvements over the previous state of the arts (\eg, from 60.0\% to 63.8\% in segment-level visual metric), which demonstrates the effectiveness of our approach. Code and trained models are publicly available at \url{https://github.com/MCG-NJU/JoMoLD}.



Paperid:1420
Authors:Pengwan Yang; Yuki M. Asano; Pascal Mettes; Cees G. M. Snoek
Title: Less than Few: Self-Shot Video Instance Segmentation
Abstract:
The goal of this paper is to bypass the need for labelled examples in few-shot video understanding at run time. While proven effective, in many practical video settings even labelling a few examples appears unrealistic. This is especially true as the level of details in spatio-temporal video understanding and with it, the complexity of annotations continues to increase. Rather than performing few-shot learning with a human oracle to provide a few densely labelled support videos, we propose to automatically learn to find appropriate support videos given a query. We call this self-shot learning and we outline a simple self-supervised learning method to generate an embedding space well-suited for unsupervised retrieval of relevant samples. To showcase this novel setting, we tackle, for the first time, video instance segmentation in a self-shot (and few-shot) setting, where the goal is to segment instances at the pixel-level across the spatial and temporal domains. We provide strong baseline performances that utilize a novel transformer-based model and show that self-shot learning can even surpass few-shot and can be positively combined for further performance gains. Experiments on new benchmarks show that our approach achieves strong performance, is competitive to oracle support in some settings, scales to large unlabelled video collections, and can be combined in a semi-supervised setting.



Paperid:1421
Authors:Luchuan Song; Zheng Fang; Xiaodan Li; Xiaoyi Dong; Zhenchao Jin; Yuefeng Chen; Siwei Lyu
Title: Adaptive Face Forgery Detection in Cross Domain
Abstract:
It is necessary to develop effective face forgery detection methods with constantly evolving technologies in synthesizing realistic faces which raises serious risks on malicious face tampering. A large and growing body of literature has investigated deep learning-based approaches, especially those taking frequency clues into consideration, have achieved remarkable progress on detecting fake faces. The method based on frequency clues result in the inconsistency across frames and make the final detection result unstable even in the same deepfake video. So, these patterns are still inadequate and unstable. In addition to this, the inconsistency problem in the previous methods is significantly exacerbated due to the diversities among various forgery methods. To address this problem, we propose a novel deep learning framework for face forgery detection in cross domain. The proposed framework explores on mining the potential consistency through the correlated representations across multiple frames as well as the complementary clues from both RGB and frequency domains. We also introduce an instance discrimination module to determine the discriminative results center for each frame across the video, which is a strategy that adaptive adjust with during inference.



Paperid:1422
Authors:Yue Zhao; Philipp Krähenbühl
Title: Real-Time Online Video Detection with Temporal Smoothing Transformers
Abstract:
Streaming video recognition reasons about objects and their actions in every frame of a video. A good streaming recognition model captures both long-term dynamics and short-term changes of video. Unfortunately, in most existing methods, the computational complexity grows linearly or quadratically with the length of the considered dynamics. This issue is particularly pronounced in transformer-based architectures. To address this issue, we reformulate the cross-attention in a video transformer through the lens of kernel and apply two kinds of temporal smoothing kernel: A box kernel or a Laplace kernel. The resulting streaming attention reuses much of the computation from frame to frame, and only requires a constant time update each frame. Based on this idea, we build TeSTra, a Temporal Smoothing Transformer, that takes in arbitrarily long inputs with constant caching and computing overhead. Specifically, it runs 6x faster than equivalent sliding-window based transformers with 2,048 frames in a streaming setting. Furthermore, thanks to the increased temporal span, TeSTra achieves state-of-the-art results on THUMOS’14 and EPIC-Kitchen-100, two standard online action detection and action anticipation datasets. A real-time version of TeSTra outperforms all but one prior approaches on the THUMOS’14 dataset.



Paperid:1423
Authors:Feng Cheng; Gedas Bertasius
Title: TALLFormer: Temporal Action Localization with a Long-Memory Transformer
Abstract:
Most modern approaches in temporal action localization divide this problem into two parts: (i) short-term feature extraction and (ii) long-range temporal boundary localization. Due to the high GPU memory cost caused by processing long untrimmed videos, many methods sacrifice the representational power of the short-term feature extractor by either freezing the backbone or using a small spatial video resolution. This issue becomes even worse with the recent video transformer models, many of which have quadratic memory complexity. To address these issues, we propose TALLFormer, a memory-efficient and end-to-end trainable Temporal Action Localization Transformer with Long-term memory. Our long-term memory mechanism eliminates the need for processing hundreds of redundant video frames during each training iteration, thus, significantly reducing the GPU memory consumption and training time. These efficiency savings allow us (i) to use a powerful video transformer feature extractor without freezing the backbone or reducing the spatial video resolution, while (ii) also maintaining long-range temporal boundary localization capability. With only RGB frames as input and no external action recognition classifier, TALLFormer outperforms previous state-of-the-arts by a large margin, achieving an average mAP of 59.1% on THUMOS14 and 35.6% on ActivityNet-1.3. The code is public available: https://github.com/klauscc/TALLFormer.



Paperid:1424
Authors:Guolei Sun; Yun Liu; Hao Tang; Ajad Chhatkuli; Le Zhang; Luc Van Gool
Title: Mining Relations among Cross-Frame Affinities for Video Semantic Segmentation
Abstract:
The essence of video semantic segmentation (VSS) is how to leverage temporal information for prediction. Previous efforts are mainly devoted to developing new techniques to calculate the cross-frame affinities such as optical flow and attention. Instead, this paper contributes from a different angle by mining relations among cross-frame affinities, upon which better temporal information aggregation could be achieved. We explore relations among affinities in two aspects: single-scale intrinsic correlations and multi-scale relations. Inspired by traditional feature processing, we propose Single-scale Affinity Refinement (SAR) and Multi-scale Affinity Aggregation (MAA). To make it feasible to execute MAA, we propose a Selective Token Masking (STM) strategy to select a subset of consistent reference tokens for different scales when calculating affinities, which also improves the efficiency of our method. At last, the cross-frame affinities strengthened by SAR and MAA are adopted for adaptively aggregating temporal information. Our experiments demonstrate that the proposed method performs favorably against state-of-the-art VSS methods.



Paperid:1425
Authors:Medhini Narasimhan; Arsha Nagrani; Chen Sun; Michael Rubinstein; Trevor Darrell; Anna Rohrbach; Cordelia Schmid
Title: TL;DW? Summarizing Instructional Videos with Task Relevance \& Cross-Modal Saliency
Abstract:
YouTube users looking for instructions for a specific task may spend a long time browsing content trying to find the right video that matches their needs. Creating a visual summary (abridged version of a video) provides viewers with a quick overview and massively reduces search time. In this work, we focus on summarizing instructional videos, an under-explored area of video summarization. In comparison to generic videos, instructional videos can be parsed into semantically meaningful segments that correspond to important steps of the demonstrated task. Existing video summarization datasets rely on manual frame-level annotations, making them subjective and limited in size. To overcome this, we first automatically generate pseudo summaries for a corpus of instructional videos by exploiting two key assumptions: (i) relevant steps are likely to appear in multiple videos of the same task (Task Relevance), and (ii) they are more likely to be described by the demonstrator verbally (Cross-Modal Saliency). We propose an instructional video summarization network that combines a context-aware temporal video encoder and a segment scoring transformer. Using pseudo summaries as weak supervision, our network constructs a visual summary for an instructional video given only video and transcribed speech. To evaluate our model, we collect a high-quality test set, WikiHow Summaries, by scraping WikiHow articles that contain video demonstrations and visual depictions of steps allowing us to obtain the ground-truth summaries. We outperform several baselines and a state-of-the-art video summarization model on this new benchmark.



Paperid:1426
Authors:Megha Nawhal; Akash Abdu Jyothi; Greg Mori
Title: Rethinking Learning Approaches for Long-Term Action Anticipation
Abstract:
Action anticipation involves predicting future actions having observed the initial portion of a video. Typically, the observed video is processed as a whole to obtain a video-level representation of the ongoing activity in the video, which is then used for future prediction. We introduce ANTICIPATR which performs long-term action anticipation leveraging segment-level representations learned using individual segments from different activities, in addition to a video-level representation. We propose a two-stage learning approach to train a novel transformer-based model that uses these two types of representations to directly predict a set of future action instances over any given anticipation duration. Results on Breakfast, 50Salads, Epic-Kitchens-55, and EGTEA Gaze+ datasets demonstrate the effectiveness of our approach.



Paperid:1427
Authors:Yuxuan Liang; Pan Zhou; Roger Zimmermann; Shuicheng Yan
Title: DualFormer: Local-Global Stratified Transformer for Efficient Video Recognition
Abstract:
While transformers have shown great potential on video recognition with their strong capability of capturing long-range dependencies, they often suffer high computational costs induced by the self-attention to the huge number of 3D tokens. In this paper, we present a new transformer architecture termed DualFormer, which can efficiently perform space-time attention for video recognition. Concretely, DualFormer stratifies the full space-time attention into dual cascaded levels, i.e., to first learn fine-grained local interactions among nearby 3D tokens, and then to capture coarse-grained global dependencies between the query token and global pyramid contexts. Different from existing methods that apply space-time factorization or restrict attention computations within local windows for improving efficiency, our local-global stratification strategy can well capture both short- and long-range spatiotemporal dependencies, and meanwhile greatly reduces the number of keys and values in attention computation to boost efficiency. Experimental results verify the superiority of DualFormer on five video benchmarks against existing methods. In particular, DualFormer achieves 82.9%/85.2% top-1 accuracy on Kinetics-400/600 with 1000G inference FLOPs which is at least 3.2x fewer than existing methods with similar performance. We have released the source code at https://github.com/sail-sg/dualformer.



Paperid:1428
Authors:Gensheng Pei; Fumin Shen; Yazhou Yao; Guo-Sen Xie; Zhenmin Tang; Jinhui Tang
Title: Hierarchical Feature Alignment Network for Unsupervised Video Object Segmentation
Abstract:
Optical flow is an easily conceived and precious cue for advancing unsupervised video object segmentation (UVOS). Most of the previous methods directly extract and fuse the motion and appearance features for segmenting target objects in the UVOS setting. However, optical flow is intrinsically an instantaneous velocity of all pixels among consecutive frames, thus making the motion features not aligned well with the primary objects among the corresponding frames. To solve the above challenge, we propose a concise, practical, and efficient architecture for appearance and motion feature alignment, dubbed hierarchical feature alignment network (HFAN). Specifically, the key merits in HFAN are the sequential Feature AlignMent (FAM) module and the Feature AdaptaTion (FAT) module, which are leveraged for processing the appearance and motion features hierarchically. FAM is capable of aligning both appearance and motion features with the primary object semantic representations, respectively. Further, FAT is explicitly designed for the adaptive fusion of appearance and motion features to achieve a desirable trade-off between cross-modal features. Extensive experiments demonstrate the effectiveness of the proposed HFAN, which reaches a new state-of-the-art performance on DAVIS-16, achieving 88.7 J&F Mean, i.e., a relative improvement of 3.5% over the best published result.



Paperid:1429
Authors:Hang Wang; Penghao Zhou; Chong Zhou; Zhao Zhang; Xing Sun
Title: PAC-Net: Highlight Your Video via History Preference Modeling
Abstract:
Autonomous highlight detection is crucial for video editing and video browsing on social media platforms. General video highlight detection aims at extracting the most interesting segments from the entire video. However, interest is subjective among different users. A naive solution is to train a model for each user but it is not practical due to the huge training expense. In this work, we propose a Preference-Adaptive Classification (PAC-Net) framework, which can model users’ personalized preferences from their user history. Specifically, we design a Decision Boundary Customizer (DBC) module to dynamically generate the user-adaptive highlight classifier from the preference-related user history. In addition, we introduce the Mini-History (Mi-Hi) mechanism to capture more fine-grained user-specific preferences. The final highlight prediction is jointly decided by the user’s multiple preferences. Extensive experiments demonstrate that the proposed PAC-Net achieves state-of-the-art performance on the public benchmark dataset, whilst using substantially smaller networks. Code will be made publicly available.



Paperid:1430
Authors:Fida Mohammad Thoker; Hazel Doughty; Piyush Bagad; Cees G. M. Snoek
Title: How Severe Is Benchmark-Sensitivity in Video Self-Supervised Learning?
Abstract:
Despite the recent success of video self-supervised learning models, there is much still to be understood about their generalization capability. In this paper, we investigate how sensitive video self-supervised learning is to the current conventional benchmark and whether methods generalize beyond the canonical evaluation setting. We do this across four different factors of sensitivity: domain, samples, actions and task. Our study which encompasses over 500 experiments on 7 video datasets, 9 self-supervised methods and 6 video understanding tasks, reveals that current benchmarks in video self-supervised learning are not good indicators of generalization along these sensitivity factors. Further, we find that self-supervised methods considerably lag behind vanilla supervised pre-training, especially when domain shift is large and the amount of available downstream samples are low. From our analysis we distill the SEVERE-benchmark, a subset of our experiments, and discuss its implication for evaluating the generalizability of representations obtained by existing and future self-supervised video learning methods. Code is available at https://github.com/fmthoker/SEVERE-BENCHMARK.



Paperid:1431
Authors:Young Hwi Kim; Hyolim Kang; Seon Joo Kim
Title: A Sliding Window Scheme for Online Temporal Action Localization
Abstract:
Most online video understanding tasks aim to immediately process each streaming frame and output predictions frame-by-frame. For extension to instance-level predictions of existing online video tasks, Online Temporal Action Localization (On-TAL) has been recently proposed. However, simple On-TAL approaches of grouping per-frame predictions have limitations due to the lack of instance-level context. To this end, we propose Online Anchor Transformer (OAT) to extend the anchor-based action localization model to the online setting. We also introduce an online-applicable post-processing method that suppresses repetitive action proposals. Evaluations of On-TAL on THUMOS’14, MUSES, and BBDB show significant improvements in terms of mAP, and our model shows comparable performance to the state-of-the-art offline TAL methods with a minor change of the post-processing method. In addition to mAP evaluation, we additionally present a new online-oriented metric of early detection for On-TAL, and measure the responsiveness of each On-TAL approach.



Paperid:1432
Authors:Lin Geng Foo; Tianjiao Li; Hossein Rahmani; Qiuhong Ke; Jun Liu
Title: ERA: Expert Retrieval and Assembly for Early Action Prediction
Abstract:
Early action prediction aims to successfully predict the class label of an action before it is completely performed. This is a challenging task because the beginning stages of different actions can be very similar, with only minor subtle differences for discrimination. In this paper, we propose a novel Expert Retrieval and Assembly (ERA) module that retrieves and assembles a set of experts most specialized at using discriminative subtle differences, to distinguish an input sample from other highly similar samples. To encourage our model to effectively use subtle differences for early action prediction, we push experts to discriminate exclusively between samples that are highly similar, forcing these experts to learn to use subtle differences that exist between those samples. Additionally, we design an effective Expert Learning Rate Optimization method that balances the experts’ optimization and leads to better performance. We evaluate our ERA module on four public action datasets and achieve state-of-the-art performance. Code will be released.



Paperid:1433
Authors:Varshanth Rao; Md Ibrahim Khalil; Haoda Li; Peng Dai; Juwei Lu
Title: Dual Perspective Network for Audio-Visual Event Localization
Abstract:
The Audio-Visual Event Localization (AVEL) problem involves tackling three core sub-tasks: the creation of efficient audio-visual representations using cross-modal guidance, the formation of short-term temporal feature aggregations, and its accumulation to achieve long-term dependency resolution. These sub-tasks are often performed by tailored modules, where the limited inter-module interaction restricts feature learning to a serialized manner. Past works have traditionally viewed videos as temporally sequenced multi-modal streams. We improve and extend on this view by proposing a novel architecture, the Dual Perspective Network (DPNet), that - (1) additionally operates on an intuitive graph perspective of a video to simultaneously facilitate cross-modal guidance and short-term temporal aggregation using a Graph NeuralNetwork (GNN), (2) deploys a Temporal Convolutional Network (TCN)to achieve long-term dependency resolution, and (3) encourages interactive feature learning using an acyclic feature refinement process that alternates between the GNN and TCN. Further, we introduce the RelationalGraph Convolutional Transformer, a novel GNN integrated into the DP-Net, to express and attend each segment node’s relational representation with its different relational neighborhoods. Lastly, we diversify the input to the DPNet through a new video augmentation technique called Replicate and Link, which outputs semantically identical video blends whose graph representations can be linked to that of the source videos. Experiments reveal that our DPNet framework outperforms prior state-of-the-art methods by large margins for the AVEL task on the public AVE dataset, while extensive ablation studies corroborate the efficacy of each proposed method.



Paperid:1434
Authors:Boyang Xia; Wenhao Wu; Haoran Wang; Rui Su; Dongliang He; Haosen Yang; Xiaoran Fan; Wanli Ouyang
Title: NSNet: Non-Saliency Suppression Sampler for Efficient Video Recognition
Abstract:
It is challenging for artificial intelligence systems to achieve accurate video recognition under the scenario of low computation costs. Adaptive inference based efficient video recognition methods typically preview videos and focus on salient parts to reduce computation costs. Most existing works focus on complex networks learning with video classification based objectives. Taking all frames as positive samples, few of them pay attention to the discrimination between positive samples (salient frames) and negative samples (non-salient frames) in supervisions. To fill this gap, in this paper, we propose a novel Non-saliency Suppression Network (NSNet), which effectively suppresses the responses of non-salient frames. Specifically, on the frame level, effective pseudo labels that can distinguish between salient and non-salient frames are generated to guide the frame saliency learning. On the video level, a temporal attention module is learned under dual video-level supervisions on both the salient and the non-salient representations. Saliency measurements from both two levels are combined for exploitation of multi-granularity complementary information. Extensive experiments conducted on four well-known benchmarks verify our NSNet not only achieves the state-of-the-art accuracy-efficiency trade-off but also present a significantly faster (2.4 4.3x) practical inference speed than state-of-the-art methods. Our project page is at https://lawrencexia2008.github.io/projects/nsnet.



Paperid:1435
Authors:Jiabo Huang; Hailin Jin; Shaogang Gong; Yang Liu
Title: Video Activity Localisation with Uncertainties in Temporal Boundary
Abstract:
Current methods for video activity localisation over time assume implicitly that activity temporal boundaries labelled for model training are determined and precise. However, in unscripted natural videos, different activities mostly transit smoothly, so that it is intrinsically ambiguous to determine in labelling precisely when an activity starts and ends over time. Such uncertainties in temporal labelling are currently ignored in model training, resulting in learning mis-matched video-text correlation with poor generalisation in test. In this work, we solve this problem by introducing Elastic Moment Bounding (EMB) to accommodate flexible and adaptive activity temporal boundaries towards modelling universally interpretable video-text correlation with tolerance to underlying temporal uncertainties in pre-fixed annotations. Specifically, we construct elastic boundaries adaptively by mining and discovering frame-wise temporal endpoints that can maximise the alignment between video segments and query sentences. To enable both more accurate matching (segment content attention) and more robust localisation (segment elastic boundaries), we optimise the selection of frame-wise endpoints subject to segment-wise contents by a novel Guided Attention mechanism. Extensive experiments on three video activity localisation benchmarks demonstrate compellingly the EMB’s advantages over existing methods without modelling uncertainty.



Paperid:1436
Authors:Boyang Xia; Zhihao Wang; Wenhao Wu; Haoran Wang; Jungong Han
Title: Temporal Saliency Query Network for Efficient Video Recognition
Abstract:
Efficient video recognition is a hot-spot research topic with the explosive growth of multimedia data on the Internet and mobile devices. Most existing methods select the salient frames without awareness of the class-specific saliency scores, which neglect the implicit association between the saliency of frames and its belonging category. To alleviate this issue, we devise a novel Temporal Saliency Query (TSQ) mechanism, which introduces class-specific information to provide fine-grained cues for saliency measurement. Specifically, we model the class-specific saliency measuring process as a query-response task. For each category, the common pattern of it is employed as a query and the most salient frames are responded to it. Then, the calculated similarities are adopted as the frame saliency scores. To achieve it, we propose a Temporal Saliency Query Network (TSQNet) that includes two instantiations of the TSQ mechanism based on visual appearance similarities and textual event-object relations. Afterward, cross-modality interactions are imposed to promote the information exchange between them. Finally, we use the class-specific saliencies of the most confident categories generated by two modalities to perform the selection of salient frames. Extensive experiments demonstrate the effectiveness of our method by achieving state-of-the-art results on ActivityNet, FCVID and Mini-Kinetics datasets. Our project page is at https://lawrencexia2008.github.io/projects/tsqnet.



Paperid:1437
Authors:Guanxiong Sun; Yang Hua; Guosheng Hu; Neil Robertson
Title: Efficient One-Stage Video Object Detection by Exploiting Temporal Consistency
Abstract:
Recently, one-stage detectors have achieved competitive accuracy and faster speed compared with traditional two-stage detectors on image data. However, in the field of video object detection (VOD), most existing VOD methods are still based on two-stage detectors. Moreover, directly adapting existing VOD methods to one-stage detectors introduces unaffordable computational costs. In this paper, we first analyse the computational bottlenecks of using one-stage detectors for VOD. Based on the analysis, we present a simple yet efficient framework to address the computational bottlenecks and achieve efficient one-stage VOD by exploiting the temporal consistency in video frames. Specifically, our method consists of a location prior network to filter out background regions and a size prior network to skip unnecessary computations on low-level feature maps for specific frames. We test our method on various modern one-stage detectors and conduct extensive experiments on the ImageNet VID dataset. Excellent experimental results demonstrate the superior effectiveness, efficiency, and compatibility of our method. The code is available at https://github.com/guanxiongsun/EOVOD.



Paperid:1438
Authors:Guodong Ding; Angela Yao
Title: Leveraging Action Affinity and Continuity for Semi-Supervised Temporal Action Segmentation
Abstract:
We present a semi-supervised learning approach to the temporal action segmentation task. The goal of the task is to temporally detect and segment actions in long, untrimmed procedural videos, where only a small set of videos are densely labelled, and a large collection of videos are unlabelled. To this end, we propose two novel loss functions for the unlabelled data: an action affinity loss and an action continuity loss. The action affinity loss guides the unlabelled samples learning by imposing the action priors induced from the labelled set. Action continuity loss enforces the temporal continuity of actions, which also provides frame-wise classification supervision. In addition, we propose an Adaptive Boundary Smoothing (ABS) approach to build coarser action boundaries for more robust and reliable learning. The proposed loss functions and ABS were evaluated on three benchmarks. Results show that they significantly improved action segmentation performance with a low amount (5% and 10%) of labelled data and achieved comparable results to full supervision with 50% labelled data. Furthermore, ABS succeeded in boosting performance when integrated into fully-supervised learning.



Paperid:1439
Authors:James Hong; Haotian Zhang; Michaël Gharbi; Matthew Fisher; Kayvon Fatahalian
Title: "Spotting Temporally Precise, Fine-Grained Events in Video"
Abstract:
We introduce the task of spotting temporally precise, fine-grained events in video (detecting the precise moment in time events occur). Precise spotting requires models to reason globally about the full-time scale of actions and locally to identify subtle frame-to-frame appearance and motion differences that identify events during these actions. Surprisingly, we find that top performing solutions to prior video understanding tasks such as action detection and segmentation do not simultaneously meet both requirements. In response, we propose E2E-Spot, a compact, end-to-end model that performs well on the precise spotting task and can be trained quickly on a single GPU. We demonstrate that E2E-Spot significantly outperforms recent baselines adapted from the video action detection, segmentation, and spotting literature to the precise spotting task. Finally, we contribute new annotations and splits to several fine-grained sports action datasets to make these datasets suitable for future work on precise spotting.



Paperid:1440
Authors:Nadine Behrmann; S. Alireza Golestaneh; Zico Kolter; Jürgen Gall; Mehdi Noroozi
Title: Unified Fully and Timestamp Supervised Temporal Action Segmentation via Sequence to Sequence Translation
Abstract:
This paper introduces a unified framework for video action segmentation via sequence to sequence (seq2seq) translation in a fully and timestamp supervised setup. In contrast to current state-of-the-art frame-level prediction methods, we view action segmentation as a seq2seq translation task, i.e., mapping a sequence of video frames to a sequence of action segments. Our proposed method involves a series of modifications and auxiliary loss functions on the standard Transformer seq2seq translation model to cope with long input sequences opposed to short output sequences and relatively few videos. We incorporate an auxiliary supervision signal for the encoder via a frame-wise loss and propose a separate alignment decoder for an implicit duration prediction. Finally, we extend our framework to the timestamp supervised setting via our proposed constrained k-medoids algorithm to generate pseudo-segmentations. Our proposed framework performs consistently on both fully and timestamp supervised settings, outperforming or competing state-of-the-art on several datasets.



Paperid:1441
Authors:Junke Wang; Xitong Yang; Hengduo Li; Li Liu; Zuxuan Wu; Yu-Gang Jiang
Title: Efficient Video Transformers with Spatial-Temporal Token Selection
Abstract:
Video transformers have achieved impressive results on major video recognition benchmarks, however they suffer from high computational cost. In this paper, we present STTS, a token selection framework that dynamically selects a few informative tokens in both temporal and spatial dimensions conditioned on input video samples. Specifically, we formulate token selection as a ranking problem, which estimates the importance of each token through a lightweight scorer network and only those with top scores will be used for downstream evaluation. In the temporal dimension, we keep the frames that are most relevant to the action categories, while in the spatial dimension, we identify the most discriminative region in feature maps without affecting the spatial context used in a hierarchical way in most video transformers. Since the decision of token selection is non-differentiable, we employ a perturbed-maximum based differentiable Top-K operator for end-to-end training. We mainly conduct extensive experiments on Kinetics-400 with a recently introduced video transformer backbone, MViT. Our framework achieves similar results while requiring 20% less computation. We also demonstrate that our approach is generic for different transformer architectures and video datasets.



Paperid:1442
Authors:Md Mohaiminul Islam; Gedas Bertasius
Title: Long Movie Clip Classification with State-Space Video Models
Abstract:
Most modern video recognition models are designed to operate on short video clips (e.g., 5-10s in length). Thus, it is challenging to apply such models to long movie understanding tasks, which typically require sophisticated long-range temporal reasoning. The recently introduced video transformers partially address this issue by using long-range temporal self-attention. However, due to the quadratic cost of self-attention, such models are often costly and impractical to use. Instead, we propose ViS4mer, an efficient long-range video model that combines the strengths of self-attention and the recently introduced structured state-space sequence (S4) layer. Our model uses a standard Transformer encoder for short-range spatiotemporal feature extraction, and a multi-scale temporal S4 decoder for subsequent long-range temporal reasoning. By progressively reducing the spatiotemporal feature resolution and channel dimension at each decoder layer, ViS4mer learns complex long-range spatiotemporal dependencies in a video. Furthermore, ViS4mer is 2.63x faster and requires 8x less GPU memory than the corresponding pure self-attention-based model. Additionally, ViS4mer achieves state-of-the-art results in 6 out of 9 long-form movie video classification tasks on the Long Video Understanding (LVU) benchmark. Furthermore, we show that our approach successfully generalizes to other domains, achieving competitive results on the Breakfast and the COIN procedural activity datasets. The code is publicly available.



Paperid:1443
Authors:Chen Ju; Tengda Han; Kunhao Zheng; Ya Zhang; Weidi Xie
Title: Prompting Visual-Language Models for Efficient Video Understanding
Abstract:
Image-based visual-language (I-VL) pre-training has shown great success for learning joint visual-textual representations from large-scale web data, revealing remarkable ability for zero-shot generalisation. This paper presents a simple but strong baseline to efficiently adapt the pre-trained I-VL model for video understanding tasks, with minimal training. Specifically, we propose to optimise a few random vectors, termed as continuous prompt vectors, that convert video-related tasks into the same format as the pre-training objectives. In addition, to bridge the gap between static images and videos, temporal information is encoded with lightweight Transformers stacking on top of frame-wise visual features. Experimentally, we conduct extensive ablation studies to analyse the critical components. On ten public benchmarks of action recognition, action localisation, and text-video retrieval, across closed-set, few-shot, and zero-shot scenarios, we achieve competitive or state-of-the-art performance to existing methods, despite optimising significantly fewer parameters. Due to space limitation, we refer the readers to the arXiv version at https://arxiv.org/abs/2112.04478.



Paperid:1444
Authors:Huan Li; Ping Wei; Jiapeng Li; Zeyu Ma; Jiahui Shang; Nanning Zheng
Title: Asymmetric Relation Consistency Reasoning for Video Relation Grounding
Abstract:
Video relation grounding has attracted growing attention in the fields of video understanding and multimodal learning. While the past years have witnessed remarkable progress in this issue, the difficulties of multi-instance and complex temporal reasoning make it still a challenging task. In this paper, we propose a novel Asymmetric Relation Consistency (ARC) reasoning model to solve the video relation grounding problem. To overcome the multi-instance confusion problem, an asymmetric relation reasoning method and a novel relation consistency loss are proposed to ensure the consistency of the relationships across multiple instances. In order to precisely localize the relation instance in temporal context, a transformer-based relation reasoning module is proposed. Our model is trained in a weakly-supervised manner. The proposed method was tested on the challenging video relation dataset. Experiments manifest that the performance of our method outperforms the state-of-the-art methods by a large margin. Extensive ablation studies also prove the effectiveness and strength of the proposed method.



Paperid:1445
Authors:Jiacheng Li; Ruize Han; Haomin Yan; Zekun Qian; Wei Feng; Song Wang
Title: Self-Supervised Social Relation Representation for Human Group Detection
Abstract:
Human group detection, which splits crowd of people into groups, is an important step for video-based human social activity analysis. The core of human group detection is the human social relation representation and division. In this paper, we propose a new two-stage multi-head framework for human group detection. In the first stage, we propose a human behavior simulator head to learn the social relation feature embedding, which is self-supervised trained by leveraging the socially grounded multi-person behavior relationship. In the second stage, based on the social relation embedding, we develop a self-attention inspired network for human group detection. Remarkable performance on two state-of-the-art large-scale benchmarks, i.e., PANDA and JRDB-Group, verifies the effectiveness of the proposed framework. Benefiting from the self-supervised social relation embedding, our method can provide promising results with very few (labeled) training data. We have released the source code to the public.



Paperid:1446
Authors:Seong Hyeon Park; Jihoon Tack; Byeongho Heo; Jung-Woo Ha; Jinwoo Shin
Title: K-Centered Patch Sampling for Efficient Video Recognition
Abstract:
For decades, it has been a common practice to choose a subset of video frames for reducing the computational burden of a video understanding model. In this paper, we argue that this popular heuristic might be sub-optimal under recent transformer-based models. Specifically, inspired by that transformers are built upon patches of video frames, we propose to sample patches rather than frames using the greedy K-center search, i.e., the farthest patch to what has been chosen so far is sampled iteratively. We then show that a transformer trained with the selected video patches can outperform its baseline trained with the video frames sampled in the traditional way. Furthermore, by adding a certain spatiotemporal structuredness condition, the proposed K-centered patch sampling can be even applied to the recent sophisticated video transformers, boosting their performance further. We demonstrate the superiority of our method on Something-Something and Kinetics datasets.



Paperid:1447
Authors:Guy Erez; Ron Shapira Weber; Oren Freifeld
Title: A Deep Moving-Camera Background Model
Abstract:
In video analysis, background models have many applications such as background/foreground separation, change detection, anomaly detection, tracking, and more. However, while learning such a model in a video captured by a static camera is a fairly-solved task, in the case of a Moving-camera Background Model (MCBM), the success has been far more modest due to algorithmic and scalability challenges that arise due to the camera motion. Thus, existing MCBMs are limited in their scope and their supported camera-motion types. These hurdles also impeded the employment, in this unsupervised task, of end-to-end solutions based on deep learning (DL). Moreover, existing MCBMs usually model the background either on the domain of a typically-large panoramic image or in an online fashion. Unfortunately, the former creates several problems, including poor scalability, while the latter prevents the recognition and leveraging of cases where the camera revisits previously-seen parts of the scene. This paper proposes a new method, called DeepMCBM, that eliminates all the aforementioned issues and achieves state-of-the-art results. Concretely, first we identify the difficulties associated with joint alignment of video frames in general and in a DL setting in particular. Next, we propose a new strategy for joint alignment that lets us use a spatial transformer net with neither a regularization nor any form of specialized (and non-differentiable) initialization. Coupled with an autoencoder conditioned on unwarped robust central moments (obtained from the joint alignment), this yields an end-to-end regularization-free MCBM that supports a broad range of camera motions and scales gracefully. We demonstrate DeepMCBM’s utility on a variety of videos, including ones beyond the scope of other methods. Our code is available at https://github.com/BGU-CS-VIL/DeepMCBM.



Paperid:1448
Authors:Eitan Kosman; Dotan Di Castro
Title: GraphVid: It Only Takes a Few Nodes to Understand a Video
Abstract:
We propose a concise representation of videos that encode perceptually meaningful features into graphs. With this representation, we aim to leverage the large amount of redundancies in videos and save computations. First, we construct superpixel-based graph representations of videos by considering superpixels as graph nodes and create spatial and temporal connections between adjacent superpixels. Then, we leverage Graph Convolutional Networks to process this representation and predict the desired output. As a result, we are able to train models with much fewer parameters, which translates into short training periods and a reduction in computation resource requirements. A comprehensive experimental study on the publicly available datasets Kinetics-400 and Charades shows that the proposed method is highly cost-effective and uses limited commodity hardware during training and inference. It reduces the computational requirements 10-fold while achieving results that are comparable to state-of-the-art methods. We believe that the proposed approach is a promising direction that could open the door to solving video understanding more efficiently and enable more resource limited users to thrive in this research field.



Paperid:1449
Authors:Amirhossein Habibian; Haitam Ben Yahia; Davide Abati; Efstratios Gavves; Fatih Porikli
Title: Delta Distillation for Efficient Video Processing
Abstract:
This paper aims to accelerate video stream processing, such as object detection and semantic segmentation, by leveraging the temporal redundancies that exist between video frames. Instead of relying on explicit motion alignment, such as optical flow warping, we propose a novel knowledge distillation schema coined as Delta Distillation. In our proposal, the student learns the variations in the teacher’s intermediate features over time. We demonstrate that these temporal variations can be effectively distilled due to the temporal redundancies within video frames. During inference, both teacher and student cooperate for providing predictions: the former by providing initial representations extracted only on the key-frame, and the latter by iteratively estimating and applying deltas for the successive frames. Moreover, we consider various design choices to learn optimal student architectures including an end-to-end learnable architecture search. By extensive experiments on a wide range of architectures, including the most efficient ones, we demonstrate that delta distillation sets a new state of the art in terms of accuracy vs. efficiency tradeoff for semantic segmentation and object detection in videos. Finally, we show that, as a by-product, delta distillation improves the temporal consistency of the teacher model.



Paperid:1450
Authors:David Junhao Zhang; Kunchang Li; Yali Wang; Yunpeng Chen; Shashwat Chandra; Yu Qiao; Luoqi Liu; Mike Zheng Shou
Title: MorphMLP: An Efficient MLP-Like Backbone for Spatial-Temporal Representation Learning
Abstract:
Recently, MLP-Like networks have been revived for image recognition. However, whether it is possible to build a generic MLP-Like architecture on video domain has not been explored, due to complex spatial-temporal modeling with large computation burden. To fill this gap, we present an efficient self-attention free backbone, namely MorphMLP, which flexibly leverages the concise Fully-Connected (FC) layer for video representation learning. Specifically, a MorphMLP block consists of two key layers in sequence, i.e., MorphFC.s and MorphFC.t, for spatial and temporal modeling respectively. MorphFC.s can effectively capture core semantics in each frame, by progressive token interaction along both height and width dimensions. Alternatively, MorphFC.t can adaptively learn long-term dependency over frames, by temporal token aggregation on each spatial location. With such multi-dimension and multi-scale factorization, our MorphMLP block can achieve a great accuracy-computation balance. Finally, we evaluate our MorphMLP on a number of popular video benchmarks. Compared with the recent state-of-the-art models, MorphMLP significantly reduces computation but with better accuracy, e.g., MorphMLP-S only uses 50% GFLOPs of VideoSwin-T but achieves 0.9% top-1 improvement on Kinetics400, under ImageNet1K pretraining. MorphMLP-B only uses 43% GFLOPs of MViT-B but achieves 2.4% top-1 improvement on SSV2, even though MorphMLP-B is pretrained on ImageNet1K while MViT-B is pretrained on Kinetics400. Moreover, our method adapted to the image domain outperforms previous SOTA MLP-Like architectures.



Paperid:1451
Authors:Honglu Zhou; Asim Kadav; Aviv Shamsian; Shijie Geng; Farley Lai; Long Zhao; Ting Liu; Mubbasir Kapadia; Hans Peter Graf
Title: COMPOSER: Compositional Reasoning of Group Activity in Videos with Keypoint-Only Modality
Abstract:
Group Activity Recognition detects the activity collectively performed by a group of actors, which requires compositional reasoning of actors and objects. We approach the task by modeling the video as tokens that represent the multi-scale semantic concepts in the video. We propose COMPOSER, a Multiscale Transformer based architecture that performs attention-based reasoning over tokens at each scale and learns group activity compositionally. In addition, prior works suffer from scene biases with privacy and ethical concerns. We only use the keypoint modality which reduces scene biases and prevents acquiring detailed visual data that may contain private or biased information of users. We improve the multiscale representations in COMPOSER by clustering the intermediate scale representations, while maintaining consistent cluster assignments between scales. Finally, we use techniques such as auxiliary prediction and data augmentations tailored to the keypoint signals to aid model training. We demonstrate the model’s strength and interpretability on two widely-used datasets (Volleyball and Collective Activity). COMPOSER achieves up to +5.4% improvement with just the keypoint modality.



Paperid:1452
Authors:Zizhang Li; Mengmeng Wang; Huaijin Pi; Kechun Xu; Jianbiao Mei; Yong Liu
Title: E-NeRV: Expedite Neural Video Representation with Disentangled Spatial-Temporal Context
Abstract:
Recently, the image-wise implicit neural representation of videos, NeRV, has gained popularity for its promising results and swift speed compared to regular pixel-wise implicit representations. However, the redundant parameters within the network structure can cause a large model size when scaling up for desirable performance. The key reason of this phenomenon is the coupled formulation of NeRV, which outputs the spatial and temporal information of video frames directly from the frame index input. In this paper, we propose E-NeRV, which dramatically expedites NeRV by decomposing the image-wise implicit neural representation into separate spatial and temporal context. Under the guidance of this new formulation, our model greatly reduces the redundant model parameters, while retaining the representation ability. We experimentally find that our method can improve the performance to a large extent with fewer parameters, resulting in a more than 8x faster speed on convergence. Code is available at https://github.com/kyleleey/E-NeRV.



Paperid:1453
Authors:Guanxiong Sun; Yang Hua; Guosheng Hu; Neil Robertson
Title: TDViT: Temporal Dilated Video Transformer for Dense Video Tasks
Abstract:
Deep video models, for example, 3D CNNs or video transformers, have achieved promising performance on sparse video tasks, i.e., predicting one result per video. However, challenges arise when adapting existing deep video models to dense video tasks, i.e., predicting one result per frame. Specifically, these models are expensive for deployment, less effective when handling redundant frames and difficult to capture long-range temporal correlations. To overcome these issues, we propose a Temporal Dilated Video Transformer (TDViT) that consists of carefully-designed temporal dilated transformer blocks (TDTB). TDTB can efficiently extract spatiotemporal representations and effectively alleviate the negative effect of temporal redundancy. Furthermore, by using hierarchical TDTBs, our approach obtains an exponentially expanded temporal receptive field and therefore can model long-range dynamics. Extensive experiments are conducted on two different dense video benchmarks, i.e., ImageNet VID for video object detection and YouTube VIS for video instance segmentation. Excellent experimental results demonstrate the superior efficiency, effectiveness, and compatibility of our method. The code is available at https://github.com/guanxiongsun/TDViT.



Paperid:1454
Authors:Woobin Im; Sebin Lee; Sung-Eui Yoon
Title: Semi-Supervised Learning of Optical Flow by Flow Supervisor
Abstract:
A training pipeline for optical flow CNNs consists of a pretraining stage on a synthetic dataset followed by a fine tuning stage on a target dataset. However, obtaining ground-truth flows from a target video requires a tremendous effort. This paper proposes a practical fine tuning method to adapt a pretrained model to a target dataset without ground truth flows, which has not been explored extensively. Specifically, we propose a flow supervisor for self-supervision, which consists of parameter separation and a student output connection. This design is aimed at stable convergence and better accuracy over conventional self-supervision methods which are unstable on the fine tuning task. Experimental results show the effectiveness of our method compared to different self-supervision methods for semi-supervised learning. In addition, we achieve meaningful improvements over state-of-the-art optical flow models on Sintel and KITTI benchmarks by exploiting additional unlabeled datasets. Code is available at https://github.com/iwbn/flow-supervisor.



Paperid:1455
Authors:Nikita Dvornik; Isma Hadji; Hai Pham; Dhaivat Bhatt; Brais Martinez; Afsaneh Fazly; Allan D. Jepson
Title: Flow Graph to Video Grounding for Weakly-Supervised Multi-step Localization
Abstract:
In this work, we consider the problem of weakly-supervised multi-step localization in instructional videos. An established approach to this problem is to rely on a given list of steps. However, in reality, there is often more than one way to execute a procedure successfully, by following the set of steps in slightly varying orders. Thus, for successful localization in a given video, recent works require the actual order of procedure steps in the video, to be provided by human annotators at both training and test times. Instead, here, we only rely on generic procedural text that is not tied to a specific video. We represent the various ways to complete the procedure by transforming the list of instructions into a procedure flow graph which captures the partial order of steps. Using the flow graphs reduces both training and test time annotation requirements. To this end, we introduce the new problem of flow graph to video grounding. In this setup, we seek the optimal step ordering consistent with the procedure flow graph and a given video. To solve this problem, we propose a new algorithm - Graph2Vid - that infers the actual ordering of steps in the video and simultaneously localizes them. To show the advantage of our proposed formulation, we extend the CrossTask dataset with procedure flow graph information. Our experiments show that Graph2Vid is both more efficient than the baselines and yields strong step localization results, without the need for step order annotation.



Paperid:1456
Authors:Yiheng Li; Connelly Barnes; Kun Huang; Fang-Lue Zhang
Title: Deep 360° Optical Flow Estimation Based on Multi-Projection Fusion
Abstract:
Optical flow computation is essential in the early stages of the video processing pipeline. This paper focuses on a less explored problem in this area, the 360° optical flow estimation using deep neural networks to support the increasingly popular VR applications. To address the distortions of panoramic representations when applying convolutional neural networks, we propose a novel multi-projection fusion framework that fuses the optical flow predicted by the models trained using different projection methods. It learns to combine the complementary information in the optical flow results under different projections. We also build the first large-scale panoramic optical flow dataset to support the training of neural networks and the evaluation of panoramic optical flow estimation methods. The experiment results on our dataset demonstrate that our method outperforms the existing methods and other alternative deep networks that were developed for processing 360° content.



Paperid:1457
Authors:Fanyi Xiao; Joseph Tighe; Davide Modolo
Title: MaCLR: Motion-Aware Contrastive Learning of Representations for Videos
Abstract:
We present MaCLR, a novel method to explicitly perform cross-modal self-supervised video representations learning from visual and motion modalities. Compared to previous video representation learn- ing methods that mostly focus on learning motion cues implicitly from RGB inputs, MaCLR enriches standard contrastive learning objectives for RGB video clips with a cross-modal learning objective between a Motion pathway and a Visual pathway. We show that the representation learned with our MaCLR method focuses more on foreground motion re- gions and thus generalizes better to downstream tasks. To demonstrate this, we evaluate MaCLR on five datasets for both action recognition and action detection, and demonstrate state-of-the-art self-supervised perfor- mance on all datasets. Furthermore, we show that MaCLR representa- tion can be as effective as representations learned with full supervision on UCF101 and HMDB51 action recognition, and even outperform the supervised representation for action recognition on VidSitu and SSv2, and action detection on AVA.



Paperid:1458
Authors:Kyle Min; Sourya Roy; Subarna Tripathi; Tanaya Guha; Somdeb Majumdar
Title: Learning Long-Term Spatial-Temporal Graphs for Active Speaker Detection
Abstract:
Active speaker detection (ASD) in videos with multiple speakers is a challenging task as it requires learning effective audiovisual features and spatial-temporal correlations over long temporal windows. In this paper, we present SPELL, a novel spatial-temporal graph learning framework that can solve complex tasks such as ASD. To this end, each person in a video frame is first encoded in a unique node for that frame. Nodes corresponding to a single person across frames are connected to encode their temporal dynamics. Nodes within a frame are also connected to encode inter-person relationships. Thus, SPELL reduces ASD to a node classification task. Importantly, SPELL is able to reason over long temporal contexts for all nodes without relying on computationally expensive fully connected graph neural networks. Through extensive experiments on the AVA-ActiveSpeaker dataset, we demonstrate that learning graph-based representations can significantly improve the active speaker detection performance owing to its explicit spatial and temporal structure. SPELL outperforms all previous state-of-the-art approaches while requiring significantly lower memory and computational resources. Our code is publicly available: https://github.com/SRA2/SPELL"



Paperid:1459
Authors:Ziyi Lin; Shijie Geng; Renrui Zhang; Peng Gao; Gerard de Melo; Xiaogang Wang; Jifeng Dai; Yu Qiao; Hongsheng Li
Title: Frozen CLIP Models Are Efficient Video Learners
Abstract:
Video recognition has been dominated by the end-to-end learning paradigm - first initializing a video recognition model with weights of a pretrained image model and then conducting end-to-end training on videos. This enables the video network to benefit from the pretrained image model. However, this requires substantial computation and memory resources for finetuning on videos and the alternative of directly using pretrained image features without finetuning the image backbone leads to subpar results. Fortunately, recent advances in Contrastive Vision-Language Pre-training (CLIP) pave the way for a new route for visual recognition tasks. Pretrained on large open-vocabulary image-text pair data, these models learn powerful visual representations with rich semantics. In this paper, we present Efficient Video Learning (EVL) - an efficient framework for directly training high-quality video recognition models with frozen CLIP features. Specifically, we employ a lightweight Transformer decoder and learn a query token to dynamically collect frame-level spatial features from the CLIP image encoder. Furthermore, we adopt a local temporal module in each decoder layer to discover temporal clues from adjacent frames and their attention maps. We show that despite being efficient to train with a frozen backbone, our models learn high quality video representations on a variety of video recognition datasets. Code is available at https://github.com/OpenGVLab/efficient-video-recognition.



Paperid:1460
Authors:Jiafei Duan; Samson Yu; Soujanya Poria; Bihan Wen; Cheston Tan
Title: PIP: Physical Interaction Prediction via Mental Simulation with Span Selection
Abstract:
Accurate prediction of physical interaction outcomes is a crucial component of human intelligence and is important for safe and efficient deployments of robots in the real world. While there are existing vision-based intuitive physics models that learn to predict physical interaction outcomes, they mostly focus on generating short sequences of future frames based on physical properties (e.g. mass, friction and velocity) extracted from visual inputs or a latent space. However, there is a lack of intuitive physics models that are tested on long physical interaction sequences with multiple interactions among different objects. We hypothesize that selective temporal attention during approximate mental simulations helps humans in physical interaction outcome prediction. With these motivations, we propose a novel scheme: \textbf{P}hysical \textbf{I}nteraction \textbf{P}rediction via Mental Simulation with Span Selection (PIP). It utilizes a deep generative model to model approximate mental simulations by generating future frames of physical interactions before employing selective temporal attention in the form of span selection for predicting physical interaction outcomes. To the best of our knowledge, attention has not been used with deep learning to tackle intuitive physics. For model evaluation, we further propose the large-scale SPACE+ dataset of synthetic videos with long sequences of three prime physical interactions in a 3D environment. Our experiments show that PIP outperforms human, baseline, and related intuitive physics models that utilize mental simulation. Furthermore, PIP’s span selection module effectively identifies the frames indicating key physical interactions among objects, allowing for added interpretability, and does not require labor-intensive frame annotations. PIP is available on https://sites.google.com/view/piphysics.



Paperid:1461
Authors:Heeseung Yun; Sehun Lee; Gunhee Kim
Title: Panoramic Vision Transformer for Saliency Detection in 360° Videos
Abstract:
360° video saliency detection is one of the challenging benchmarks for 360° video understanding since non-negligible distortion and discontinuity occur in the projection of any format of 360° videos, and capture-worthy viewpoint in the omnidirectional sphere is ambiguous by nature. We present a new framework named Panoramic Vision Transformer (PAVER). We design the encoder using Vision Transformer with deformable convolution, which enables us not only to plug pretrained models from normal videos into our architecture without additional modules or finetuning but also to perform geometric approximation only once, unlike previous deep CNN-based approaches. Thanks to its powerful encoder, PAVER can learn the saliency from three simple relative relations among local patch features, outperforming state-of-the-art models for the Wild360 benchmark by large margins without supervision or auxiliary information like class activation. We demonstrate the utility of our saliency prediction model with the omnidirectional video quality assessment task in VQA-ODV, where we consistently improve performance without any form of supervision, including head movement.



Paperid:1462
Authors:Aditi Basu Bal; Ramy Mounir; Sathyanarayanan Aakur; Sudeep Sarkar; Anuj Srivastava
Title: Bayesian Tracking of Video Graphs Using Joint Kalman Smoothing and Registration
Abstract:
Graph-based representations are becoming increasingly popular for representing and analyzing video data, especially in object tracking and scene understanding applications. Accordingly, an essential tool in this approach is to generate statistical inferences for graphical time series associated with videos. This paper develops a Kalman-smoothing method for estimating graphs from noisy, cluttered, and incomplete data. The main challenge here is to find and preserve the registration of nodes (salient detected objects) across time frames when the data has noise and clutter due to false and missing nodes. First, we introduce a quotient-space representation of graphs that incorporates temporal registration of nodes, and we use that metric structure to impose a dynamical model on graph evolution. Then, we derive a Kalman smoother, adapted to the quotient space geometry, to estimate dense, smooth trajectories of graphs. We demonstrate this framework using simulated data and actual video graphs extracted from the Multiview Extended Video with Activities (MEVA) dataset. This framework successfully estimates graphs despite the noise, clutter, and missed detections. Keywords: motion tracking, graph representations, video graphs, quotient metrics, Kalman smoothing, nonlinear manifolds.



Paperid:1463
Authors:Jingcheng Ni; Nan Zhou; Jie Qin; Qian Wu; Junqi Liu; Boxun Li; Di Huang
Title: Motion Sensitive Contrastive Learning for Self-Supervised Video Representation
Abstract:
Contrastive learning has shown great potential in video representation learning. However, existing approaches fail to sufficiently exploit short-term motion dynamics, which are crucial to various down-stream video understanding tasks. In this paper, we propose Motion Sensitive Contrastive Learning (MSCL) that injects the motion information captured by optical flows into RGB frames to strengthen feature learning. To achieve this, in addition to clip-level global contrastive learning, we develop Local Motion Contrastive Learning (LMCL) with frame-level contrastive objectives across the two modalities. Moreover, we introduce Flow Rotation Augmentation (FRA) to generate extra motion-shuffled negative samples and Motion Differential Sampling (MDS) to accurately screen training samples. Extensive experiments on standard benchmarks validate the effectiveness of the proposed method. With the commonly-used 3D ResNet-18 as the backbone, we achieve the top-1 accuracies of 91.5\% on UCF101 and 50.3\% on Something-Something v2 for video classification, and a 65.6\% Top-1 Recall on UCF101 for video retrieval, notably improving the state-of-the-art.



Paperid:1464
Authors:Fuchen Long; Zhaofan Qiu; Yingwei Pan; Ting Yao; Chong-Wah Ngo; Tao Mei
Title: Dynamic Temporal Filtering In Video Models
Abstract:
Video temporal dynamics is conventionally modeled with 3D spatial-temporal kernel or its factorized version comprised of 2D spatial kernel and 1D temporal kernel. The modeling power, nevertheless, is limited by the fixed window size and static weights of a kernel along the temporal dimension. The pre-determined kernel size severely limits the temporal receptive fields and the fixed weights treat each spatial location across frames equally, resulting in sub-optimal solution for long-range temporal modeling in natural scenes. In this paper, we present a new recipe of temporal feature learning, namely Dynamic Temporal Filter (DTF), that novelly performs spatial-aware temporal modeling in frequency domain with large temporal receptive field. Specifically, DTF dynamically learns a specialized frequency filter for every spatial location to model its long-range temporal dynamics. Meanwhile, the temporal feature of each spatial location is also transformed into frequency feature spectrum via 1D Fast Fourier Transform (FFT). The spectrum is modulated by the learnt frequency filter, and then transformed back to temporal domain with inverse FFT. In addition, to facilitate the learning of frequency filter in DTF, we perform frame-wise aggregation to enhance the primary temporal feature with its temporal neighbors by inter-frame correlation. It is feasible to plug DTF block into ConvNets and Transformer, yielding DTF-Net and DTF-Transformer. Extensive experiments conducted on three datasets demonstrate the superiority of our proposals. More remarkably, DTF-Transformer achieves an accuracy of 83.5% on Kinetics-400 dataset. Source code is available at \url{https://github.com/FuchenUSTC/DTF}.



Paperid:1465
Authors:Renrui Zhang; Wei Zhang; Rongyao Fang; Peng Gao; Kunchang Li; Jifeng Dai; Yu Qiao; Hongsheng Li
Title: Tip-Adapter: Training-Free Adaption of CLIP for Few-Shot Classification
Abstract:
Contrastive Vision-Language Pre-training, known as CLIP, has provided a new paradigm for learning visual representations using large-scale image-text pairs. It shows impressive performance on downstream tasks by zero-shot knowledge transfer. To further enhance CLIP’s adaption capability, existing methods proposed to fine-tune additional learnable modules, which significantly improves the few-shot performance but introduces extra training time and computational resources. In this paper, we propose a training-free adaption method for CLIP to conduct few-shot classification, termed as Tip-Adapter, which not only inherits the training-free advantage of zero-shot CLIP but also performs comparably to those training-required approaches. Tip-Adapter constructs the adapter via a key-value cache model from the few-shot training set, and updates the prior knowledge encoded in CLIP by feature retrieval. On top of that, the performance of Tip-Adapter can be further boosted to be state-of-the-art on ImageNet by fine-tuning the cache model for 10$\times$ fewer epochs than existing methods, which is both effective and efficient. We conduct extensive experiments of few-shot classification on 11 datasets to demonstrate the superiority of our proposed methods. Code is released at https://github.com/gaopengcuhk/Tip-Adapter.



Paperid:1466
Authors:Lianyu Hu; Liqing Gao; Zekang Liu; Wei Feng
Title: Temporal Lift Pooling for Continuous Sign Language Recognition
Abstract:
Pooling methods are necessities for modern neural networks for increasing receptive fields and lowering down computational costs. However, commonly used hand-crafted pooling approaches, e.g. max pooling and average pooling, may not well preserve discriminative features. While many researchers have elaborately designed various pooling variants in spatial domain to handle these limitations with much progress, the temporal aspect is rarely visited where directly applying hand-crafted methods or these specialized spatial variants may not be optimal. In this paper, we derive temporal lift pooling (TLP) from the Lifting Scheme in signal processing to intelligently downsample features of different temporal hierarchies. The Lifting Scheme factorizes input signals into various sub-bands with different frequency, which can be viewed as different temporal movement patterns. Our TLP is a three-stage procedure, which performs signal decomposition, component weighting and information fusion to generate a refined downsized feature map. We select a typical temporal task with long sequences, i.e. continuous sign language recognition (CSLR), as our testbed to verify the effectiveness of TLP. Experiments on two large-scale datasets show TLP outperforms hand-crafted methods and specialized spatial variants by a large margin (1.5%) with similar computational overhead. As a robust feature extractor, TLP exhibits great generalizability upon multiple backbones on various datasets and achieves new state-of-the-art results on two large-scale CSLR datasets. Visualizations further demonstrate the mechanism of TLP in correcting gloss borders.



Paperid:1467
Authors:Yang Jiao; Shaoxiang Chen; Zequn Jie; Jingjing Chen; Lin Ma; Yu-Gang Jiang
Title: MORE: Multi-Order RElation Mining for Dense Captioning in 3D Scenes
Abstract:
3D dense captioning is a recently-proposed novel task, where point clouds contain more geometric information than the 2D counterpart. However, it is also more challenging due to the higher complexity and wider variety of inter-object relations contained in point clouds. Existing methods only treat such relations as by-products of object feature learning in graphs without specifically encoding them, which leads to sub-optimal results. In this paper, aiming at improving 3D dense captioning via capturing and utilizing the complex relations in the 3D scene, we propose MORE, a Multi-Order RElation mining model, to support generating more descriptive and comprehensive captions. Technically, our MORE encodes object relations in a progressive manner since complex relations can be deduced from a limited number of basic ones. We first devise a novel Spatial Layout Graph Convolution (SLGC), which semantically encodes several first-order relations as edges of a graph constructed over 3D object proposals. Next, from the resulting graph, we further extract multiple triplets which encapsulate basic first-order relations as the basic unit, and construct several Object-centric Triplet Attention Graphs (OTAG) to infer multi-order relations for every target object. The updated node features from OTAG are aggregated and fed into the caption decoder to provide abundant relational cues, so that captions including diverse relations with context objects can be generated. Extensive experiments on the Scan2Cap dataset prove the effectiveness of our proposed MORE and its components, and we also outperform the current state-of-the-art method. Our code is available at https://github.com/SxJyJay/MORE.



Paperid:1468
Authors:Mengxue Qu; Yu Wu; Wu Liu; Qiqi Gong; Xiaodan Liang; Olga Russakovsky; Yao Zhao; Yunchao Wei
Title: SiRi: A Simple Selective Retraining Mechanism for Transformer-Based Visual Grounding
Abstract:
In this paper, we investigate how to achieve better referring visual grounding with modern vision-language transformers, and propose a simple yet powerful Selective Retraining (SiRi) mechanism. Particularly, SiRi conveys a significant principle to the research of visual grounding, i.e, a better initialized vision-language encoder would help the model converge to a better local minimum, advancing the performance accordingly. With such a principle, we continually update the parameters of the encoder as the training goes on, while periodically re-initialize the rest parameters to compel the model to be better optimized based on an enhanced encoder. With such a simple training mechanism, our SiRi can significantly outperform previous approaches on three popular benchmarks. Additionally, we reveal that SiRi performs surprisingly superior even with limited training data. More importantly, the effectiveness of SiRi, are further verified by other model and other V-L tasks. Code is available in the supplementary materials.



Paperid:1469
Authors:Jun Wang; Abhir Bhalerao; Yulan He
Title: Cross-Modal Prototype Driven Network for Radiology Report Generation
Abstract:
Radiology report generation (RRG) aims to describe automatically a radiology image with human-like language and could potentially support the work of radiologists, reducing the burden of manual reporting. Previous approaches often adopt an encoder-decoder architecture and focus on single-modal feature learning, while few studies explore cross-modal feature interaction. Here we propose a Cross-modal PROtotype driven NETwork (XPRONET) to promote cross-modal pattern learning and exploit it to improve the task of radiology report generation. This is achieved by three well-designed, fully differentiable and complementary modules: a shared cross-modal prototype matrix to record the cross-modal prototypes; a cross-modal prototype network to learn the cross-modal prototypes and embed the cross-modal information into the visual and textual features; and an improved multi-label contrastive loss to enable and enhance multi-label prototype learning. XPRONET obtains substantial improvements on the IU-Xray and MIMIC-CXR benchmarks, where its performance exceeds recent state-of-the-art approaches by a large margin on IU-Xray and comparable performance on MIMIC-CXR.



Paperid:1470
Authors:Chuan Guo; Xinxin Zuo; Sen Wang; Li Cheng
Title: TM2T: Stochastic and Tokenized Modeling for the Reciprocal Generation of 3D Human Motions and Texts
Abstract:
Inspired by the strong ties between vision and language, the two intimate human sensing and communication modalities, our paper aims to explore the generation of 3D human full-body motions from texts, as well as its reciprocal task, shorthanded for text2motion and motion2text, respectively. To tackle the existing challenges, especially to enable the generation of multiple distinct motions from the same text, and to avoid the undesirable production of trivial motionless pose sequences, we propose the use of motion token, a discrete and compact motion representation, where motions and texts could then be considered on one level playing ground, as the motion and text tokens. Moreover, our motion2text module is integrated into the inverse alignment process of our text2motion training pipeline, where a significant deviation of synthesized text (text2motion-2text) from the input text would be penalized by a large training loss; empirically this is shown to achieve improved performance. Finally, the mappings in-between the two modalities of motions and texts are facilitated by adapting the neural model for machine translation (NMT) to our context. Autoregressive modeling on the underlying distribution of discrete motion tokens further enables the production of non-deterministic motions from texts. Overall our approach is flexible, and could be used for both text2motion and motion2text tasks. Empirical evaluations on two benchmark datasets demonstrate the superior performance of our approach over a variety of state-of-the-art methods on both tasks.



Paperid:1471
Authors:Chaoyang Zhu; Yiyi Zhou; Yunhang Shen; Gen Luo; Xingjia Pan; Mingbao Lin; Chao Chen; Liujuan Cao; Xiaoshuai Sun; Rongrong Ji
Title: SeqTR: A Simple Yet Universal Network for Visual Grounding
Abstract:
In this paper, we propose a simple yet universal network termed SeqTR for visual grounding tasks, e.g., phrase localization, referring expression comprehension (REC) and segmentation (RES). The canonical paradigms for visual grounding often require substantial expertise in designing network architectures and loss functions, making them hard to generalize across tasks. To simplify and unify the modeling, we cast visual grounding as a point prediction problem conditioned on image and text inputs, where either the bounding box or binary mask is represented as a sequence of discrete coordinate tokens. Under this paradigm, visual grounding tasks are unified in our SeqTR network without task-specific branches or heads, e.g., the convolutional mask decoder for RES, which greatly reduces the complexity of multi-task modeling. In addition, SeqTR also shares the same optimization objective for all tasks with a simple cross-entropy loss, further reducing the complexity of deploying hand-crafted loss functions. Experiments on five benchmark datasets demonstrate that the proposed SeqTR outperforms (or is on par with) the existing state-of-the-arts, proving that a simple yet universal approach for visual grounding is indeed feasible. Source code is available at https://github.com/sean-zhuh/SeqTR.



Paperid:1472
Authors:Laura Hanu; James Thewlis; Yuki M. Asano; Christian Rupprecht
Title: VTC: Improving Video-Text Retrieval with User Comments
Abstract:
Multi-modal retrieval is an important problem for many applications, such as recommendation and search. Current benchmarks and even datasets are often manually constructed and consist of mostly clean samples where all modalities are well-correlated with the content. Thus, current video-text retrieval literature largely focuses on video titles or audio transcripts, while ignoring user comments, since users often tend to discuss topics only vaguely related to the video. Despite the ubiquity of user comments online, there is currently no multi-modal representation learning datasets that includes comments. In this paper, we a) introduce a new dataset of videos, titles and comments; b) present an attention-based mechanism that allows the model to learn from sometimes irrelevant data such as comments; c) show that by using comments, our method is able to learn better, more contextualised, representations for image, video and audio representations.



Paperid:1473
Authors:Xiao Han; Licheng Yu; Xiatian Zhu; Li Zhang; Yi-Zhe Song; Tao Xiang
Title: FashionViL: Fashion-Focused Vision-and-Language Representation Learning
Abstract:
Large-scale Vision-and-Language (V+L) pre-training for representation learning has proven to be effective in boosting various downstream V+L tasks. However, when it comes to the fashion domain, existing V+L methods are inadequate as they overlook the unique characteristics of both fashion V+L data and downstream tasks. In this work, we propose a novel fashion-focused V+L representation learning framework, dubbed as FashionViL. It contains two novel fashion-specific pre-training tasks designed particularly to exploit two intrinsic attributes with fashion V+L data. First, in contrast to other domains where a V+L datum contains only a single image-text pair, there could be multiple images in the fashion domain. We thus propose a Multi-View Contrastive Learning task for pulling closer the visual representation of one image to the compositional multimodal representation of another image+text. Second, fashion text (e.g., product description) often contains rich fine-grained concepts (attributes/noun phrases). To capitalize this, a Pseudo-Attributes Classification task is introduced to encourage the learned unimodal (visual/textual) representations of the same concept to be adjacent. Further, fashion V+L tasks uniquely include ones that do not conform to the common one-stream or two-stream architectures (e.g., text-guided image retrieval). We thus propose a flexible, versatile V+L model architecture consisting of a modality-agnostic Transformer so that it can be flexibly adapted to any downstream tasks. Extensive experiments show that our FashionViL achieves new state of the art across five downstream tasks. Code is available at https://github.com/BrandonHanx/mmf.



Paperid:1474
Authors:Aisha Urooj; Hilde Kuehne; Chuang Gan; Niels Da Vitoria Lobo; Mubarak Shah
Title: Weakly Supervised Grounding for VQA in Vision-Language Transformers
Abstract:
Transformers for visual-language representation learning have been getting a lot of interest and shown tremendous performance on visual question answering (VQA) and grounding. However, most systems that show good performance of those tasks still rely on pre-trained object detectors during training, which limits their applicability to the object classes available for those detectors. To mitigate this limitation, this paper focuses on the problem of weakly supervised grounding in the context of visual question answering in transformers. Our approach leverages capsules by transforming each visual token into a capsule representation in the visual encoder; it then uses activations from language self-attention layers as a text-guided selection module to mask those capsules before they are forwarded to the next layer. We evaluate our approach on the challenging GQA as well as VQA-HAT dataset for VQA grounding. Our experiments show that: while removing the information of masked objects from standard transformer architectures leads to a significant drop in performance, the integration of capsules significantly improves the grounding ability of such systems and provides new state-of-the-art results compared to other approaches in the field.



Paperid:1475
Authors:Liliane Momeni; Hannah Bull; K R Prajwal; Samuel Albanie; Gül Varol; Andrew Zisserman
Title: Automatic Dense Annotation of Large-Vocabulary Sign Language Videos
Abstract:
Recently, sign language researchers have turned to sign language interpreted TV broadcasts, comprising (i) a video of continuous signing and (ii) subtitles corresponding to the audio content, as a readily available and large-scale source of training data. One key challenge in the usability of such data is the lack of sign annotations. Previous work exploiting such weakly-aligned data only found sparse correspondences between keywords in the subtitle and individual signs. In this work, we propose a simple, scalable framework to vastly increase the density of automatic annotations. Our contributions are the following: (1) we significantly improve previous annotation methods by making use of synonyms and subtitle-signing alignment; (2) we show the value of pseudo-labelling from a sign recognition model as a way of sign spotting; (3) we propose a novel approach for increasing our annotations of known and unknown classes based on in domain exemplars; (4) on the BOBSL BSL sign language corpus, we increase the number of confident automatic annotations from 670K to 5M. We make these annotations publicly available to support the sign language research community.



Paperid:1476
Authors:Yuying Ge; Yixiao Ge; Xihui Liu; Jinpeng Wang; Jianping Wu; Ying Shan; Xiaohu Qie; Ping Luo
Title: MILES: Visual BERT Pre-training with Injected Language Semantics for Video-Text Retrieval
Abstract:
Dominant pre-training work for video-text retrieval mainly adopt the ""dual-encoder"" architectures to enable efficient retrieval, where two separate encoders are used to contrast global video and text representations, but ignore detailed local semantics. The recent success of image BERT pre-training with masked visual modeling that promotes the learning of local visual context, motivates a possible solution to address the above limitation. In this work, we for the first time investigate masked visual modeling in video-text pre-training with the ""dual-encoder"" architecture. We perform Masked visual modeling with Injected LanguagE Semantics (MILES) by employing an extra snapshot video encoder as an evolving ""tokenizer"" to produce reconstruction targets for masked video patch prediction. Given the corrupted video, the video encoder is trained to recover text-aligned features of the masked patches via reasoning with the visible regions along the spatial and temporal dimensions, which enhances the discriminativeness of local visual features and the fine-grained cross-modality alignment. Our method outperforms state-of-the-art methods for text-to-video retrieval on four datasets with both zero-shot and fine-tune evaluation protocols. Our approach also surpasses the baseline models significantly on zero-shot action recognition, which can be cast as video-to-text retrieval.



Paperid:1477
Authors:Yuxuan Wang; Difei Gao; Licheng Yu; Weixian Lei; Matt Feiszli; Mike Zheng Shou
Title: "GEB+: A Benchmark for Generic Event Boundary Captioning, Grounding and Retrieval"
Abstract:
Cognitive science has shown that humans perceive videos in terms of events separated by the state changes of dominant subjects. State changes trigger new events and are one of the most useful among the large amount of redundant information perceived. However, previous research focuses on the overall understanding of segments without evaluating the fine-grained status changes inside. In this paper, we introduce a new dataset called Kinetic-GEB+. The dataset consists of over 170k boundaries associated with captions describing status changes in the generic events in 12K videos. Upon this new dataset, we propose three tasks supporting the development of a more fine-grained, robust, and human-like understanding of videos through status changes. We evaluate many representative baselines in our dataset, where we also design a new TPD (Temporal-based Pairwise Difference) Modeling method for visual difference and achieve significant performance improvements. Besides, the results show there are still formidable challenges for current methods in the utilization of different granularities, representation of visual difference, and the accurate localization of status changes. Further analysis shows that our dataset can drive developing more powerful methods to understand status changes and thus improve video level comprehension. The dataset is available at https://github.com/Yuxuan-W/GEB-Plus"



Paperid:1478
Authors:Wei Suo; Mengyang Sun; Kai Niu; Yiqi Gao; Peng Wang; Yanning Zhang; Qi Wu
Title: A Simple and Robust Correlation Filtering Method for Text-Based Person Search
Abstract:
Text-based person search aims to associate pedestrian images with natural language descriptions. In this task, extracting differentiated representations and aligning them among identities and descriptions is an essential yet challenging problem. Most of the previous methods depend on additional language parsers or vision techniques to select the relevant regions or words from noise inputs. But there exists heavy computation cost and inevitable error accumulation. Meanwhile, simply using horizontal segmentation images to obtain local-level features would harm the reliability of models as well. In this paper, we present a novel end-to-end Simple and Robust Correlation Filtering (SRCF) method which can effectively extract key clues and adaptively align the discriminative features. Different from previous works, our framework focuses on computing the similarity between templates and inputs. In particular, we design two different types of filtering modules (i.e., denoising filters and dictionary filters) to extract crucial features and establish multi-modal mappings. Extensive experiments have shown that our method improves the robustness of the model and achieves better performance on the two text-based person search datasets. Source code is available at https://github.com/Suo-Wei/SRCF.



Paperid:1479
Authors:Benedikt Boecking; Naoto Usuyama; Shruthi Bannur; Daniel C. Castro; Anton Schwaighofer; Stephanie Hyland; Maria Wetscherek; Tristan Naumann; Aditya Nori; Javier Alvarez-Valle; Hoifung Poon; Ozan Oktay
Title: Making the Most of Text Semantics to Improve Biomedical Vision-Language Processing
Abstract:
Multi-modal data abounds in biomedicine, such as radiology images and reports. Interpreting this data at scale is essential for improving clinical care and accelerating clinical research. Biomedical text with its complex semantics poses additional challenges in vision-language modelling compared to the general domain, and previous work has used insufficiently adapted models that lack domain-specific language understanding. In this paper, we show that principled textual semantic modelling can substantially improve contrastive learning in self-supervised vision-language processing. We release a language model that achieves state-of-the-art results in radiology natural-language inference through its improved vocabulary and novel language pretraining objective leveraging semantics and discourse characteristics in radiology reports. Further, we propose a self-supervised joint vision-language approach with a focus on better text modelling. It establishes new state of the art results on a wide range of publicly available benchmarks, in part by leveraging our new domain-specific language model. We release a new dataset with locally-aligned phrase grounding annotations by radiologists to facilitate the study of complex semantic modelling in biomedical vision-language processing. A broad evaluation, including on this new dataset, shows that our contrastive learning approach, aided by textual-semantic modelling, outperforms prior methods in segmentation tasks, despite only using a global-alignment objective.



Paperid:1480
Authors:Shipeng Yan; Lanqing Hong; Hang Xu; Jianhua Han; Tinne Tuytelaars; Zhenguo Li; Xuming He
Title: Generative Negative Text Replay for Continual Vision-Language Pretraining
Abstract:
Vision-language pre-training (VLP) has attracted increasing attention recently. With a large amount of image-text pairs, VLP models trained with contrastive loss have achieved impressive performance in various tasks, especially the zero-shot generalization on downstream datasets. In practical applications, however, massive data are usually collected in a streaming fashion, requiring VLP models to continuously integrate novel knowledge from incoming data. In this work, we focus on learning a VLP model with sequential data chunks of image-text pairs. To tackle the catastrophic forgetting issue in this multi-modal continual learning setting, we first introduce pseudo text replay that generates hard negative texts conditioned on the training images in memory, which not only preserves learned knowledge but also improves the diversity of negative samples in the contrastive loss. Moreover, we propose multi-modal knowledge distillation between images and texts to align the instance-wise prediction between models. We incrementally pre-train our model on the both instance and class incremental splits of Conceptual Caption dataset, and evaluate the model on zero-shot image classification and image-text retrieval tasks. Our method consistently outperforms the existing baselines with a large margin, which demonstrates its superiority. Notably, we realize an average performance boost of $4.60\%$ on image-classification downstream datasets for class incremental split.



Paperid:1481
Authors:Junbin Xiao; Pan Zhou; Tat-Seng Chua; Shuicheng Yan
Title: Video Graph Transformer for Video Question Answering
Abstract:
This paper proposes a Video Graph Transformer (VGT) model for Video Quetion Answering (VideoQA). VGT’s uniqueness are two-fold: 1) it designs a dynamic graph transformer module which encodes video by explicitly capturing the visual objects, their relations, and dynamics for complex spatio-temporal reasoning; and 2) it exploits disentangled video and text Transformers for relevance comparison between the video and text to perform QA, instead of entangled cross-modal Transformer for answer classification. Vision-text communication is done by additional cross-modal interaction modules. With more reasonable video encoding and QA solution, we show that VGT can achieve much better performances on VideoQA tasks that challenge dynamic relation reasoning than prior arts in the pretraining-free scenario. Its performances even surpass those models that are pretrained with millions of external data. We further show that VGT can also benefit a lot from self-supervised cross-modal pretraining, yet with orders of magnitude smaller data. These results clearly demonstrate the effectiveness and superiority of VGT, and reveal its potential for more data-efficient pretraining. With comprehensive analyses and some heuristic observations, we hope that VGT can promote VQA research beyond coarse recognition/description towards fine-grained relation reasoning in realistic videos.



Paperid:1482
Authors:Kun Yan; Lei Ji; Chenfei Wu; Jianmin Bao; Ming Zhou; Nan Duan; Shuai Ma
Title: Trace Controlled Text to Image Generation
Abstract:
Text to Image generation is a fundamental and inevitable challenging task for visual linguistic modeling. The recent surge in this area such as DALL·E has shown breathtaking technical breakthroughs, however, it still lacks a precise control of the spatial relation corresponding to semantic text. To tackle this problem, mouse trace paired with text provides an interactive way, in which users can describe the imagined image with natural language while drawing traces to locate those they want. However, this brings the challenges of both controllability and compositionality of the generation. Motivated by this, we propose a Trace Controlled Text to Image Generation model (TCTIG), which takes trace as a bridge between semantic concepts and spatial conditions. Moreover, we propose a set of new technique to enhance the controllability and compositionality of generation, including trace guided re-weighting loss (TGR) and semantic aligned augmentation (SAA). In addition, we establish a solid benchmark for the trace-controlled text-to-image generation task, and introduce several new metrics to evaluate both the controllability and compositionality of the model. Upon that, we demonstrate TCTIG’s superior performance and further present the fruitful qualitative analysis of our model.



Paperid:1483
Authors:AJ Piergiovanni; Kairo Morton; Weicheng Kuo; Michael S. Ryoo; Anelia Angelova
Title: Video Question Answering with Iterative Video-Text Co-Tokenization
Abstract:
Video question answering is a challenging task that requires understanding jointly the language input, the visual information in individual video frames, as well as the temporal information about the events occurring in the video. In this paper, we propose a novel multi-stream video encoder for video question answering that uses multiple video inputs and a new video-text iterative co-tokenization approach to answer a variety of questions related to videos. We experimentally evaluate the model on several datasets, such as MSRVTT-QA, MSVD-QA, IVQA, outperforming the previous state-of-the-art by large margins. Simultaneously, our model reduces the required GFLOPs from 150-360 to only 67, producing a highly efficient video question answering model.



Paperid:1484
Authors:Long Chen; Yuhang Zheng; Jun Xiao
Title: Rethinking Data Augmentation for Robust Visual Question Answering
Abstract:
Data Augmentation (DA) --- generating extra training samples beyond original training set --- has been widely-used in today’s unbiased VQA models to mitigate the language biases. Current mainstream DA strategies are synthetic-based methods, which synthesize new samples by either editing some visual regions/words, or re-generating them from scratch. However, these synthetic samples are always unnatural and error-prone. To avoid this issue, a recent DA work composes new augmented samples by randomly pairing pristine images and other human-written questions. Unfortunately, to guarantee augmented samples have reasonable ground-truth answers, they manually design a set of heuristic rules for several question types, which extremely limits its generalization abilities. To this end, we propose a new Knowledge Distillation based Data Augmentation for VQA, dubbed KDDAug. Specifically, we first relax the requirements of reasonable image-question pairs, which can be easily applied to any question types. Then, we design a knowledge distillation (KD) based answer assignment to generate pseudo answers for all composed image-question pairs, which are robust to both in-domain and out-of-distribution settings. Since KDDAug is a model-agnostic DA strategy, it can be seamlessly incorporated into any VQA architectures. Extensive ablation studies on multiple backbones and benchmarks have demonstrated the effectiveness and generalization abilities of KDDAug.



Paperid:1485
Authors:Zhen Wang; Long Chen; Wenbo Ma; Guangxing Han; Yulei Niu; Jian Shao; Jun Xiao
Title: Explicit Image Caption Editing
Abstract:
Given an image and a reference caption, the image caption editing task aims to correct the misalignment errors and generate a refined caption. However, all existing caption editing works are implicit models, ie, they directly produce the refined captions without explicit connections to the reference captions. In this paper, we introduce a new task: Explicit Caption Editing (ECE). ECE models explicitly generate a sequence of edit operations, and this edit operation sequence can translate the reference caption into a refined one. Compared to the implicit editing, ECE has multiple advantages: 1) Explainable: it can trace the whole editing path. 2) Editing Efficient: it only needs to modify a few words. 3) Human-like: it resembles the way that humans perform caption editing, and tries to keep original sentence structures. To solve this new task, we propose the first ECE model: TIger. TIger is a non-autoregressive transformer-based model, consisting of three modules: Tagger_del, Tagger_add, and Inserter. Specifically, Tagger_del decides whether each word should be preserved or not, Tagger_add decides where to add new words, and Inserter predicts the specific word for adding. To further facilitate ECE research, we propose two new ECE benchmarks by re-organizing two existing datasets, dubbed COCO-EE and Flickr30K-EE, respectively. Extensive ablations on both two benchmarks have demonstrated the effectiveness of TIger.



Paperid:1486
Authors:Jiachang Hao; Haifeng Sun; Pengfei Ren; Jingyu Wang; Qi Qi; Jianxin Liao
Title: Can Shuffling Video Benefit Temporal Bias Problem: A Novel Training Framework for Temporal Grounding
Abstract:
Temporal grounding aims to locate a target video moment that semantically corresponds to the given sentence query in an untrimmed video. However, recent works find that existing methods suffer a severe temporal bias problem. These methods do not reason the target moment locations based on the visual-textual semantic alignment but over-rely on the temporal biases of queries in training sets. To this end, this paper proposes a novel training framework for grounding models to use shuffled videos to address temporal bias problem without losing grounding accuracy. Our framework introduces two auxiliary tasks, cross-modal matching and temporal order discrimination, to promote the grounding model training. The cross-modal matching task leverages the content consistency between shuffled and original videos to force the grounding model to mine visual contents to semantically match queries. The temporal order discrimination task leverages the difference in temporal order to strengthen the understanding of long-term temporal contexts. Extensive experiments on Charades-STA and ActivityNet Captions demonstrate the effectiveness of our method for mitigating the reliance on temporal biases and strengthening the model’s generalization ability against the different temporal distributions. Code is available at https://github.com/haojc/ShufflingVideosForTSG.



Paperid:1487
Authors:Spencer Whitehead; Suzanne Petryk; Vedaad Shakib; Joseph Gonzalez; Trevor Darrell; Anna Rohrbach; Marcus Rohrbach
Title: Reliable Visual Question Answering: Abstain Rather Than Answer Incorrectly
Abstract:
Machine learning has advanced dramatically, narrowing the accuracy gap to humans in multimodal tasks like visual question answering (VQA). However, while humans can say ""I don’t know"" when they are uncertain (i.e., abstain from answering a question), such ability has been largely neglected in multimodal research, despite the importance of this problem to the usage of VQA in real settings. In this work, we promote a problem formulation for reliable VQA, where we prefer abstention over providing an incorrect answer. We first enable abstention capabilities for several VQA models, and analyze both their coverage, the portion of questions answered, and risk, the error on that portion. For that, we explore several abstention approaches. We find that although the best performing models achieve over 71% accuracy on the VQA v2 dataset, introducing the option to abstain by directly using a model’s softmax scores limits them to answering less than 8% of the questions to achieve a low risk of error (i.e., 1%). This motivates us to utilize a multimodal selection function to directly estimate the correctness of the predicted answers, which we show can increase the coverage by, for example, 2.4x from 6.8% to 16.3% at 1% risk. While it is important to analyze both coverage and risk, these metrics have a trade-off which makes comparing VQA models challenging. To address this, we also propose an Effective Reliability metric for VQA that places a larger cost on incorrect answers compared to abstentions. This new problem formulation, metric, and analysis for VQA provide the groundwork for building effective and reliable VQA models that have the self-awareness to abstain if and only if they don’t know the answer.



Paperid:1488
Authors:Van-Quang Nguyen; Masanori Suganuma; Takayuki Okatani
Title: GRIT: Faster and Better Image Captioning Transformer Using Dual Visual Features
Abstract:
Current state-of-the-art methods for image captioning employ region-based features, as they provide object-level information that is essential to describe the content of images; they are usually extracted by an object detector such as Faster R-CNN. However, they have several issues, such as lack of contextual information, the risk of inaccurate detection, and the high computational cost. The first two could be resolved by additionally using grid-based features. However, how to extract and fuse these two types of features is uncharted. This paper proposes a Transformer-only neural architecture, dubbed GRIT (Grid- and Region-based Image captioning Transformer), that effectively utilizes the two visual features to generate better captions. GRIT replaces the CNN-based detector employed in previous methods with a DETR-based one, making it computationally faster. Moreover, its monolithic design consisting only of Transformers enables end-to-end training of the model. This innovative design and the integration of the dual visual features bring about significant performance improvement. The experimental results on several image captioning benchmarks show that GRIT outperforms previous methods in inference accuracy and speed.



Paperid:1489
Authors:Sunjae Yoon; Ji Woo Hong; Eunseop Yoon; Dahyun Kim; Junyeong Kim; Hee Suk Yoon; Chang D. Yoo
Title: Selective Query-Guided Debiasing for Video Corpus Moment Retrieval
Abstract:
Video moment retrieval (VMR) aims to localize target moments in untrimmed videos pertinent to a given textual query. Existing retrieval systems tend to rely on retrieval bias as a shortcut and thus, fail to sufficiently learn multi-modal interactions between query and video. This retrieval bias stems from learning frequent co-occurrence patterns between query and moments, which spuriously correlate objects (e.g., a pencil) referred in the query with moments (e.g., scene of writing with a pencil) where the objects frequently appear in the video, such that they converge into biased moment predictions. Although recent debiasing methods have focused on removing this retrieval bias, we argue that these biased predictions sometimes should be preserved because there are many queries where biased predictions are rather helpful. To conjugate this retrieval bias, we propose a Selective Query-guided Debiasing network (SQuiDNet), which incorporates the following two main properties: (1) Biased Moment Retrieval that intentionally uncovers the biased moments inherent in objects of the query and (2) Selective Query-guided Debiasing that performs selective debiasing guided by the meaning of the query. Our experimental results on three moment retrieval benchmarks (i.e., TVR, ActivityNet, DiDeMo) show the effectiveness of SQuiDNet and qualitative analysis shows improved interpretability.



Paperid:1490
Authors:Cheng Shi; Sibei Yang
Title: Spatial and Visual Perspective-Taking via View Rotation and Relation Reasoning for Embodied Reference Understanding
Abstract:
Embodied Reference Understanding studies the reference understanding in an embodied fashion, where a receiver requires to locate a target object referred to by both language and gesture of the sender in a shared physical environment. Its main challenge lies in how to make the receiver with the egocentric view access spatial and visual information relative to the sender to judge how objects are oriented around and seen from the sender, i.e., spatial and visual perspective-taking. In this paper, we propose a REasoning from your Perspective (REP) method to tackle the challenge by modeling relations between the receiver and the sender as well as the sender and the objects via the proposed novel view rotation and relation reasoning. Specifically, view rotation first rotates the receiver to the position of the sender by constructing an embodied 3D coordinate system with the position of the sender as the origin. Then, it changes the orientation of the receiver to the orientation of the sender by encoding the body orientation and gesture of the sender. Relation reasoning models both the nonverbal and verbal relations between the sender and the objects by multi-modal cooperative reasoning in gesture, language, visual content, and spatial position. Experiment results demonstrate the effectiveness of REP, which consistently surpasses all existing state-of-the-art algorithms by a large margin, i.e., +5.22% absolute accuracy in terms of Prec@0.5 on YouRefIt. Code is available at https://github.com/ChengShiest/REP-ERU"



Paperid:1491
Authors:Zihang Meng; David Yang; Xuefei Cao; Ashish Shah; Ser-Nam Lim
Title: Object-Centric Unsupervised Image Captioning
Abstract:
Image captioning is a longstanding problem in the field of computer vision and natural language processing. To date, researchers have produced impressive state-of-the-art performance in the age of deep learning. Most of these state-of-the-art, however, requires large volume of annotated image-caption pairs in order to train their models. When given an image dataset of interests, practitioner needs to annotate the caption for each image in the training set and this process needs to happen for each newly collected image dataset. In this paper, we explore the task of unsupervised image captioning which utilizes unpaired images and texts to train the model so that the texts can come from different sources than the images. A main school of research on this topic that has been shown to be effective is to construct pairs from the images and texts in the training set according to their overlap of objects. Unlike in the supervised setting, these constructed pairings are however not guaranteed to have fully overlapping set of objects. Our work in this paper overcomes this by harvesting objects corresponding to a given sentence from the training set, even if they don’t belong to the same image. When used as input to a transformer, such mixture of objects enable larger if not full object coverage, and when supervised by the corresponding sentence, produced results that outperform current state of the art unsupervised methods by a significant margin. Building upon this finding, we further show that (1) additional information on relationship between objects and attributes of objects also helps in boosting performance; and (2) our method also extends well to non-English image captioning, which usually suffers from a scarcer level of annotations. Our findings are supported by strong empirical results. Our code is available at \href{https://github.com/zihangm/obj-centric-unsup-caption}{https://github.com/zihangm/obj-centric-unsup-caption}"



Paperid:1492
Authors:Quan Cui; Boyan Zhou; Yu Guo; Weidong Yin; Hao Wu; Osamu Yoshie; Yubo Chen
Title: Contrastive Vision-Language Pre-training with Limited Resources
Abstract:
Pioneering dual-encoder pre-training works (e.g., CLIP and ALIGN) have revealed the potential of aligning multi-modal representations with contrastive learning. However, these works require a tremendous amount of data and computational resources (e.g., billion-level web data and hundreds of GPUs), which prevent researchers with limited resources from reproduction and further exploration. To this end, we propose a stack of novel methods, which significantly cut down the heavy resource dependency and allow us to conduct dual-encoder multi-modal representation alignment with limited resources. Besides, we provide a reproducible baseline of competitive results, namely ZeroVL, with only 14M publicly accessible academic datasets and 8 V100 GPUs. Additionally, we collect 100M web data for pre-training, and achieve comparable or superior results than state-of-the-art methods, further proving the effectiveness of our methods on large-scale data. We hope that this work will provide useful data points and experience for future research in contrastive vision-language pre-training. Code is available at https://github.com/zerovl/ZeroVL.



Paperid:1493
Authors:Sheng Fang; Shuhui Wang; Junbao Zhuo; Xinzhe Han; Qingming Huang
Title: Learning Linguistic Association towards Efficient Text-Video Retrieval
Abstract:
Text-video retrieval attracts growing attention recently. A dominant approach is to learn a common space for aligning two modalities. However, video deliver richer content than text in general situations and captions usually miss certain events or details in the video. The information imbalance between two modalities makes it difficult to align their representations. In this paper, we propose a general framework, LINguistic ASsociation (LINAS), which utilizes the complementarity between captions corresponding to the same video. Concretely, we first train a teacher model taking extra relevant captions as inputs, which can aggregate language semantics for obtaining more comprehensive text representations. Since the additional captions are inaccessible during inference, Knowledge Distillation is employed to train a student model with a single caption as input. We further propose Adaptive Distillation strategy, which allows the student model to adaptively learn the knowledge from the teacher model. This strategy also suppresses the spurious relations introduced during the linguistic association. Extensive experiments demonstrate the effectiveness and efficiency of LINAS with various baseline architectures on benchmark datasets.



Paperid:1494
Authors:Zanming Huang; Zhongkai Shangguan; Jimuyang Zhang; Gilad Bar; Matthew Boyd; Eshed Ohn-Bar
Title: ASSISTER: Assistive Navigation via Conditional Instruction Generation
Abstract:
We introduce a novel vision-and-language navigation (VLN) task of learning to provide real-time guidance to a blind follower situated in complex dynamic navigation scenarios. Towards exploring real-time information needs and fundamental challenges in our novel modeling task, we first collect a multi-modal real-world benchmark with in-situ Orientation and Mobility (O&M) instructional guidance. Subsequently, we leverage the real-world study to inform the design of a larger-scale simulation benchmark, thus enabling comprehensive analysis of limitations in current VLN models. Motivated by how sighted O&M guides seamlessly and safely support the awareness of individuals with visual impairments when collaborating on navigation tasks, we present ASSISTER, an imitation-learned agent that can embody such effective guidance. The proposed assistive VLN agent is conditioned on navigational goals and commands for generating instructional sentences that are coherent with the surrounding visual scene, while also carefully accounting for the immediate assistive navigation task. Altogether, our introduced evaluation and training framework takes a step towards scalable development of the next generation of seamless, human-like assistive agents.



Paperid:1495
Authors:Zhaowei Cai; Gukyeong Kwon; Avinash Ravichandran; Erhan Bas; Zhuowen Tu; Rahul Bhotika; Stefano Soatto
Title: X-DETR: A Versatile Architecture for Instance-Wise Vision-Language Tasks
Abstract:
In this paper, we study the challenging instance-wise vision-language tasks, where the free-form language is required to align with the objects instead of the whole image. To address these tasks, we propose X-DETR, whose architecture has three major components: an object detector, a language encoder, and vision-language alignment. The vision and language streams are independent until the end and they are aligned using an efficient dot-product operation. The whole network is trained end-to-end, such that the detector is optimized for the vision-language tasks instead of an off-the-shelf component. To overcome the limited size of paired object-language annotations, we leverage other weak types of supervision to expand the knowledge coverage. This simple yet effective architecture of X-DETR shows good accuracy and fast speeds for multiple instance-wise vision-language tasks, e.g., 16.4 AP on LVIS detection of 1.2K categories at 20 frames per second without using any LVIS annotation during training. The code is available at https://github.com/amazon-research/cross-modal-detr.



Paperid:1496
Authors:Wenhao Cheng; Xingping Dong; Salman Khan; Jianbing Shen
Title: Learning Disentanglement with Decoupled Labels for Vision-Language Navigation
Abstract:
Vision-and-Language Navigation (VLN) requires an agent to follow complex natural language instructions and perceive the visual environment for real-world navigation. Intuitively, we find that instruction disentanglement for each viewpoint along the agent’s path is critical for accurate navigation. However, most methods only utilize the whole complex instruction or inaccurate sub-instructions due to the lack of accurate disentanglement as an intermediate supervision stage. To address this problem, we propose a new Disentanglement framework with Decoupled Labels (DDL) for VLN. Firstly, we manually extend the benchmark dataset Room-to-Room with landmark- and action-aware labels in order to provide fine-grained information for each viewpoint. Furthermore, to enhance the generalization ability, we propose a Decoupled Label Speaker module to generate pseudo-labels for augmented data and reinforcement training. To fully use the proposed fine-grained labels, we design a Disentangled Decoding Module to guide discriminative feature extraction and help alignment of multi-modalities. To reveal the generality of our proposed method, we apply it on a LSTM-based model and two recent Transformer-based models. Extensive experiments on two VLN benchmarks (i.e., R2R and R4R) demonstrate the effectiveness of our approach, achieving better performance than previous state-of-the-art methods.



Paperid:1497
Authors:Qingpei Guo; Kaisheng Yao; Wei Chu
Title: Switch-BERT: Learning to Model Multimodal Interactions by Switching Attention and Input
Abstract:
The ability to model intra-modal and inter-modal interactions is fundamental in multimodal machine learning. The current state-of-the-art models usually adopt deep learning models with fixed structures. They can achieve exceptional performances on specific tasks, but face a particularly challenging problem of modality mismatch because of diversity of input modalities and their fixed structures. In this paper, we present Switch-BERT for joint vision and language representation learning to address this problem. Switch-BERT extends BERT architecture by introducing learnable layer-wise and cross-layer interactions. It learns to optimize attention from a set of attention modes representing these interactions. One specific property of the model is that it learns to attend outputs from various depths, therefore mitigates the modality mismatch problem. We present extensive experiments on visual question answering, image-text retrieval and referring expression comprehension experiments. Results confirm that, whereas alternative architectures including ViLBERT and UNITER may excel in particular tasks, Switch-BERT can consistently achieve better or comparable performances than the current state-of-the-art models in these tasks. Ablation studies indicate that the proposed model achieves superior performances due to its ability in learning task-specific multimodal interactions.



Paperid:1498
Authors:Bowen Li
Title: Word-Level Fine-Grained Story Visualization
Abstract:
Story visualization aims to generate a sequence of images to narrate each sentence in a multi-sentence story with a global consistency across dynamic scenes and characters. Current works still struggle with output images’ quality and consistency, and rely on additional semantic information or auxiliary captioning networks. To address these challenges, we first introduce a new sentence representation, which incorporates word information from all story sentences to mitigate the inconsistency problem. Then, we propose a new discriminator with fusion features and further extend the spatial attention to improve image quality and story consistency. Extensive experiments on different datasets and human evaluation demonstrate the superior performance of our approach, compared to state-of-the-art methods, neither using segmentation masks nor auxiliary captioning networks.



Paperid:1499
Authors:Qi Zhang; Yuqing Song; Qin Jin
Title: Unifying Event Detection and Captioning as Sequence Generation via Pre-training
Abstract:
Dense video captioning aims to generate corresponding text descriptions for a series of events in the untrimmed video, which can be divided into two sub-tasks, event detection and event captioning. Unlike previous works that tackle the two sub-tasks separately, recent works have focused on enhancing the inter-task association between the two sub-tasks. However, designing inter-task interactions for event detection and captioning is not trivial due to the large differences in their task specific solutions. Besides, previous event detection methods normally ignore temporal dependencies between events, leading to event redundancy or inconsistency problems. To tackle above the two defects, in this paper, we define event detection as a sequence generation task and propose a unified pre-training and fine-tuning framework to naturally enhance the inter-task association between event detection and captioning. Since the model predicts each event with previous events as context, the inter-dependency between events is fully exploited and thus our model can detect more diverse and consistent events in the video. Experiments on the ActivityNet dataset show that our model outperforms the state-of-the-art methods, and can be further boosted when pre-trained on extra large-scale video-text data. Code is available at https://github.com/QiQAng/UEDVC.



Paperid:1500
Authors:Chuang Lin; Yi Jiang; Jianfei Cai; Lizhen Qu; Gholamreza Haffari; Zehuan Yuan
Title: Multimodal Transformer with Variable-Length Memory for Vision-and-Language Navigation
Abstract:
Vision-and-Language Navigation (VLN) is a task that an agent is required to follow a language instruction to navigate to the goal position, which relies on the ongoing interactions with the environment during moving. Recent Transformer-based VLN methods have made great progress benefiting from the direct connections between visual observations and language instructions via the multimodal cross-attention mechanism. However, these methods usually represent temporal context as a fixed-length vector by using an LSTM decoder or using manually designed hidden states to build a recurrent Transformer. Considering a single fixed-length vector is often insufficient to capture long-term temporal context, in this paper, we introduce Multimodal Transformer with Variable-length Memory (MTVM) for visually-grounded natural language navigation by modeling the temporal context explicitly. Specifically, MTVM enables the agent to keep track of the navigation trajectory by directly storing activations in the previous time step in a memory bank. To further boost the performance, we propose a memory-aware consistency loss to help learn a better joint representation of temporal context with random masked instructions. We evaluate MTVM on popular R2R and CVDN datasets. Our model improves Success Rate on R2R test set by 2% and reduces Goal Process by 1.5m on CVDN test set.



Paperid:1501
Authors:Christopher Thomas; Yipeng Zhang; Shih-Fu Chang
Title: Fine-Grained Visual Entailment
Abstract:
Visual entailment is a recently proposed multimodal reasoning task where the goal is to predict the logical relationship of a piece of text to an image. In this paper, we propose an extension of this task, where the goal is to predict the logical relationship of fine-grained knowledge elements within a piece of text to an image. Unlike prior work, our method is inherently explainable and makes logical predictions at different levels of granularity. Because we lack fine-grained labels to train our method, we propose a novel multi-instance learning approach which learns a fine-grained labeling using only sample-level supervision. We also impose novel semantic structural constraints which ensure that fine-grained predictions are internally semantically consistent. We evaluate our method on a new dataset of manually annotated knowledge elements and show that our method achieves 68.18% accuracy at this challenging task while significantly outperforming several strong baselines. Finally, we present extensive qualitative results illustrating our method’s predictions and the visual evidence our method relied on. Our code and annotated dataset can be found at the enclosed link.



Paperid:1502
Authors:Ayush Jain; Nikolaos Gkanatsios; Ishita Mediratta; Katerina Fragkiadaki
Title: Bottom Up Top down Detection Transformers for Language Grounding in Images and Point Clouds
Abstract:
Most models tasked to ground referential utterances in 2D and 3D scenes learn to select the referred object from a pool of object proposals provided by a pre-trained detector. This is limiting because an utterance may refer to visual entities at various levels of granularity, such as the chair, the leg of the chair, or the tip of the front leg of the chair, which may be missed by the detector. We propose a language grounding model that attends on the referential utterance and on the object proposal pool computed from a pre-trained detector to decode referenced objects with a detection head, without selecting them from the pool. In this way, it is helped by powerful pre-trained object detectors without being restricted by their misses. We call our model Bottom Up Top Down DEtection TRansformers (BUTD-DETR) because it uses both language guidance (top down) and objectness guidance (bottom-up) to ground referential utterances in images and point clouds. Moreover, BUTD-DETR casts object detection as referential grounding and uses object labels as language prompts to be grounded in the visual scene, augmenting supervision for the referential grounding task in this way. The proposed model sets a new state-of-the-art across popular 3D language grounding benchmarks with significant performance gains over previous 3D approaches (12.6% on SR3D, 11.6% on NR3D and 6.3% on ScanRefer). When applied in 2D images, it performs on par with the previous state of the art. We ablate the design choices of our model and quantify their contribution to performance. Our code and checkpoints can be found at the project website https://butd-detr.github.io"



Paperid:1503
Authors:Yifeng Zhang; Ming Jiang; Qi Zhao
Title: New Datasets and Models for Contextual Reasoning in Visual Dialog
Abstract:
Visual Dialog (VD) is a vision-language task that requires AI systems to maintain a natural question-answering dialog about visual contents. Using the dialog history as contexts, VD models have achieved promising performance on public benchmarks. However, prior VD datasets do not provide sufficient contextually dependent questions that require knowledge from the dialog history to answer. As a result, advanced VQA models can still perform well without considering the dialog context. In this work, we focus on developing new datasets and models to highlight the role of contextual reasoning in VD. We define a hierarchy of contextual patterns to represent and organize the dialog context, enabling quantitative analyses of contextual dependencies and designs of new VD datasets and models. We then develop two new datasets, namely CLEVR-VD and GQA-VD, offering context-rich dialogs over synthetic and realistic images, respectively. Furthermore, we propose a novel neural module network method featuring contextual reasoning in VD. We demonstrate the effectiveness of our proposed datasets and method with experimental results and model comparisons across different datasets. Our code and data are available at https://github.com/SuperJohnZhang/ContextVD.



Paperid:1504
Authors:Joanna Hong; Minsu Kim; Yong Man Ro
Title: VisageSynTalk: Unseen Speaker Video-to-Speech Synthesis via Speech-Visage Feature Selection
Abstract:
The goal of this work is to reconstruct speech from a silent talking face video. Recent studies have shown impressive performance on synthesizing speech from silent talking face videos. However, they have not explicitly considered on varying identity characteristics of different speakers, which place a challenge in the video-to-speech synthesis, and this becomes more critical in unseen-speaker settings. Our approach is to separate the speech content and the visage-style from a given silent talking face video. By guiding the model to independently focus on modeling the two representations, we can obtain the speech of high intelligibility from the model even when the input video of an unseen subject is given. To this end, we introduce speech-visage selection that separates the speech content and the speaker identity from the visual features of the input video. The disentangled representations are jointly incorporated to synthesize speech through visage-style based synthesizer which generates speech by coating the visage-styles while maintaining the speech content. Thus, the proposed framework brings the advantage of synthesizing the speech containing the right content even with the silent talking face video of an unseen subject. We validate the effectiveness of the proposed framework on the GRID, TCD-TIMIT volunteer, and LRW datasets.



Paperid:1505
Authors:Matan Levy; Rami Ben-Ari; Dani Lischinski
Title: Classification-Regression for Chart Comprehension
Abstract:
Chart question answering (CQA) is a task used for assessing chart comprehension, which is fundamentally different from understanding natural images. CQA requires analyzing the relationships between the textual and the visual components of a chart, in order to answer general questions or infer numerical values. Most existing CQA datasets and models are based on simplifying assumptions that often enable surpassing human performance. In this work, we address this outcome and propose a new model that jointly learns classification and regression. Our language-vision setup uses co-attention transformers to capture the complex real-world interactions between the question and the textual elements. We validate our design with extensive experiments on the realistic PlotQA dataset, outperforming previous approaches by a large margin, while showing competitive performance on FigureQA. Our model is particularly well suited for realistic questions with out-of-vocabulary answers that require regression.



Paperid:1506
Authors:Benita Wong; Joya Chen; You Wu; Stan Weixian Lei; Dongxing Mao; Difei Gao; Mike Zheng Shou
Title: AssistQ: Affordance-Centric Question-Driven Task Completion for Egocentric Assistant
Abstract:
A long-standing goal of intelligent assistants such as AR glasses/robots has been to assist users in affordance-centric real-world scenarios, such as ""how can I run the microwave for 1 minute?”. However, there is still no clear task definition and suitable benchmarks. In this paper, we define a new task called Affordance-centric Question-driven Task Completion, where the AI assistant should learn from instructional videos to provide step-by-step help in the user’s view. To support the task, we constructed AssistQ, a new dataset comprising 531 question-answer samples from 100 newly filmed instructional videos. We also developed a novel Question-to-Actions (Q2A) model to address the AQTC task and validate it on the AssistQ dataset. The results show that our model significantly outperforms several VQA-related baselines while still having large room for improvement. We expect our task and dataset to advance Egocentric AI Assistant’s development. Our project page is available at: https://showlab.github.io/assistq/.



Paperid:1507
Authors:Weicheng Kuo; Fred Bertsch; Wei Li; AJ Piergiovanni; Mohammad Saffar; Anelia Angelova
Title: FindIt: Generalized Localization with Natural Language Queries
Abstract:
We propose FindIt, a simple and versatile framework that unifies a variety of visual grounding and localization tasks including referring expression comprehension, text-based localization, and object detection. Key to our architecture is an efficient multi-scale fusion module that unifies the disparate localization requirements across the tasks. In addition, we discover that a standard object detector is surprisingly effective in unifying these tasks without a need for task-specific design, losses, or pre-computed detections. Our end-to-end trainable framework responds flexibly and accurately to a wide range of referring expression, localization or detection queries for zero, one, or multiple objects. Jointly trained on these tasks, FindIt outperforms the state of the art on both referring expression and text-based localization, and shows competitive performance on object detection. Finally, FindIt generalizes better to out-of-distribution data and novel categories compared to strong single-task baselines. All of these are accomplished by a single, unified and efficient model. The code will be released at: https://github.com/google-research/google-research/tree/master/findit.



Paperid:1508
Authors:Zhengyuan Yang; Zhe Gan; Jianfeng Wang; Xiaowei Hu; Faisal Ahmed; Zicheng Liu; Yumao Lu; Lijuan Wang
Title: UniTAB: Unifying Text and Box Outputs for Grounded Vision-Language Modeling
Abstract:
We propose UniTAB that Unifies Text And Box outputs for grounded vision-language (VL) modeling. Grounded VL tasks such as grounded captioning require the model to generate a text description and align predicted words with object regions. To achieve this, models must generate desired text and box outputs together, and meanwhile indicate the alignments between words and boxes. In contrast to existing solutions that use multiple separate modules for different outputs, UniTAB represents both text and box outputs with a shared token sequence, and introduces a special token to naturally indicate word-box alignments in the sequence. UniTAB thus could provide a more comprehensive and interpretable image description, by freely grounding generated words to object regions. On grounded captioning, UniTAB presents a simpler solution with a single output head, and significantly outperforms state of the art in both grounding and captioning evaluations. On general VL tasks that have different desired output formats (i.e., text, box, or their combination), UniTAB with a single network achieves better or comparable performance than task-specific state of the art. Experiments cover 7 VL benchmarks, including grounded captioning, visual grounding, image captioning, and visual question answering. Furthermore, UniTAB’s unified multi-task network and the task-agnostic output sequence design make the model parameter efficient and generalizable to new tasks.



Paperid:1509
Authors:Golnaz Ghiasi; Xiuye Gu; Yin Cui; Tsung-Yi Lin
Title: Scaling Open-Vocabulary Image Segmentation with Image-Level Labels
Abstract:
We design an open-vocabulary image segmentation model to organize an image into meaningful regions indicated by arbitrary texts. Recent works (CLIP and ALIGN), despite attaining impressive open-vocabulary classification accuracy with image-level caption labels, are unable to segment visual concepts with pixels. We argue that these models miss an important step of visual grouping, which organizes pixels into groups before learning visual-semantic alignments. We propose OpenSeg to address the above issue while still making use of scalable image-level supervision of captions. First, it learns to propose segmentation masks for possible organizations. Then it learns visual-semantic alignments by aligning each word in a caption to one or a few predicted masks. We find the mask representations are the key to support learning image segmentation from captions, making it possible to scale up the dataset and vocabulary sizes. OpenSeg significantly outperforms the recent open-vocabulary method of LSeg by +19.9 mIoU on PASCAL dataset, thanks to its scalability.



Paperid:1510
Authors:Jack Hessel; Jena D. Hwang; Jae Sung Park; Rowan Zellers; Chandra Bhagavatula; Anna Rohrbach; Kate Saenko; Yejin Choi
Title: The Abduction of Sherlock Holmes: A Dataset for Visual Abductive Reasoning
Abstract:
Humans have remarkable capacity to reason abductively and hypothesize about what lies beyond the literal content of an image. By identifying concrete visual clues scattered throughout a scene, we almost can’t help but draw probable inferences beyond the literal scene based on our everyday experience and knowledge about the world. For example, if we see a “20 mph” sign alongside a road, we might assume the street sits in a residential area (rather than on a highway), even if no houses are pictured. Can machines perform similar visual reasoning? We present Sherlock, an annotated corpus of 103K images for testing machine capacity for abductive reasoning beyond literal image contents. We adopt a free-viewing paradigm: participants first observe and identify salient clues within images (e.g., objects, actions) and then provide a plausible inference about the scene, given the clue. In total, we collect 363K (clue, inference) pairs, which form a first-of-its-kind abductive visual reasoning dataset. Using our corpus, we test three complementary axes of abductive reasoning. We evaluate the capacity of models to: i) retrieve relevant inferences from a large candidate corpus; ii) localize evidence for inferences via bounding boxes, and iii) compare plausible inferences to match human judgments on a newly-collected diagnostic corpus of 19K Likert-scale judgments. While we find that fine-tuning CLIP-RN50x64 with a multitask objective outperforms strong baselines, significant headroom exists between model performance and human agreement. Data, models, and leaderboard available at http://visualabduction.com/.



Paperid:1511
Authors:Minsu Kim; Hyunjun Kim; Yong Man Ro
Title: Speaker-Adaptive Lip Reading with User-Dependent Padding
Abstract:
Lip reading aims to predict speech based on lip movements alone. As it focuses on visual information to model the speech, its performance is inherently sensitive to personal lip appearances and movements. This makes the lip reading models show degraded performance when they are applied to unseen speakers due to the mismatch between training and testing conditions. Speaker adaptation technique aims to reduce this mismatch between train and test speakers, thus guiding a trained model to focus on modeling the speech content without being intervened by the speaker variations. In contrast to the efforts made in audio-based speech recognition for decades, the speaker adaptation methods have not well been studied in lip reading. In this paper, to remedy the performance degradation of lip reading model on unseen speakers, we propose a speaker-adaptive lip reading method, namely user-dependent padding. The user-dependent padding is a speaker-specific input that can participate in the visual feature extraction stage of a pre-trained lip reading model. Therefore, the lip appearances and movements information of different speakers can be considered during the visual feature encoding, adaptively for individual speakers. Moreover, the proposed method does not need 1) any additional layers, 2) to modify the learned weights of the pre-trained model, and 3) the speaker label of train data used during pre-train. It can directly adapt to unseen speakers by learning the user-dependent padding only, in a supervised or unsupervised manner. Finally, to alleviate the speaker information insufficiency in public lip reading databases, we label the speaker of a well-known audio-visual database, LRW, and design an unseen-speaker lip reading scenario named LRW-ID. The effectiveness of the proposed method is verified on sentence- and word-level lip reading, and we show it can further improve the performance of a well-trained model with large speaker variations.



Paperid:1512
Authors:Tan M. Dinh; Rang Nguyen; Binh-Son Hua
Title: TISE: Bag of Metrics for Text-to-Image Synthesis Evaluation
Abstract:
In this paper, we conduct a study on the state-of-the-art methods for text-to-image synthesis and propose a framework to evaluate these methods. We consider syntheses where an image contains a single or multiple objects. Our study outlines several issues in the current evaluation pipeline: (i) for image quality assessment, a commonly used metric, e.g., Inception Score (IS), is often either miscalibrated for the single-object case or misused for the multi-object case; (ii) for text relevance and object accuracy assessment, there is an overfitting phenomenon in the existing R-precision (RP) and Semantic Object Accuracy (SOA) metrics, respectively; (iii) for multi-object case, many vital factors for evaluation, e.g., object fidelity, positional alignment, counting alignment, are largely dismissed; (iv) the ranking of the methods based on current metrics is highly inconsistent with real images. To overcome these issues, we propose a combined set of existing and new metrics to systematically evaluate the methods. For existing metrics, we offer an improved version of IS named IS* by using temperature scaling to calibrate the confidence of the classifier used by IS; we also propose a solution to mitigate the overfitting issues of RP and SOA. For new metrics, we develop counting alignment, positional alignment, object-centric IS, and object-centric FID metrics for evaluating the multi-object case. We show that benchmarking with our bag of metrics results in a highly consistent ranking among existing methods that is well-aligned with human evaluation. As a by-product, we create AttnGAN++, a simple but strong baseline for the benchmark by stabilizing the training of AttnGAN using spectral normalization. We also release our toolbox, so-called TISE, for advocating fair and consistent evaluation of text-to-image models.



Paperid:1513
Authors:Morgan Heisler; Amin Banitalebi-Dehkordi; Yong Zhang
Title: SemAug: Semantically Meaningful Image Augmentations for Object Detection through Language Grounding
Abstract:
Data augmentation is an essential technique in improving the generalization of deep neural networks. The majority of existing image-domain augmentations either rely on geometric and structural transformations, or apply different kinds of photometric distortions. In this paper, we propose an effective technique for image augmentation by injecting contextually meaningful knowledge into the scenes. Our method of semantically meaningful image augmentation for object detection via language grounding, SemAug, starts by calculating semantically appropriate new objects that can be placed into relevant locations in the image (the what and where problems). Then it embeds these objects into their relevant target locations, thereby promoting diversity of object instance distribution. Our method allows for introducing new object instances and categories that may not even exist in the training set. Furthermore, it does not require the additional overhead of training a context network, so it can be easily added to existing architectures. Our comprehensive set of evaluations showed that the proposed method is very effective in improving the generalization, while the overhead is negligible. In particular, for a wide range of model architectures, our method achieved 2-4% and 1-2% mAP improvements for the task of object detection on the Pascal VOC and COCO datasets, respectively. Code is available as supplementary.



Paperid:1514
Authors:Myungsub Choi
Title: Referring Object Manipulation of Natural Images with Conditional Classifier-Free Guidance
Abstract:
We introduce the problem of referring object manipulation (ROM), which aims to generate photo-realistic image edits regarding two textual descriptions: 1) a text referring to an object in the input image and 2) a text describing how to manipulate the referred object. A successful ROM model would enable users to simply use natural language to manipulate images, removing the need for learning sophisticated image editing software. We present one of the first approach to address this challenging multi-modal problem by combining a referring image segmentation method with a text-guided diffusion model. Specifically, we propose a conditional classifier-free guidance scheme to better guide the diffusion process along the direction from the referring expression to the target prompt. In addition, we provide a new localized ranking method and further improvements to make the generated edits more robust. Experimental results show that the proposed framework can serve as a simple but strong baseline for referring object manipulation. Also, comparisons with several baseline text-guided diffusion models demonstrate the effectiveness of our conditional classifier-free guidance technique.



Paperid:1515
Authors:Reuben Tan; Bryan A. Plummer; Kate Saenko; JP Lewis; Avneesh Sud; Thomas Leung
Title: NewsStories: Illustrating Articles with Visual Summaries
Abstract:
Recent self-supervised approaches have used large-scale image-text datasets to learn powerful representations that transfer to many tasks without finetuning. These methods often assume that there is one-to-one correspondence between its images and their (short) captions. However, many tasks require reasoning about multiple images and long text narratives, such as describing news articles with visual summaries. Thus, we explore a novel setting where the goal is to learn a self-supervised visual-language representation that is robust to varying text length and the number of images. In addition, unlike prior work which assumed captions have a literal relation to the image, we assume images only contain loose illustrative correspondence with the text. To explore this problem, we introduce a large-scale multimodal dataset containing over 31M articles, 22M images and 1M videos. We show that state-of-the-art image-text alignment methods are not robust to longer narratives with multiple images. Finally, we introduce an intuitive baseline that outperforms these methods on zero-shot image-set retrieval by 10% on the GoodNews dataset.



Paperid:1516
Authors:Amita Kamath; Christopher Clark; Tanmay Gupta; Eric Kolve; Derek Hoiem; Aniruddha Kembhavi
Title: Webly Supervised Concept Expansion for General Purpose Vision Models
Abstract:
General purpose vision (GPV) systems are models that are designed to solve a wide array of visual tasks without requiring architectural changes. Today, GPVs primarily learn both skills and concepts from large fully supervised datasets. Scaling GPVs to tens of thousands of concepts by acquiring data to learn each concept for every skill quickly becomes prohibitive. This work presents an effective and inexpensive alternative: learn skills from supervised datasets, learn concepts from web image search, and leverage a key characteristic of GPVs -- the ability to transfer visual knowledge across skills. We use a dataset of 1M+ images spanning 10k+ visual concepts to demonstrate webly-supervised concept expansion for two existing GPVs (GPV-1 and VL-T5) on 3 benchmarks - 5 COCO based datasets (80 primary concepts), a newly curated series of 5 datasets based on the OpenImages and VisualGenome repositories ( 500 concepts) and the Web-derived dataset (10k+ concepts). We also propose a new architecture, GPV-2 that supports a variety of tasks -- from vision tasks like classification and localization to vision+language tasks like QA and captioning to more niche ones like human-object interaction detection. GPV-2 benefits hugely from web data and outperforms GPV-1 and VL-T5 across these benchmarks.



Paperid:1517
Authors:Kaiwen Zhou; Xin Eric Wang
Title: FedVLN: Privacy-Preserving Federated Vision-and-Language Navigation
Abstract:
Data privacy is a central problem for embodied agents that can perceive the environment, communicate with humans, and act in the real world. While helping humans complete tasks, the agent may observe and process sensitive information of users, such as house environments, human activities, etc. In this work, we introduce privacy-preserving embodied agent learning for the task of Vision-and-Language Navigation (VLN), where an embodied agent navigates house environments by following natural language instructions. We view each house environment as a local client, which shares nothing other than local updates with the cloud server and other clients, and propose a novel federated vision-and-language navigation (FedVLN) framework to protect data privacy during both training and pre-exploration. Particularly, we propose a decentralized training strategy to limit the data of each client to its local model training and a federated pre-exploration method to do partial model aggregation to improve model generalizability to unseen environments. Extensive results on R2R and RxR datasets show that under our FedVLN framework, decentralized VLN models achieve comparable results with centralized training while protecting seen environment privacy, and federated pre-exploration significantly outperforms centralized pre-exploration while preserving unseen environment privacy. Code is available at https://github.com/eric-ai-lab/FedVLN.



Paperid:1518
Authors:Haoran Wang; Dongliang He; Wenhao Wu; Boyang Xia; Min Yang; Fu Li; Yunlong Yu; Zhong Ji; Errui Ding; Jingdong Wang
Title: CODER: Coupled Diversity-Sensitive Momentum Contrastive Learning for Image-Text Retrieval
Abstract:
Image-Text Retrieval (ITR) is challenging in bridging visual and lingual modalities. Contrastive learning has been adopted by most prior arts. Except for limited amount of negative image-text pairs, the capability of constrastive learning is restricted by manually weighting negative pairs as well as unawareness of external knowledge. In this paper, we propose our novel Coupled Diversity-Sensitive Momentum Constrastive Learning (CODER) for improving cross-modal representation. Firstly, a novel diversity-sensitive contrastive learning (DCL) architecture is invented. We introduce dynamic dictionaries for both modalities to enlarge the scale of image-text pairs, and diversity-sensitiveness is achieved by adaptive negative pair weighting. Furthermore, two branches are designed in CODER. One learns instance-level embeddings from image/text, and it also generates pseudo online clustering labels for its input image/text based on their embeddings. Meanwhile, the other branch learns to query from commonsense knowledge graph to form concept-level descriptors for both modalities. Afterwards, both branches leverage DCL to align the cross-modal embedding spaces while an extra pseudo clustering label prediction loss is utilized to promote concept-level representation learning for the second branch. Extensive experiments conducted on two popular benchmarks, i.e. MSCOCO and Flicker30K, validate CODER remarkably outperforms the state-of-the-art approaches.



Paperid:1519
Authors:Tsu-Jui Fu; Xin Eric Wang; William Yang Wang
Title: Language-Driven Artistic Style Transfer
Abstract:
Despite having promising results, style transfer, which requires preparing style images in advance, may result in lack of creativity and accessibility. Following human instruction, on the other hand, is the most natural way to perform artistic style transfer that can significantly improve controllability for visual effect applications. We introduce a new task---language-driven artistic style transfer (LDAST)---to manipulate the style of a content image, guided by a text. We propose contrastive language visual artist (CLVA) that learns to extract visual semantics from style instructions and accomplish LDAST by the patch-wise style discriminator. The discriminator considers the correlation between language and patches of style images or transferred results to jointly embed style instructions. CLVA further compares contrastive pairs of content image and style instruction to improve the mutual relativeness. The results from the same content image can preserve consistent content structures. Besides, they should present analogous style patterns from style instructions that contain similar visual semantics. The experiments show that our CLVA is effective and achieves superb transferred results on LDAST.



Paperid:1520
Authors:Zaid Khan; Vijay Kumar B G; Xiang Yu; Samuel Schulter; Manmohan Chandraker; Yun Fu
Title: Single-Stream Multi-level Alignment for Vision-Language Pretraining
Abstract:
Self-supervised vision-language pretraining from pure images and text with a contrastive loss is effective, but ignores fine-grained alignment due to a dual-stream architecture that aligns image and text representations only on a global level. Earlier, supervised, non-contrastive methods were capable of finer-grained alignment, but required dense annotations that were not scalable. We propose a single stream architecture that aligns images and language at multiple levels: global, fine-grained patch-token, and conceptual/semantic, using two novel tasks: symmetric cross-modality reconstruction (XMM) and a pseudo-labeled key word prediction (PSL). In XMM, we mask input tokens from one modality and use cross-modal information to reconstruct the masked token, thus improving fine-grained alignment between the two modalities. In PSL, we use attention to select keywords in a caption, use a momentum encoder to recommend other important keywords that are missing from the caption but represented in the image, and then train the visual encoder to predict the presence of those keywords, helping it learn semantic concepts that are essential for grounding a textual token to an image region. We demonstrate competitive performance and improved data efficiency on image-text retrieval, grounding, visual question answering/reasoning against larger models and models trained on more data. Code and models available at zaidkhan.me/SIMLA.



Paperid:1521
Authors:Liuwan Zhu; Rui Ning; Jiang Li; Chunsheng Xin; Hongyi Wu
Title: Most and Least Retrievable Images in Visual-Language Query Systems
Abstract:
This is the first work to introduce the Most Retrievable Image(MRI) and Least Retrievable Image(LRI) concepts in modern text-to-image retrieval systems. An MRI is associated with and thus can be retrieved by many unrelated texts, while an LRI is disassociated from and thus not retrievable by related texts. Both of them have important practical applications and implications. Due to their one-to-many nature, it is fundamentally challenging to construct MRI and LRI. This research addresses this nontrivial problem by developing novel and effective loss functions to craft perturbations that essentially corrupt feature correlation between visual and language spaces, thus enabling MRI and LRI. The proposed schemes are implemented based on CLIP, a state-of-the-art image and text representation model, to demonstrate MRI and LRI and their application in privacy-preserved image sharing and malicious advertisement. They are evaluated by extensive experiments based on the modern visual-language models on multiple benchmarks, including Paris, ImageNet, Flickr30k, and MSCOCO. The experimental results show the effectiveness and robustness of the proposed schemes for constructing MRI and LRI.



Paperid:1522
Authors:Dekun Wu; He Zhao; Xingce Bao; Richard P. Wildes
Title: Sports Video Analysis on Large-Scale Data
Abstract:
This paper investigates the modeling of automated machine description on sports video, which has seen much progress recently. Nevertheless, state-of-the-art approaches fall quite short of capturing how human experts analyze sports scenes. There are several major reasons: (1) The used dataset is collected from non-official providers, which naturally creates a gap between models trained on those datasets and real-world applications; (2) previously proposed methods require extensive annotation efforts (i.e., player and ball segmentation at pixel level) on localizing useful visual features to yield acceptable results; (3) very few public datasets are available. In this paper, we propose a novel large-scale NBA dataset for Sports Video Analysis (NSVA) with a focus on captioning, to address the above challenges. We also design a unified approach to process raw videos into a stack of meaningful features with minimum labelling efforts, showing that cross modeling on such features using a transformer architecture leads to strong performance. In addition, we demonstrate the broad application of NSVA by addressing two additional tasks, namely fine-grained sports action recognition and salient player identification. Code and dataset are available at https://github.com/jackwu502/NSVA.



Paperid:1523
Authors:Seonwoo Min; Nokyung Park; Siwon Kim; Seunghyun Park; Jinkyu Kim
Title: Grounding Visual Representations with Texts for Domain Generalization
Abstract:
Reducing the representational discrepancy between source and target domains is a key component to maximize the model generalization. In this work, we advocate for leveraging natural language supervision for the domain generalization task. We introduce two modules to ground visual representations with texts containing typical reasoning of humans: (1) Visual and Textual Joint Embedder and (2) Textual Explanation Generator. The former learns the image-text joint embedding space where we can ground high-level class-discriminative information into the model. The latter leverages an explainable model and generates explanations justifying the rationale behind its decision. To the best of our knowledge, this is the first work to leverage the vision-and-language cross-modality approach for the domain generalization task. Our experiments with a newly created CUB-DG benchmark dataset demonstrate that cross-modality supervision can be successfully used to ground domain-invariant visual representations and improve the model generalization. Furthermore, in the large-scale DomainBed benchmark, our proposed method achieves state-of-the-art results and ranks 1st in average performance for five multi-domain datasets. The dataset and codes are available at https://github.com/mswzeus/GVRT.



Paperid:1524
Authors:Joaquín Ossandón; Benjamín Earle; Alvaro Soto
Title: Bridging the Visual Semantic Gap in VLN via Semantically Richer Instructions
Abstract:
The Visual-and-Language Navigation (VLN) task requires understanding a textual instruction to navigate a natural indoor environment using only visual information. While this is a trivial task for most humans, it is still an open problem for AI models. In this work, we hypothesize that poor use of the visual information available is at the core of the low performance of current models. To support this hypothesis, we provide experimental evidence showing that state-of-the-art models are not severely affected when they receive just limited or even no visual data, indicating a strong overfitting to the textual instructions. To encourage a more suitable use of the visual information, we propose a new data augmentation method that fosters the inclusion of more explicit visual information in the generation of textual navigational instructions. Our main intuition is that current VLN datasets include textual instructions that are intended to inform an expert navigator, such as a human, but not a beginner visual navigational agent, such as a randomly initialized DL model. Specifically, to bridge the visual semantic gap of current VLN datasets, we take advantage of metadata available for the Matterport3D dataset that, among others, includes information about object labels that are present in the scenes. Training a state-of-the-art model with the new set of instructions increase its performance by 8% in terms of success rate on unseen environments, demonstrating the advantages of the proposed data augmentation method.



Paperid:1525
Authors:Adyasha Maharana; Darryl Hannan; Mohit Bansal
Title: StoryDALL-E: Adapting Pretrained Text-to-Image Transformers for Story Continuation
Abstract:
Recent advances in text-to-image synthesis have led to large pretrained transformers with excellent capabilities to generate visualizations from a given text. However, these models are ill-suited for specialized tasks like story visualization, which requires an agent to produce a sequence of images given a corresponding sequence of captions, forming a narrative. Moreover, we find that the story visualization task fails to accommodate generalization to unseen plots and characters in new narratives. Hence, we first propose the task of story continuation, where the generated visual story is conditioned on a source image, allowing for better generalization to narratives with new characters. Then, we enhance or ‘retro-fit’ the pretrained text-to-image synthesis models with task-specific modules for (a) sequential image generation and (b) copying relevant elements from an initial frame. We explore full-model finetuning, as well as prompt-based tuning for parameter-efficient adaptation, of the pretrained model. We evaluate our approach StoryDALL-E on two existing datasets, PororoSV and FlintstonesSV, and introduce a new dataset DiDeMoSV collected from a video-captioning dataset. We also develop a model StoryGANc based on Generative Adversarial Networks (GAN) for story continuation, and compare with the StoryDALL-E model to demonstrate the advantages of our approach. We show that our retro-fitting approach outperforms GAN-based models for story continuation. We also demonstrate that the ‘retro-fitting’ approach facilitates copying of visual elements from the source image and improved continuity in visual frames. Finally, our analysis suggests that pretrained transformers struggle with comprehending narratives containing multiple characters, and translating them into appropriate imagery. Our work encourages future research into story continuation and large-scale models for the task.



Paperid:1526
Authors:Katherine Crowson; Stella Biderman; Daniel Kornis; Dashiell Stander; Eric Hallahan; Louis Castricato; Edward Raff
Title: VQGAN-CLIP: Open Domain Image Generation and Editing with Natural Language Guidance
Abstract:
Image generation and manipulation requires technical expertise to use, inhibiting adoption. Current methods rely heavily on training to a specific domain (e.g., only faces), manual work or algorithm tuning to latent vector discovery, and manual effort in mask selection to alter only a part of an image. We address all of these usability constraints while producing images of high visual and semantic quality through a unique combination of OpenAI’s CLIP (Radford et al., 2021), VQGAN (Esser et al., 2021), and a generation augmentation strategy to produce VQGAN-CLIP. This allows generation and manipulation of images using natural language text, without further training on any domain datasets. We demonstrate on a variety of tasks how VQGAN-CLIP produces higher visual quality outputs than prior, less flexible approaches like minDALL-E (Kakaobrain, 2021) and Open-Edit (Liu, 2020), despite not being trained for the tasks presented.



Paperid:1527
Authors:Xian Liu; Yinghao Xu; Qianyi Wu; Hang Zhou; Wayne Wu; Bolei Zhou
Title: Semantic-Aware Implicit Neural Audio-Driven Video Portrait Generation
Abstract:
Animating high-fidelity video portrait with speech audio is crucial for virtual reality and digital entertainment. While most previous studies rely on accurate explicit structural information, recent works explore the implicit scene representation of Neural Radiance Fields (NeRF) for realistic generation. In order to capture the inconsistent motions as well as the semantic difference between human head and torso, some work models them via two individual sets of NeRF, leading to unnatural results. In this work, we propose Semantic-aware Speaking Portrait NeRF (SSP-NeRF), which creates delicate audio-driven portraits using one unified set of NeRF. The proposed model can handle the detailed local facial semantics and the global head-torso relationship through two semantic-aware modules. Specifically, we first propose a Semantic-Aware Dynamic Ray Sampling module with an additional parsing branch that facilitates audio-driven volume rendering. Moreover, to enable portrait rendering in one unified neural radiance field, a Torso Deformation module is designed to stabilize the large-scale non-rigid torso motions. Extensive evaluations demonstrate that our proposed approach renders realistic video portraits. Demo video and more resources can be found in https://alvinliu0.github.io/projects/SSP-NeRF"



Paperid:1528
Authors:Juan León Alcázar; Moritz Cordes; Chen Zhao; Bernard Ghanem
Title: End-to-End Active Speaker Detection
Abstract:
Abstract. Recent advances in the Active Speaker Detection (ASD) problem build upon a two stage process: feature extraction and spatio-temporal context aggregation. In this paper, we propose an end-to-end ASD workflow where feature learning and contextual predictions are jointly learned. Our end-to-end trainable network simultaneously learns multi-modal embeddings and aggregates spatio-temporal context. This results in more suitable feature representations and improved performance in the ASD task. We also introduce interleaved graph neural network (iGNN) blocks, which split the message passing according to the main sources of context in the ASD problem. Experiments show that the aggregated features from the iGNN blocks are more suitable for ASD, resulting in state-of-the art performance. Finally, we design a weakly-supervised strategy, which demonstrates that the ASD problem can also be approached by utilizing audiovisual data but relying exclusively on audio annotations. We achieve this by modelling the direct relationship between the audio signal and the possible sound sources (speakers), as well as introducing a contrastive loss.



Paperid:1529
Authors:Dingkang Yang; Shuai Huang; Shunli Wang; Yang Liu; Peng Zhai; Liuzhen Su; Mingcheng Li; Lihua Zhang
Title: Emotion Recognition for Multiple Context Awareness
Abstract:
Understanding emotion in context is a rising hotspot in the computer vision community. Existing methods lack reliable context semantics to mitigate uncertainty in expressing emotions and fail to model multiple context representations complementarily. To alleviate these issues, we present a context-aware emotion recognition framework that combines four complementary contexts. The first context is multimodal emotion recognition based on facial expression, facial landmarks, gesture and gait. Secondly, we adopt the channel and spatial attention modules to obtain the emotion semantics of the scene context. Inspired by sociology theory, we explore the emotion transmission between agents by constructing relationship graphs in the third context. Meanwhile, we propose a novel agent-object context, which aggregates emotion cues from the interactions between surrounding agents and objects in the scene to mitigate the ambiguity of prediction. Finally, we introduce an adaptive relevance fusion module for learning the shared representations among multiple contexts. Extensive experiments show that our approach outperforms the state-of-the-art methods on both EMOTIC and GroupWalk datasets. We also release a dataset annotated with diverse emotion labels, Human Emotion in Context (HECO). In practice, we compare with the existing methods on the HECO, and our approach obtains a higher classification average precision of 50.65% and a lower regression mean error rate of 0.7. The project is available at https://heco2022.github.io/.



Paperid:1530
Authors:Ayan Kumar Bhunia; Aneeshan Sain; Parth Hiren Shah; Animesh Gupta; Pinaki Nath Chowdhury; Tao Xiang; Yi-Zhe Song
Title: Adaptive Fine-Grained Sketch-Based Image Retrieval
Abstract:
The recent focus on Fine-Grained Sketch-Based Image Retrieval (FG-SBIR) has shifted towards generalising a model to new categories without any training data from them. In real-world applications, however, a trained FG-SBIR model is often applied to both new categories and different human sketchers, i.e., different drawing styles. Although this complicates the generalisation problem, fortunately, a handful of examples are typically available, enabling the model to adapt to the new category/style. In this paper, we offer a novel perspective -- instead of asking for a model that generalises, we advocate for one that quickly adapts, with just very few samples during testing (in a few-shot manner). To solve this new problem, we introduce a novel model-agnostic meta-learning (MAML) based framework with several key modifications: (1) As a retrieval task with a margin-based contrastive loss, we simplify the MAML training in the inner loop to make it more stable and tractable. (2) The margin in our contrastive loss is also meta-learned with the rest of the model. (3) Three additional regularisation losses are introduced in the outer loop, to make the meta-learned FG-SBIR model more effective for category/style adaptation. Extensive experiments on public datasets suggest a large gain over generalisation and zero-shot based approaches, and a few strong few-shot baselines.



Paperid:1531
Authors:Ye Zhu; Kyle Olszewski; Yu Wu; Panos Achlioptas; Menglei Chai; Yan Yan; Sergey Tulyakov
Title: Quantized GAN for Complex Music Generation from Dance Videos
Abstract:
We present Dance2Music-GAN (D2M-GAN), a novel adversarial multi-modal framework that generates complex musical samples conditioned on dance videos. Our proposed framework takes dance video frames and human body motions as input, and learns to generate music samples that plausibly accompany the corresponding input. Unlike most existing conditional music generation works that generate specific types of mono-instrumental sounds using symbolic audio representations (e.g., MIDI), and that usually rely on pre-defined musical synthesizers, in this work we generate dance music in complex styles (e.g., pop, breaking, etc.) by employing a Vector Quantized (VQ) audio representation, and leverage both its generality and high abstraction capacity of its symbolic and continuous counterparts. By performing an extensive set of experiments on multiple datasets, and following a comprehensive evaluation protocol, we assess the generative qualities of our proposal against alternatives. The attained quantitative results, which measure the music consistency, beats correspondence, and music diversity, demonstrate the effectiveness of our proposed method. Last but not least, we curate a challenging dance-music dataset of in-the-wild TikTok videos, which we use to further demonstrate the efficacy of our approach in real-world applications -- and which we hope to serve as a starting point for relevant future research.



Paperid:1532
Authors:Hu Wang; Jianpeng Zhang; Yuanhong Chen; Congbo Ma; Jodie Avery; Louise Hull; Gustavo Carneiro
Title: Uncertainty-Aware Multi-modal Learning via Cross-Modal Random Network Prediction
Abstract:
Multi-modal learning focuses on training models by equally combining multiple input data modalities during the prediction process. However, this equal combination can be detrimental to the prediction accuracy because different modalities are usually accompanied by varying levels of uncertainty. Using such uncertainty to combine modalities has been studied by a couple of approaches, but with limited success because these approaches are either designed to deal with specific classification or segmentation problems and cannot be easily translated into other tasks, or suffer from numerical instabilities. In this paper, we propose a new Uncertainty-aware Multi-modal Learner that estimates uncertainty by measuring feature density via Cross-modal Random Network Prediction (CRNP). CRNP is designed to require little adaptation to translate between different prediction tasks, while having a stable training process. From a technical point of view, CRNP is the first approach to explore random network prediction to estimate uncertainty and to combine multi-modal data. Experiments on two 3D multi-modal medical image segmentation tasks and three 2D multi-modal computer vision classification tasks show the effectiveness, adaptability and robustness of CRNP. Also, we provide an extensive discussion on different fusion functions and visualization to validate the proposed model.



Paperid:1533
Authors:Shentong Mo; Pedro Morgado
Title: Localizing Visual Sounds the Easy Way
Abstract:
Unsupervised audio-visual source localization aims at localizing visible sound sources in a video without relying on ground-truth localization for training. Previous works often seek high audio-visual similarities for likely positive (sounding) regions and low similarities for likely negative regions. However, accurately distinguishing between sounding and non-sounding regions is challenging without manual annotations. In this work, we propose a simple yet effective approach for Easy Visual Sound Localization, namely EZ-VSL, without relying on the construction of positive and/or negative regions during training. Instead, we align audio and visual spaces by seeking audio-visual representations that are aligned in, at least, one location of the associated image, while not matching other images, at any location. We also introduce a novel object-guided localization scheme at inference time for improved precision. Our simple and effective framework achieves state-of-the-art performance on two popular benchmarks, Flickr SoundNet and VGG-Sound Source. In particular, we improve the CIoU of the Flickr SoundNet test set from 76.80% to 83.94%, and on the VGG-Sound Source dataset from 34.60% to 38.85%.



Paperid:1534
Authors:Tingle Li; Yichen Liu; Andrew Owens; Hang Zhao
Title: Learning Visual Styles from Audio-Visual Associations
Abstract:
From the patter of rain to the crunch of snow, the sounds we hear often convey the visual textures that appear within a scene. In this paper, we present a method for learning visual styles from unlabeled audio-visual data. Our model learns to manipulate the texture of a scene to match a sound, a problem we term audio-driven image stylization. Given a dataset of paired audio-visual data, we learn to modify input images such that, after manipulation, they are more likely to co-occur with a given input sound. In quantitative and qualitative evaluations, our sound-based model outperforms label-based approaches. We also show that audio can be an intuitive representation for manipulating images, as adjusting a sound’s volume or mixing two sounds together results in predictable changes to visual style.



Paperid:1535
Authors:Jae-Ho Choi; Ki-Bong Kang; Kyung-Tae Kim
Title: Remote Respiration Monitoring of Moving Person Using Radio Signals
Abstract:
Non-contact respiration rate measurement (nRRM), which aims to monitor one’s breathing status without any contact with the skin, can be utilized in various remote applications (e.g., telehealth or emergency detection). The existing nRRM approaches mainly analyze fine details from videos to extract minute respiration signals; however, they have practical limitations in that the head or body of a subject must be quasi-stationary. In this study, we examine the task of estimating the respiration signal of a non-stationary subject (a person with large body movements or even walking around) based on radio signals. The key idea is that the received radio signals retain both the reflections from human global motion (GM) and respiration in a mixed form, while preserving the GM-only components at the same time. During training, our model leverages a novel multi-task adversarial learning (MTAL) framework to capture the mapping from radio signals to respiration while excluding the GM components in a self-supervised manner. We test the proposed model based on the newly collected and released datasets under real-world conditions. This study is the first realization of the nRRM task for moving/occluded scenarios, and also outperforms the state-of-the-art baselines even when the person sits still.



Paperid:1536
Authors:Karren Yang; Michael Firman; Eric Brachmann; Clément Godard
Title: Camera Pose Estimation and Localization with Active Audio Sensing
Abstract:
In this work, we show how to estimate a device’s position and orientation indoors by echolocation, i.e., by interpreting the echoes of an audio signal that the device itself emits. Established visual localization methods rely on the device’s camera and yield excellent accuracy if unique visual features are in view and depicted clearly. We argue that audio sensing can offer complementary information to vision for device localization, since audio is invariant to adverse visual conditions and can reveal scene information beyond a camera’s field of view. We first propose a strategy for learning an audio representation that captures the scene geometry around a device using supervision transfer from vision. Subsequently, we leverage this audio representation to complement vision in three device localization tasks: relative pose estimation, place recognition, and absolute pose regression. Our proposed methods outperform state-of-the-art vision models on new audio-visual benchmarks for the Replica and Matterport3D datasets.



Paperid:1537
Authors:Samuel Yu; Peter Wu; Paul Pu Liang; Ruslan Salakhutdinov; Louis-Philippe Morency
Title: PACS: A Dataset for Physical Audiovisual Commonsense Reasoning
Abstract:
In order for AI to be safely deployed in real-world scenarios such as hospitals, schools, and the workplace, it must be able to robustly reason about the physical world. Fundamental to this reasoning is physical common sense: understanding the physical properties and affordances of available objects, how they can be manipulated, and how they interact with other objects. Physical commonsense reasoning is fundamentally a multi-sensory task, since physical properties are manifested through multiple modalities - two of them being vision and acoustics. Our paper takes a step towards real-world physical commonsense reasoning by contributing PACS: the first audiovisual benchmark annotated for physical commonsense attributes. PACS contains 13,400 question-answer pairs, involving 1,377 unique physical commonsense questions and 1,526 videos. Our dataset provides new opportunities to advance the research field of physical reasoning by bringing audio as a core component of this multimodal problem. Using PACS, we evaluate multiple state-of-the-art models on our new challenging task. While some models show promising results (70% accuracy), they all fall short of human performance (95% accuracy). We conclude the paper by demonstrating the importance of multimodal reasoning and providing possible avenues for future research.



Paperid:1538
Authors:Juan F. Montesinos; Venkatesh S. Kadandale; Gloria Haro
Title: VoViT: Low Latency Graph-Based Audio-Visual Voice Separation Transformer
Abstract:
This paper presents an audio-visual approach for voice separation which produces state-of-the-art results at a low latency in two scenarios: speech and singing voice. The model is based on a two-stage network. Motion cues are obtained with a lightweight graph convolutional network that processes face landmarks. Then, both audio and motion features are fed to an audio-visual transformer which produces a fairly good estimation of the isolated target source. In a second stage, the predominant voice is enhanced with an audio-only network. We present different ablation studies and comparison to state-of-the-art methods. Finally, we explore the transferability of models trained for speech separation in the task of singing voice separation. The demos, code, and weights are available in https://ipcv.github.io/VoViT/"



Paperid:1539
Authors:Zhenqiang Ying; Deepti Ghadiyaram; Alan Bovik
Title: Telepresence Video Quality Assessment
Abstract:
Efficient and accurate video quality tools are needed to monitor and perceptually optimize telepresence traffic streamed via Zoom, Webex, Meet, etc. However, existing models are limited in their prediction capabilities on multi-modal, live streaming telepresence content. Here we address the significant challenges of Telepresence Video Quality Assessment (TVQA) in several ways. First, we mitigated the dearth of subjectively labeled data by collecting ~2k telepresence videos from different countries, on which we crowdsourced ~80k subjective quality labels. Using this new resource, we created a first-of-a-kind online video quality prediction framework for live streaming, using a multi-modal learning framework with separate pathways to compute visual and audio quality predictions. Our all-in-one model is able to provide accurate quality predictions at the patch, frame, clip, and audiovisual levels. Our model achieves state-of-the-art performance on both existing quality databases and our new TVQA database, at a considerably lower computational expense, making it an attractive solution for mobile and embedded systems.



Paperid:1540
Authors:Roman Bachmann; David Mizrahi; Andrei Atanov; Amir Zamir
Title: MultiMAE: Multi-modal Multi-task Masked Autoencoders
Abstract:
We propose a pre-training strategy called Multi-modal Multi-task Masked Autoencoders (MultiMAE). It differs from standard Masked Autoencoding in two key aspects: I) it can ‘optionally’ accept additional modalities of information in the input besides the RGB image (hence “multi-modal""""), and II) its training objective accordingly includes predicting multiple outputs besides the RGB image (hence “multi-task""""). We make use of masking (across image patches and input modalities) to make training MultiMAE tractable as well as to ensure cross-modality predictive coding is indeed learned by the network. We show this pre-training strategy leads to a flexible, simple, and efficient framework with improved transfer results to downstream tasks. In particular, the same exact pre-trained network can be flexibly used when additional information besides RGB images is available or when no information other than RGB is available - in all configurations yielding competitive to or significantly better results than the baselines. To avoid needing training datasets with multiple modalities and tasks, we train MultiMAE ‘entirely using pseudo-labeling’, which makes the framework widely applicable to any RGB dataset. The experiments are performed on multiple transfer tasks (image classification, semantic segmentation, depth estimation) on different datasets (ImageNet, ADE20K, Taskonomy, Hypersim, NYUv2), and the results show an intriguingly impressive capability by the network in cross-modal/task predictive coding and transfer. Code, pre-trained models, and interactive visualizations are available at https://multimae.epfl.ch.



Paperid:1541
Authors:Efthymios Tzinis; Scott Wisdom; Tal Remez; John R. Hershey
Title: AudioScopeV2: Audio-Visual Attention Architectures for Calibrated Open-Domain On-Screen Sound Separation
Abstract:
We introduce AudioScopeV2, a state-of-the-art universal audio-visual on-screen sound separation system which is capable of learning to separate sounds and associate them with on-screen objects by looking at in-the-wild videos. We identify several limitations of previous work on audio-visual on-screen sound separation, including the coarse resolution of spatio-temporal attention, poor convergence of the audio separation model, limited variety in training and evaluation data, and failure to account for the trade off between preservation of on-screen sounds and suppression of off-screen sounds. We provide solutions to all of these issues. Our proposed cross-modal and self-attention network architectures capture audio-visual dependencies at a finer resolution over time, and we also propose efficient separable variants that are capable of scaling to longer videos without sacrificing much performance. We also find that pre-training the separation model only on audio greatly improves results. For training and evaluation, we collected new human annotations of onscreen sounds from a large database of in-the-wild videos (YFCC100M). This new dataset is more diverse and challenging. Finally, we propose a calibration procedure that allows exact tuning of on-screen reconstruction versus off-screen suppression, which greatly simplifies comparing performance between models with different operating points. Overall, our experimental results show marked improvements in on-screen separation performance under much more general conditions than previous methods with minimal additional computational complexity.



Paperid:1542
Authors:Jinxing Zhou; Jianyuan Wang; Jiayi Zhang; Weixuan Sun; Jing Zhang; Stan Birchfield; Dan Guo; Lingpeng Kong; Meng Wang; Yiran Zhong
Title: Audio—Visual Segmentation
Abstract:
We propose to explore a new problem called audio-visual segmentation (AVS), in which the goal is to output a pixel-level map of the object(s) that produce sound at the time of the image frame. To facilitate this research, we construct the first audio-visual segmentation benchmark (AVSBench), providing pixel-wise annotations for the sounding objects in audible videos. Two settings are studied with this benchmark: 1) semi-supervised audio-visual segmentation with a single sound source and 2) fully-supervised audio-visual segmentation with multiple sound sources. To deal with the AVS problem, we propose a new method that uses a temporal pixel-wise audio-visual interaction module to inject audio semantics as guidance for the visual segmentation process. We also design a regularization loss to encourage the audio-visual mapping during training. Quantitative and qualitative experiments on the AVSBench compare our approach to several existing methods from related tasks, demonstrating that the proposed method is promising for building a bridge between the audio and pixel-wise visual semantics. Code is available at https://github.com/OpenNLPLab/AVSBench.



Paperid:1543
Authors:Yeying Jin; Wenhan Yang; Robby T. Tan
Title: Unsupervised Night Image Enhancement: When Layer Decomposition Meets Light-Effects Suppression
Abstract:
Night images suffer not only from low light, but also from uneven distributions of light. Most existing night visibility enhancement methods focus mainly on enhancing low-light regions. This inevitably leads to over enhancement and saturation in bright regions, such as those regions affected by light effects (glare, floodlight, etc). To address this problem, we need to suppress the light effects in bright regions while, at the same time, boosting the intensity of dark regions. With this idea in mind, we introduce an unsupervised method that integrates a layer decomposition network and a light-effects suppression network. Given a single night image as input, our decomposition network learns to decompose shading, reflectance and light-effects layers, guided by unsupervised layer-specific prior losses. Our light-effects suppression network further suppresses the light effects and, at the same time, enhances the illumination in dark regions. This light-effects suppression network exploits the estimated light-effects layer as the guidance to focus on the light-effects regions. To recover the background details and reduce hallucination/artefacts, we propose structure and high-frequency consistency losses. Our quantitative and qualitative evaluations on real images show that our method outperforms state-of-the-art methods in suppressing night light effects and boosting the intensity of dark regions. Our data and code is available at: https://github.com/jinyeying/night-enhancement>"



Paperid:1544
Authors:Suprosanna Shit; Rajat Koner; Bastian Wittmann; Johannes Paetzold; Ivan Ezhov; Hongwei Li; Jiazhen Pan; Sahand Sharifzadeh; Georgios Kaissis; Volker Tresp; Bjoern Menze
Title: Relationformer: A Unified Framework for Image-to-Graph Generation
Abstract:
A comprehensive representation of an image requires understanding objects and their mutual relationship, especially in image-to-graph generation, e.g., road network extraction, blood-vessel network extraction, or scene graph generation. Traditionally, image-to-graph generation is addressed with a two-stage approach consisting of object detection followed by a separate relation prediction, which prevents simultaneous object-relation interaction. This work proposes a unified one-stage transformer-based framework, namely Relationformer, that jointly predicts objects and their relations. We leverage direct set-based object prediction and incorporate the interaction among the objects to learn an object-relation representation jointly. In addition to existing [obj]-tokens, we propose a novel learnable token, namely [rln]-token. Together with [obj]-tokens, [rln]-token exploits local and global semantic reasoning in an image through a series of mutual associations. In combination with the pair-wise [obj]-token, the [rln]-token contributes to a computationally efficient relation prediction. We achieve state-of-the-art performance on multiple, diverse and multi-domain datasets that demonstrate our approach’s effectiveness and generalizability.



Paperid:1545
Authors:Shruti Vyas; Chen Chen; Mubarak Shah
Title: GAMa: Cross-view Video Geo-localization
Abstract:
The existing work in cross-view geo-localization is based on images where a ground panorama is matched to an aerial image. In this work, we focus on ground videos instead of images which provides additional contextual cues which are important for this task. There are no existing datasets for this problem, therefore we propose GAMa dataset, a large-scale dataset with ground videos and corresponding aerial images. We also propose a novel approach to solve this problem. At clip-level, a short video clip is matched with corresponding aerial image and is later used to get video-level geo-localization of a long video. Moreover, we propose a hierarchical approach to further improve the clip-level geo-localization. On this challenging dataset, with unaligned images and limited field of view, our proposed method achieves a Top-1 recall rate of 19.4% and 45.1% @1.0mile. Code & dataset are available at this link.



Paperid:1546
Authors:Kengo Nakata; Youyang Ng; Daisuke Miyashita; Asuka Maki; Yu-Chieh Lin; Jun Deguchi
Title: Revisiting a kNN-based Image Classification System with High-capacity Storage
Abstract:
In existing image classification systems that use deep neural networks, the knowledge needed for image classification is implicitly stored in model parameters. If users want to update this knowledge, then they need to fine-tune the model parameters. Moreover, users cannot verify the validity of inference results or evaluate the contribution of knowledge to the results. In this paper, we investigate a system that stores knowledge for image classification, such as image feature maps, labels, and original images, not in model parameters but in external storage. Our system refers to the storage like a database when classifying input images. To increase knowledge, our system updates the database instead of fine-tuning model parameters, which avoids catastrophic forgetting in incremental learning scenarios. We revisit a kNN (k-nearest neighbor) classifier and employ it in our system. By analyzing the neighborhood samples referred by the kNN algorithm, we can interpret how knowledge learned in the past is used for inference results. Our system achieves 79.8% top-1 accuracy on the ImageNet dataset without fine-tuning model parameters after pretraining, and 90.8% accuracy on the Split CIFAR-100 dataset in the task incremental learning setting.



Paperid:1547
Authors:Hao Feng; Wengang Zhou; Jiajun Deng; Yuechen Wang; Houqiang Li
Title: Geometric Representation Learning for Document Image Rectification
Abstract:
In document image rectification, there exist rich geometric constraints between the distorted image and the ground truth one. How- ever, such geometric constraints are largely ignored in existing advanced solutions, which limits the rectification performance. To this end, we present DocGeoNet for document image rectification by introducing explicit geometric representation. Technically, two typical attributes of the document image are involved in the proposed geometric representation learning, i.e., 3D shape and textlines. Our motivation raises from the insight that 3D shape provides global unwarping cues for rectifying a distorted document image, while overlooking the local structure. On the other hand, textlines complementarily provide explicit geometric constraints for local patterns. The learned geometric representation effectively bridges the distorted image and the ground truth one. Extensive experiments show the effectiveness of our framework and demonstrate the superiority of our DocGeoNet over state-of-the-art methods on both the DocUNet Benchmark dataset and our proposed DIR300 test set.



Paperid:1548
Authors:Guoli Jia; Jufeng Yang
Title: S2-VER: Semi-Supervised Visual Emotion Recognition
Abstract:
Visual emotion recognition (VER), which plays an important role in various applications, has attracted increasing attention of researchers. Due to the ambiguous characteristic of emotion, it is hard to annotate a reliable large-scale dataset in this field. An alternative solution is semi-supervised learning (SSL), which progressively selects high-confidence samples from unlabeled data to help optimize the model. However, it is challenging to directly employ existing SSL algorithms in VER task. On the one hand, compared with object recognition, in VER task, the accuracy of the produced pseudo labels for unlabeled data drops a large margin. On the other hand, the maximum probability in the prediction is difficult to reach the fixed threshold, which leads to few unlabeled samples can be leveraged. Both of them would induce the suboptimal performance of the learned model. To address these issues, we propose S2-VER, the first SSL algorithm for VER, which consists of two com- ponents. The first component, reliable emotion label learning, aims to improve the accuracy of pseudo-labels. In detail, it generates smoothing labels by computing the similarity between the maintained emotion prototypes and the embedding of the sample. The second one is ambiguity-aware adaptive threshold strategy, which is dedicated to leveraging more unlabeled samples. Specifically, our strategy uses information entropy to measure the ambiguity of the smoothing labels, then adaptively adjusts the threshold, which is adopted to select high-confidence unlabeled samples. Extensive experiments conducted on six public datasets show that our proposed S2-VER performs favorably against the state-of-the-art approaches. The code is available at https://github.com/exped1230/S2-VER.



Paperid:1549
Authors:Ruoyu Feng; Xin Jin; Zongyu Guo; Runsen Feng; Yixin Gao; Tianyu He; Zhizheng Zhang; Simeng Sun; Zhibo Chen
Title: Image Coding for Machines with Omnipotent Feature Learning
Abstract:
Image Coding for Machines (ICM) aims to compress images for AI tasks analysis rather than meeting human perception. Learning a kind of feature that is both general (for AI tasks) and compact (for compression) is pivotal for its success. In this paper, we attempt to develop an ICM framework by learning universal features while also considering compression. We name such features as omnipotent features and the corresponding framework as Omni-ICM. Considering self-supervised learning (SSL) improves feature generalization, we integrate it with the compression task into the Omni-ICM framework to learn omnipotent features. However, it is non-trivial to coordinate semantics modeling in SSL and redundancy removing in compression, so we design a novel information filtering (IF) module between them by co-optimization of instance distinguishment and entropy minimization to adaptively drop information that is weakly related to AI tasks (e.g., some texture redundancy). Different from previous task-specific solutions, Omni-ICM could directly support AI tasks analysis based on the learned omnipotent features without joint training or extra transformation. Albeit simple and intuitive, Omni-ICM significantly outperforms existing traditional and learning-based codecs on multiple fundamental vision tasks.



Paperid:1550
Authors:Conghui Hu; Gim Hee Lee
Title: Feature Representation Learning for Unsupervised Cross-Domain Image Retrieval
Abstract:
Current supervised cross-domain image retrieval methods can achieve excellent performance. However, the cost of data collection and labeling imposes an intractable barrier to practical deployment in real applications. In this paper, we investigate the unsupervised cross-domain image retrieval task, where class labels and pairing annotations are no longer a prerequisite for training. This is an extremely challenging task because there is no supervision for both in-domain feature representation learning and cross-domain alignment. We address both challenges by introducing: 1) a new cluster-wise contrastive learning mechanism to help extract class semantic-aware features, and 2) a novel distance-of-distance loss to effectively measure and minimize the domain discrepancy without any external supervision. Experiments on the Office-Home and DomainNet datasets consistently show the superior image retrieval accuracies of our framework over state-of-the-art approaches. Our source code can be found at https://github.com/conghuihu/UCDIR.



Paperid:1551
Authors:Shilin Xu; Xiangtai Li; Jingbo Wang; Guangliang Cheng; Yunhai Tong; Dacheng Tao
Title: "Fashionformer: A Simple, Effective and Unified Baseline for Human Fashion Segmentation and Recognition"
Abstract:
Human fashion understanding is one important computer vision task since it has the comprehensive information for real-world applications. In this work, we focus on joint human fashion segmentation and attribute recognition. Contrary to the previous works that separately model each task as a multi-head prediction problem, our insight is to bridge these two tasks with one unified model via vision transformer modeling to benefit each task. In particular, we introduce the object query for segmentation and the attribute query for attribute prediction. Both queries and their corresponding features can be linked via mask prediction. Then we adopta two-stream query learning framework to learn the decoupled query representations. For attribute stream, we design a novel Multi-Layer Rendering module to explore more fine-grained features. The decoder design shares the same spirits with DETR, thus we name the proposed method \textit{Fahsionformer}. Extensive experiments on three human fashion datasets illustrate the effectiveness of our approach. In particular, our method with the same backbone achieverelative 10% improvements than previous works in case of \textit{a joint metric (AP^{{mask}}_{IoU+F_1}) for both segmentation and attribute recognition}. To the best of our knowledge, we are the first unified end-to-end vision transformer framework for human fashion analysis. We hope this simple yet effective method can serve as a new flexible baseline for fashion analysis. Code will be available at https://github.com/xushilin1/FashionFormer.



Paperid:1552
Authors:Xuqian Ren; Yifan Liu
Title: Semantic-Guided Multi-Mask Image Harmonization
Abstract:
Previous harmonization methods focus on adjusting one inharmonious region in an image based on an input mask. They may face problems when dealing with different perturbations on different semantic regions without available input masks. To deal with the problem that one image has been pasted with several foregrounds coming from different images and needs to harmonize them towards different domain directions without any mask as input, we propose a new semantic-guided multi-mask image harmonization task. Different from the previous single-mask image harmonization task, each inharmonious image is perturbed with different methods according to the semantic segmentation masks. Two challenging benchmarks, HScene and HLIP, are constructed based on 150 and 19 semantic classes, respectively. Furthermore, previous baselines focus on regressing the exact value for each pixel of the harmonized images. The generated results are in the ‘black box’ and cannot be edited. In this work, we propose a novel way to edit the inharmonious images by predicting a series of operator masks. The masks indicate the level and the position to apply a certain image editing operation, which could be the brightness, the saturation, and the color in a specific dimension. The operator masks provide more flexibility for users to edit the image further. Extensive experiments verify that the operator mask-based network can further improve those state-of-the-art methods which directly regress RGB images when the perturbations are structural. Experiments have been conducted on our constructed benchmarks to verify that our proposed operator mask-based framework can locate and modify the inharmonious regions in more complex scenes. Our code and models are available at https://github.com/XuqianRen/Semantic-guided-Multi-mask-Image-Harmonization.git.



Paperid:1553
Authors:Sagnik Das; Ke Ma; Zhixin Shu; Dimitris Samaras
Title: Learning an Isometric Surface Parameterization for Texture Unwrapping
Abstract:
In this paper, we present a novel approach to learn texture mapping for an isometrically deformed 3D surface and apply it for texture unwrapping of documents or other objects. Recent work on differentiable rendering techniques for implicit surfaces has shown high-quality 3D scene reconstruction and view synthesis results. However, these methods typically learn the appearance color as a function of the surface points and lack explicit surface parameterization. Thus they do not allow texture map extraction or texture editing. We propose an efficient method to learn surface parameterization by learning a continuous bijective mapping between 3D surface positions and 2D texture-space coordinates. Our surface parameterization network can be conveniently plugged into a differentiable rendering pipeline and trained using multi-view images and rendering loss. Using the learned parameterized implicit 3D surface we demonstrate state-of-the-art document-unwarping via texture extraction in both synthetic and real scenarios. We also show that our approach can reconstruct high-frequency textures for arbitrary objects. We further demonstrate the usefulness of our system by applying it to document and object texture editing. Code and related assets are available at: https://github.com/cvlab-stonybrook/Iso-UVField"



Paperid:1554
Authors:Rahul Duggal; Hao Zhou; Shuo Yang; Jun Fang; Yuanjun Xiong; Wei Xia
Title: Towards Regression-Free Neural Networks for Diverse Compute Platforms
Abstract:
With the shift towards on-device deep learning, ensuring a consistent behavior of an AI service across diverse compute platforms becomes tremendously important. Our work tackles the emergent problem of reducing predictive in-consistencies arising as negative flips: test samples that are correctly predicted by a less accurate on-device model, but incorrectly by a more accurate on-cloud one. We introduce REGression constrained Neural Architecture Search (REG-NAS) to design a family of highly accurate models that engender fewer negative flips. REG-NAS consists of two components: (1) A novel architecture constraint that enables a larger on-cloud model to contain all the weights of the smaller on-device one thus maximizing weight sharing. This idea stems from our observation that larger weight sharing among networks leads to similar sample-wise predictions and results in fewer negative flips; (2) A novel search reward that incorporates both Top-1 accuracy and negative flips in the architecture optimization metric. We demonstrate that REG-NAS can successfully find architecture with few negative flips, in three popular architecture search spaces. Compared to the existing state-of-the-art approach [29], REG-NAS leads to 33-48% relative reduction of negative flips.



Paperid:1555
Authors:Xiaoyu Xu; Jiayan Qiu; Xinchao Wang; Zhou Wang
Title: Relationship Spatialization for Depth Estimation
Abstract:
Considering the role played by the relationships between objects in monocular depth estimation (MDE), it can be easily told that relationships, such as ‘in front of’ and ‘behind’, provide explicit spatial priors for depth estimation. However, it is hard to answer the questions that which kinds of relationships embed with the useful spatial cues for MDE? And how much these relationships contribute to the MDE? We term the task of answering these two questions as Relationship Spatialization. To this end, we strive to spatializing the relationships by devising a novel learning-based framework. Specifically, given the monocular image, the image representations and the corresponding scene graph are firstly extracted, and the in-graph relationship representations are learnt to be obtained. Then, the relationship representations from the graph space are spatially aligned with the image representations from the visual space, followed by a redundancy elimination. Finally, we feed the concatenation of the image representations and the modified relationship representations into a depth predictor, which estimates the monocular depth with relationship spatialization. Experiments on KITTI, NYU v2 and ICL-NUIM datasets shows the effectiveness of relationship spatialization on MDE. Moreover, adopting our framework to current state-of-the-art MDE models leads to marginal improvement on most evaluation metrics.



Paperid:1556
Authors:Chenfeng Xu; Shijia Yang; Tomer Galanti; Bichen Wu; Xiangyu Yue; Bohan Zhai; Wei Zhan; Peter Vajda; Kurt Keutzer; Masayoshi Tomizuka
Title: Image2Point: 3D Point-Cloud Understanding with 2D Image Pretrained Models
Abstract:
3D point-clouds and 2D images are different visual representations of the physical world. While human vision can understand both representations, computer vision models designed for 2D image and 3D point-cloud understanding are quite different. Our paper explores the potential of transferring 2D model architectures and weights to understand 3D point-clouds, by empirically investigating the feasibility of the transfer, the benefits of the transfer, and shedding light on why the transfer works. We discover that we can indeed use the same architecture and pretrained weights of a neural net model to understand both images and point-clouds. Specifically, we transfer the image-pretrained model to a point-cloud model by copying or inflating the weights. We find that finetuning the transformed image-pretrained models (FIP) with minimal efforts --- only on input, output, and normalization layers --- can achieve competitive performance on 3D point-cloud classification, beating a wide range of point-cloud models that adopt task-specific architectures and use a variety of tricks. When finetuning the whole model, the performance gets further improved. Meanwhile, FIP improves data efficiency, reaching up to 10.0 top-1 accuracy percent on few-shot classification. It also speeds up training of point-cloud models by up to 11.1x for a target accuracy (e.g., 90 % accuracy). Lastly, we provide an explanation of the image to point-cloud transfer from the aspect of neural collapse.



Paperid:1557
Authors:Divya Kothandaraman; Tianrui Guan; Xijun Wang; Shuowen Hu; Ming Lin; Dinesh Manocha
Title: FAR: Fourier Aerial Video Recognition
Abstract:
We present a method, Fourier Activity Recognition (FAR), for UAV video activity recognition. Our formulation uses a novel Fourier object disentanglement method to innately separate out the human agent (which is typically small) from the background. Our disentanglement technique operates in the frequency domain to characterize the extent of temporal change of spatial pixels, and exploits convolution-multiplication properties of Fourier transform to map this representation to the corresponding object-background entangled features obtained from the network. To encapsulate contextual information and long-range space-time dependencies, we present a novel Fourier Attention algorithm, which emulates the benefits of self-attention by modeling the weighted outer product in the frequency domain. Our Fourier attention formulation uses much fewer computations than self-attention. We have evaluated our approach on multiple UAV datasets including UAV Human RGB, UAV Human Night, Drone Action, and NEC Drone. We demonstrate a relative improvement of 8.02% -38.69% in top-1 accuracy over prior work.



Paperid:1558
Authors:Ruocheng Wang; Yunzhi Zhang; Jiayuan Mao; Chin-Yi Cheng; Jiajun Wu
Title: Translating a Visual LEGO Manual to a Machine-Executable Plan
Abstract:
We study the problem of translating an image-based, step-by-step assembly manual created by human designers into machine-interpretable instructions. We formulate this problem as a sequential prediction task: at each step, our model reads the manual, locates the components to be added to the current shape, and infers their 3D poses. This task poses the challenge of establishing a 2D-3D correspondence between the manual image and the real 3D object, and 3D pose estimation for unseen 3D objects, since a new component to be added in a step can be an object built from previous steps. To address these two challenges, we present a novel learning-based framework, the Manual-to-Executable-Plan Network (MEPNet), which reconstructs the assembly steps from a sequence of manual images. The key idea is to integrate neural 2D keypoint detection modules and 2D-3D projection algorithms for high-precision prediction and strong generalization to unseen components. The MEPNet outperforms existing methods on three newly collected LEGO manual datasets and a Minecraft house dataset.



Paperid:1559
Authors:Junbang Liang; Ming Lin
Title: Fabric Material Recovery from Video Using Multi-Scale Geometric Auto-Encoder
Abstract:
Fabric materials are central to recreating realistic appearance of avatars in a virtual world and many VR applications, ranging from virtual try-on, teleconferencing, to character animation. We propose an end-to-end network model that uses video input to estimate the fabric materials of the garment worn by a human or an avatar in a virtual world. To achieve the high accuracy, we jointly learn human body and the garment geometry as conditions to material prediction. Due to the highly dynamic and deformable nature of cloth, general data-driven garment modeling remains a challenge. To address this problem, we propose a two-level auto-encoder to account for both global and local features of any garment geometry that would directly affect material perception. Using this network, we can also achieve smooth geometry transitioning between different garment topologies. During the estimation, we use a closed-loop optimization structure to share information between tasks and feed the learned garment features for temporal estimation of garment materials. Experiments show that our proposed network structures greatly improve the material classification accuracy by 1.5x, with applicability to unseen input. It also runs at least three orders of magnitude faster than the state-of-the-art. We demonstrate the recovered fabric materials on virtual try-on, where we recreate the entire avatar appearance, including body shape and pose, garment geometry and materials from only a single video.



Paperid:1560
Authors:Jie Ren; Wenteng Liang; Ran Yan; Luo Mai; Shiwen Liu; Xiao Liu
Title: MegBA: A GPU-Based Distributed Library for Large-Scale Bundle Adjustment
Abstract:
Large-scale Bundle Adjustment (BA) requires massive memory and computation resources which are difficult to be fulfilled by existing BA libraries. In this paper, we propose MegBA, a GPU-based distributed BA library. MegBA can provide massive aggregated memory by automatically partitioning large BA problems, and assigning the solvers of sub-problems to parallel nodes. The parallel solvers adopt distributed Precondition Conjugate Gradient and distributed Schur Elimination, so that an effective solution, which can match the precision of those computed by a single node, can be efficiently computed. To accelerate BA computation, we implement end-to-end BA computation using high-performance primitives available on commodity GPUs. MegBA exposes easy-to-use APIs that are compatible with existing popular BA libraries. Experiments show that MegBA can significantly outperform state-of-the-art BA libraries: Ceres (41.45×), RootBA (64.576×) and DeepLM (6.769×) in several large-scale BA benchmarks.



Paperid:1561
Authors:Georgios Pavlakos; Ethan Weber; Matthew Tancik; Angjoo Kanazawa
Title: The One Where They Reconstructed 3D Humans and Environments in TV Shows
Abstract:
TV shows depict a wide variety of human behaviors and have been studied extensively for their potential to be a rich source of data for many applications. However, the majority of the existing work focuses on 2D recognition tasks. In this paper, we make the observation that there is a certain persistence in TV shows, i.e., repetition of the environments and the humans, which makes possible the 3D reconstruction of this content. Building on this insight, we propose an automatic approach that operates on an entire season of a TV show and aggregates information in 3D; we build a 3D model of the environment, compute camera information, static 3D scene structure and body scale information. Then, we demonstrate how this information acts as rich 3D context that can guide and improve the recovery of 3D human pose and position in these environments. Moreover, we show that reasoning about humans and their environment in 3D enables a broad range of downstream applications: re-identification, gaze estimation, cinematography and image editing. We apply our approach on environments from seven iconic TV shows and perform an extensive evaluation of the proposed system.



Paperid:1562
Authors:Suraj Kothawade; Saikat Ghosh; Sumit Shekhar; Yu Xiang; Rishabh Iyer
Title: TALISMAN: Targeted Active Learning for Object Detection with Rare Classes and Slices Using Submodular Mutual Information
Abstract:
Deep neural networks based object detectors have shown great success in a variety of domains like autonomous vehicles, biomedical imaging, etc. It is known that their success depends on a large amount of data from the domain of interest. While deep models often perform well in terms of overall accuracy, they often struggle in performance on rare yet critical data slices. For example, data slices like ""motorcycle at night"" or ""bicycle at night"" are often rare but very critical slices for self-driving applications and false negatives on such rare slices could result in ill-fated failures and accidents. Active learning (AL) is a well-known paradigm to incrementally and adaptively build training datasets with a human in the loop. However, current AL based acquisition functions are not well-equipped to tackle real-world datasets with rare slices, since they are based on uncertainty scores or global descriptors of the image. We propose TALISMAN, a novel framework for Targeted Active Learning or object detectIon with rare slices using Submodular MutuAl iNformation. Our method uses the submodular mutual information functions instantiated using features of the region of interest (RoI) to efficiently target and acquire data points with rare slices. We evaluate our framework on the standard PASCAL VOC07+12 and BDD100K, a real-world self-driving dataset. We observe that TALISMAN outperforms other methods by in terms of average precision on rare slices, and in terms of mAP.



Paperid:1563
Authors:Junde Wu; Yu Zhang; Rao Fu; Yuanpei Liu; Jing Gao
Title: An Efficient Person Clustering Algorithm for Open Checkout-Free Groceries
Abstract:
Open checkout-free grocery is the grocery store where the customers never have to wait in line to check out. Developing a system like this is not trivial since it faces challenges of recognizing the dynamic and massive flow of people. In particular, a clustering method that can efficiently assign each snapshot to the corresponding customer is essential for the system. Motivated by unique challenges in the open checkout-free grocery, we propose an efficient and effective person clustering method. Specifically, we first propose a Crowded Sub-Graph (CSG) to localize the relationship among massive and continuous data streams. CSG is constructed by the proposed Pick-Link-Weight (PLW) strategy, which picks the nodes based on time-space information, links the nodes via trajectory information, and weighs the links by the proposed von Mises-Fisher (vMF) similarity metric. Then, to ensure that the method adapts to the dynamic and unseen person flow, we propose Graph Convolutional Network (GCN) with a simple Nearest Neighbor (NN) strategy to accurately cluster the instances of CSG. GCN is adopted to project the features into low-dimensional separable space, and NN is able to quickly produce a result in this space upon dynamic person flow. The experimental results show that the proposed method outperforms other alternative algorithms in this scenario. In practice, the whole system has been implemented and deployed in several real-world open checkout-free groceries.



Paperid:1564
Authors:Christian Joppi; Geri Skenderi; Marco Cristani
Title: POP: Mining POtential Performance of New Fashion Products via Webly Cross-Modal Query Expansion
Abstract:
We propose a data-centric pipeline able to generate exogenous observation data for the New Fashion Product Performance Forecasting (NFPPF) problem, i.e., predicting the performance of a brand-new clothing probe with no available past observations. Our pipeline manufactures the missing past starting from a single, available image of the clothing probe. It starts by expanding textual tags associated with the image, querying related fashionable or unfashionable images uploaded on the web at a specific time in the past. A binary classifier is robustly trained on these web images by confident learning, to learn what was fashionable in the past and how much the probe image conforms to this notion of fashionability. This compliance produces the POtential Performance (POP) time series, indicating how performing the probe could have been if it were available earlier. POP proves to be highly predictive for the probe’s future performance, ameliorating the sales forecasts of all state-of-the-art models on the recent VISUELLE fast-fashion dataset. We also show that POP reflects the ground-truth popularity of new styles (ensembles of clothing items) on the Fashion Forward benchmark, demonstrating that our webly-learned signal is a truthful expression of popularity, accessible by everyone and generalizable to any time of analysis. Forecasting code, data and the POP time series are available at: https://github.com/HumaticsLAB/POP-Mining-POtential-Performance"



Paperid:1565
Authors:Alessio Sampieri; Guido Maria D’Amely di Melendugno; Andrea Avogaro; Federico Cunico; Francesco Setti; Geri Skenderi; Marco Cristani; Fabio Galasso
Title: Pose Forecasting in Industrial Human-Robot Collaboration
Abstract:
Pushing back the frontiers of collaborative robots in industrial environments, we propose a new Separable-Sparse Graph Convolutional Network (SeS-GCN) for pose forecasting. For the first time, SeS-GCN bottlenecks the interaction of the spatial, temporal and channel-wise dimensions in GCNs, and it learns sparse adjacency matrices by a teacher-student framework. Compared to the state-of-the-art, it only uses 1.72% of the parameters and it is 4 times faster, while still performing comparably in forecasting accuracy on Human3.6M at 1 second in the future, which enables cobots to be aware of human operators. As a second contribution, we present a new benchmark of Cobots and Humans in Industrial COllaboration (CHICO). CHICO includes multi-view videos, 3D poses and trajectories of 20 human operators and cobots, engaging in 7 realistic industrial actions. Additionally, it reports 226 genuine collisions, taking place during the human-cobot interaction. We test SeS-GCN on CHICO for two important perception tasks in robotics: human pose forecasting, where it reaches an average error of 85.3 mm (MPJPE) at 1 sec in the future with a run time of 2.3 msec, and collision detection, by comparing the forecasted human motion with the known cobot motion, obtaining an F1-score of 0.64.



Paperid:1566
Authors:Sathyanarayanan Aakur; Sudeep Sarkar
Title: Actor-Centered Representations for Action Localization in Streaming Videos
Abstract:
Event perception tasks such as recognizing and localizing actions in streaming videos are essential for scaling to real-world application contexts. We tackle the problem of learning actor-centered representations through the notion of continual hierarchical predictive learning to localize actions in streaming videos without the need for training labels and outlines for the objects in the video. We propose a framework driven by the notion of hierarchical predictive learning to construct actor-centered features by attention-based contextualization. The key idea is that predictable features or objects do not attract attention and hence do not contribute to the action of interest. Experiments on three benchmark datasets show that the approach can learn robust representations for localizing actions using only one epoch of training, i.e., a single pass through the streaming video. We show that the proposed approach outperforms unsupervised and weakly supervised baselines while offering competitive performance to fully supervised approaches. Additionally, we extend the model to multi-actor settings to recognize group activities while localizing the multiple, plausible actors. We also show that it generalizes to out-of-domain data with limited performance degradation.



Paperid:1567
Authors:Xiufeng Xie; Ning Zhou; Wentao Zhu; Ji Liu
Title: Bandwidth-Aware Adaptive Codec for DNN Inference Offloading in IoT
Abstract:
The lightweight nature of IoT devices makes it challenging to run deep neural networks (DNNs) locally for applications like augmented reality. Recent advances in IoT communication like LTE-M have significantly boosted the link bandwidth, enabling IoT devices to stream visual data to edge servers running DNNs for inference. However, uncompressed visual data can still easily overload the IoT link, and the wireless spectrum is shared by numerous IoT devices, causing unstable link bandwidth. Mainstream codecs can reduce the traffic but at the cost of severe inference accuracy drops. Recent works on differentiable JPEG train the codec to tackle the damage to inference accuracy. But they rely on heuristic configurations in the loss function to balance the rate-accuracy tradeoff, providing no guarantee to meet the IoT bandwidth constraint. This paper presents AutoJPEG, a bandwidth-aware adaptive compression solution that learns the JPEG encoding parameters to optimize the DNN inference accuracy under bandwidth constraints. We model the compressed image size as a closed-form function of encoding parameters by analyzing the JPEG codec workflow. Furthermore, we formulate a constrained optimization framework to minimize the original DNN loss while ensuring the image size strictly meets the bandwidth constraint. Our evaluation validates AutoJPEG on various DNN models and datasets. In our experiments, AutoJPEG outperforms the mainstream codecs (like JPEG and WebP) and the state-of-the-art solutions that optimize the image codec for DNN inference.



Paperid:1568
Authors:Paritosh Parmar; Amol Gharat; Helge Rhodin
Title: Domain Knowledge-Informed Self-Supervised Representations for Workout Form Assessment
Abstract:
Maintaining proper form while exercising is important for preventing injuries and maximizing muscle mass gains. Detecting errors in workout form naturally requires estimating human’s body pose. However, off-the-shelf pose estimators struggle to perform well on the videos recorded in gym scenarios due to factors such as camera angles, occlusion from gym equipment, illumination, and clothing. To aggravate the problem, the errors to be detected in the workouts are very subtle. To that end, we propose to learn exercise-oriented image and video representations from unlabeled samples such that a small dataset annotated by experts suffices for supervised error detection. In particular, our domain knowledge-informed self-supervised approaches (pose contrastive learning and motion disentangling) exploit the harmonic motion of the exercise actions, and capitalize on the large variances in camera angles, clothes, and illumination to learn powerful representations. To facilitate our self-supervised pretraining, and supervised finetuning, we curated a new exercise dataset, Fitness-AQA (https://github.com/ParitoshParmar/Fitness-AQA), comprising of three exercises: BackSquat, BarbellRow, and OverheadPress. It has been annotated by expert trainers for multiple crucial and typically occurring exercise errors. Experimental results show that our self-supervised representations outperform off-the-shelf 2D- and 3D-pose estimators and several other baselines. We also show that our approaches can be applied to other domains/tasks such as pose estimation and dive quality assessment.



Paperid:1569
Authors:Mohan Zhou; Yalong Bai; Wei Zhang; Ting Yao; Tiejun Zhao; Tao Mei
Title: Responsive Listening Head Generation: A Benchmark Dataset and Baseline
Abstract:
We present a new listening head generation benchmark, for synthesizing responsive feedbacks of a listener (e.g., nod, smile) during a face-to-face conversation. As the indispensable complement to talking heads generation, listening head generation has seldomly been studied in literature. Automatically synthesizing listening behavior that actively responds to a talking head, is critical to applications such as digital human, virtual agents and social robots. In this work, we propose a novel dataset ""ViCo"", highlighting the listening head generation during a face-to-face conversation. A total number of 92 identities (67 speakers and 76 listeners) are involved in ViCo, featuring 483 clips in a paired ""speaking-listening"" pattern, where listeners show three listening styles based on their attitudes: positive, neutral, negative. Different from traditional speech-to-gesture or talking-head generation, listening head generation takes as input both the audio and visual signals from the speaker, and gives non-verbal feedbacks (e.g., head motions, facial expressions) in a real-time manner. Our dataset supports a wide range of applications such as human-to-human interaction, video-to-video translation, cross-modal understanding and generation. To encourage further research, we also release a listening head generation baseline, conditioning on different listening attitudes. Code & ViCo dataset: https://project.mhzhou.com/vico.



Paperid:1570
Authors:Sen Zhang; Jing Zhang; Dacheng Tao
Title: "Towards Scale-Aware, Robust, and Generalizable Unsupervised Monocular Depth Estimation by Integrating IMU Motion Dynamics"
Abstract:
Unsupervised monocular depth and ego-motion estimation has drawn extensive research attention in recent years. Although current methods have reached a high up-to-scale accuracy, they usually fail to learn the true scale metric due to the inherent scale ambiguity from training with monocular sequences. In this work, we tackle this problem and propose DynaDepth, a novel scale-aware framework that integrates information from vision and IMU motion dynamics. Specifically, we first propose an IMU photometric loss and a cross-sensor photometric consistency loss to provide dense supervision and absolute scales. To fully exploit the complementary information from both sensors, we further drive a differentiable camera-centric extended Kalman filter (EKF) to update the IMU preintegrated motions when observing visual measurements. In addition, the EKF formulation enables learning an ego-motion uncertainty measure, which is non-trivial for unsupervised methods. By leveraging IMU during training, DynaDepth not only learns an absolute scale, but also provides a better generalization ability and robustness against vision degradation such as illumination change and moving objects. We validate the effectiveness of DynaDepth by conducting extensive experiments and simulations on the KITTI and Make3D datasets.



Paperid:1571
Authors:Prasun Roy; Subhankar Ghosh; Saumik Bhattacharya; Umapada Pal; Michael Blumenstein
Title: TIPS: Text-Induced Pose Synthesis
Abstract:
In computer vision, human pose synthesis and transfer deal with probabilistic image generation of a person in a previously unseen pose from an already available observation of that person. Though researchers have recently proposed several methods to achieve this task, most of these techniques derive the target pose directly from the desired target image on a specific dataset, making the underlying process challenging to apply in real-world scenarios as the generation of the target image is the actual aim. In this paper, we first present the shortcomings of current pose transfer algorithms and then propose a novel text-based pose transfer technique to address those issues. We divide the problem into three independent stages: (a) text to pose representation, (b) pose refinement, and (c) pose rendering. To the best of our knowledge, this is one of the first attempts to develop a text-based pose transfer framework where we also introduce a new dataset DF-PASS, by adding descriptive pose annotations for the images of the DeepFashion dataset. The proposed method generates promising results with significant qualitative and quantitative scores in our experiments.



Paperid:1572
Authors:Haolin Yuan; Bo Hui; Yuchen Yang; Philippe Burlina; Neil Zhenqiang Gong; Yinzhi Cao
Title: Addressing Heterogeneity in Federated Learning via Distributional Transformation
Abstract:
Federated learning (FL) allows multiple clients to collaboratively train a deep learning model. One major challenge of FL is when data distribution is heterogeneous, i.e., differs from one client to another. Existing personalized FL algorithms are only applicable to narrow cases, e.g., one or two data classes per client, and therefore they do not satisfactorily address FL under varying levels of data heterogeneity. In this paper, we propose a novel framework, called DisTrans, to improve FL performance (i.e., model accuracy) via train and test-time distributional transformations along with a double-input-channel model structure. DisTrans works by optimizing distributional offsets and models for each FL client to shift their data distribution, and aggregates these offsets at the FL server to further improve performance in case of distributional heterogeneity. Our evaluation on multiple benchmark datasets shows that DisTrans outperforms state-of-the-art FL methods and data augmentation methods under various settings and different degrees of client distributional heterogeneity (e.g., for CelebA and 100% heterogeneity DisTrans has accuracy of 80.4% vs. 72.1% or lower for other SOTA approaches).



Paperid:1573
Authors:Shraman Pramanick; Ewa M. Nowara; Joshua Gleason; Carlos D. Castillo; Rama Chellappa
Title: Where in the World Is This Image? Transformer-Based Geo-Localization in the Wild
Abstract:
Predicting the geographic location (geo-localization) from a single ground-level RGB image taken anywhere in the world is a very challenging problem. The challenges include huge diversity of images due to different environmental scenarios, drastic variation in the appearance of the same location depending on the time of the day, weather, season, and more importantly, the prediction is made from a single image possibly having only a few geo-locating cues. For these reasons, most existing works are restricted to specific cities, imagery, or worldwide landmarks. In this work, we focus on developing an efficient solution to planet-scale single-image geo-localization. To this end, we propose TransLocator, a unified dual-branch transformer network that attends to tiny details over the entire image and produces robust feature representation under extreme appearance variations. TransLocator takes an RGB image and its semantic segmentation map as inputs, interacts between its two parallel branches after each transformer layer, and simultaneously performs geo-localization and scene recognition in a multi-task fashion. We evaluate TransLocator on four benchmark datasets - Im2GPS, Im2GPS3k, YFCC4k, YFCC26k and obtain 5.5%, 14.1%, 4.9%, 9.9% continent-level accuracy improvement over the state-of-the-art. TransLocator is also validated on real-world test images and found to be more effective than previous methods.



Paperid:1574
Authors:Guannan Guo; Qi Lin; Tao Chen; Zhenghui Feng; Zheng Wang; Jianping Li
Title: Colorization for In Situ Marine Plankton Images
Abstract:
Underwater imaging with red-NIR light illumination can avoid phototropic aggregation-induced observational deviation of marine plankton abundance under white light illumination, but this will lead to the loss of critical color information in the collected grayscale images, which is non-preferable to subsequent human and machine recognition. We present a novel deep networks-based vision system IsPlanktonCLR for automatic colorization of in situ marine plankton images. IsPlanktonCLR uses a reference module to generate self-guidance from a customized palette, which is obtained by clustering in situ plankton image colors. With this self-guidance, a parallel colorization module restores input grayscale images into their true color counterparts. Additionally, a new metric for image colorization evaluation is proposed, which can objectively reflect the color dissimilarity between comparative images. Experiments and comparisons with state-of-the-art approaches are presented to show that our method achieves a substantial improvement over previous methods on color restoration of scientific plankton image data.



Paperid:1575
Authors:Mingyu Yang; Yu Chen; Hun-Seok Kim
Title: Efficient Deep Visual and Inertial Odometry with Adaptive Visual Modality Selection
Abstract:
In recent years, deep learning-based approaches for visual-inertial odometry (VIO) have shown remarkable performance outperforming traditional geometric methods. Yet, all existing methods use both the visual and inertial measurements for every pose estimation incurring potential computational redundancy. While visual data processing is much more expensive than that for the inertial measurement unit (IMU), it may not always contribute to improving the pose estimation accuracy. In this paper, we propose an adaptive deep-learning based VIO method that reduces computational redundancy by opportunistically disabling the visual modality. Specifically, we train a policy network that learns to deactivate the visual feature extractor on the fly based on the current motion state and IMU readings. A Gumbel-Softmax trick is adopted to train the policy network to make the decision process differentiable for end-to-end system training. The learned strategy is interpretable, and it shows scenario-dependent decision patterns for adaptive complexity reduction. Experiment results show that our method achieves a similar or even better performance than the full-modality baseline with up to 78.8% computational complexity reduction for KITTI dataset evaluation. The code is available at https://github.com/mingyuyng/Visual-Selective-VIO.



Paperid:1576
Authors:Patsorn Sangkloy; Wittawat Jitkrittum; Diyi Yang; James Hays
Title: A Sketch Is Worth a Thousand Words: Image Retrieval with Text and Sketch
Abstract:
We address the problem of retrieving in-the-wild images with both a sketch and a text query. We present TASK-former (Text And SKetch transformer), an end-to-end trainable model for image retrieval using a text description and a sketch as input. We argue that both input modalities complement each other in a manner that cannot be achieved easily by either one alone. TASK-former follows the late-fusion dual-encoder approach, similar to CLIP, which allows efficient and scalable retrieval since the retrieval set can be indexed independently of the queries. We empirically demonstrate that using an input sketch (even a poorly drawn one) in addition to text considerably increases retrieval recall compared to traditional text-based image retrieval. To evaluate our approach, we collect 5,000 hand-drawn sketches for images in the test set of the COCO dataset. The collected sketches are available a https://janesjanes.github.io/tsbir/.



Paperid:1577
Authors:Tianyi Liu; Sen He; Vinodh Kumaran Jayakumar; Wei Wang
Title: A Cloud 3D Dataset and Application-Specific Learned Image Compression in Cloud 3D
Abstract:
In Cloud 3D, such as Cloud Gaming and Cloud Virtual Reality (VR), image frames are rendered and compressed (encoded) in the cloud, and sent to the clients for users to view. For low latency and high image quality, fast, high compression rate, and high-quality image compression techniques are preferable. This paper explores computation time reduction techniques for learned image compression to make it more suitable for cloud 3D. More specifically, we employed slim (low-complexity) and application-specific AI models to reduce the computation time without degrading image quality. Our approach is based on two key insights: (1) as the frames generated by a 3D application are highly homogeneous, application-specific compression models can improve the rate-distortion performance over a general model; (2) many computer-generated frames from 3D applications are less complex than natural photos, which makes it feasible to reduce the model complexity to accelerate compression computation. We evaluated our models on six gaming image datasets. The results show that our approach has similar rate-distortion performance as a state-of-the-art learned image compression algorithm, while obtaining about 5x to 9x speedup and reducing the compression time to be less than 1 second (0.74s), bringing learned image compression closer to being viable for cloud 3D. Code is available at https://github.com/cloud-graphics-rendering/AppSpecificLIC.



Paperid:1578
Authors:Yaojie Shen; Libo Zhang; Kai Xu; Xiaojie Jin
Title: AutoTransition: Learning to Recommend Video Transition Effects
Abstract:
Video transition effects are widely used in video editing to connect shots for creating cohesive and visually appealing videos. However, it is challenging for non-professionals to choose best transitions due to the lack of cinematographic knowledge and design skills. In this paper, we present the premier work on performing automatic video transitions recommendation (VTR): given a sequence of raw video shots and companion audio, recommend video transitions for each pair of neighboring shots. To solve this task, we collect a large-scale video transition dataset using publicly available video templates on editing softwares. Then we formulate VTR as a multi-modal retrieval problem from vision/audio to video transitions and propose a novel multi-modal matching framework which consists of two parts. First we learn the embedding of video transitions through a video transition classification task. Then we propose a model to learn the matching correspondence from vision/audio inputs to video transitions. Specifically, the proposed model employs a multi-modal transformer to fuse vision and audio information, as well as capture the context cues in sequential transition outputs. Through both quantitative and qualitative experiments, we clearly demonstrate the effectiveness of our method. Notably, in the comprehensive user study, our method receives comparable scores compared with professional editors while improving the video editing efficiency by 300x. We hope our work serves to inspire other researchers to work on this new task. The dataset and codes are public at https://github.com/acherstyx/AutoTransition.



Paperid:1579
Authors:Romain Loiseau; Mathieu Aubry; Loïc Landrieu
Title: Online Segmentation of LiDAR Sequences: Dataset and Algorithm
Abstract:
Roof-mounted spinning LiDAR sensors are widely used by autonomous vehicles. However, most semantic datasets and algorithms used for LiDAR sequence segmentation operate on 360° frames, causing an acquisition latency incompatible with real-time applications. To address this issue, we first introduce HelixNet, a 10 billion point dataset with fine-grained labels, timestamps, and sensor rotation information necessary to accurately assess the real-time readiness of segmentation algorithms. Second, we propose Helix4D, a compact and efficient spatio-temporal transformer architecture specifically designed for rotating LiDAR sequences. Helix4D operates on acquisition slices corresponding to a fraction of a full sensor rotation, significantly reducing the total latency. Helix4D reaches accuracy on par with the best segmentation algorithms on HelixNet and SemanticKITTI with a reduction of over 5x in terms of latency and 50x in model size. The code and data are available at: https://romainloiseau.fr/helixnet"



Paperid:1580
Authors:Jun Cen; Peng Yun; Shiwei Zhang; Junhao Cai; Di Luan; Mingqian Tang; Ming Liu; Michael Yu Wang
Title: Open-World Semantic Segmentation for LIDAR Point Clouds
Abstract:
Classical LIDAR semantic segmentation is not robust for real-world applications, e.g., autonomous driving, since it is closed-set and static. The closed-set network is only able to output labels of trained classes, even for objects never seen before, while a static network cannot update its knowledge base according to what it has seen. Therefore, we propose the open-world semantic segmentation task for LIDAR point clouds, which aims to 1) identify both old and novel classes using open-set semantic segmentation, and 2) gradually incorporate novel objects into the existing knowledge base using incremental learning without forgetting old classes. We propose a REdundAncy cLassifier (REAL) framework to provide a general architecture for both open-set semantic segmentation and incremental learning. The experimental results show that REAL can achieves state-of-the-art performance in the open-set semantic segmentation task on the SemanticKITTI and nuScenes datasets, and alleviate the catastrophic forgetting with a large margin during incremental learning.



Paperid:1581
Authors:Niklas Hanselmann; Katrin Renz; Kashyap Chitta; Apratim Bhattacharyya; Andreas Geiger
Title: KING: Generating Safety-Critical Driving Scenarios for Robust Imitation via Kinematics Gradients
Abstract:
Simulators offer the possibility of safe, low-cost development of self-driving systems. However, current driving simulators exhibit naïve behavior models for background traffic. Hand-tuned scenarios are typically added during simulation to induce safety-critical situations. An alternative approach is to adversarially perturb the background traffic trajectories. In this paper, we study this approach to safety-critical driving scenario generation using the CARLA simulator. We use a kinematic bicycle model as a proxy to the simulator’s true dynamics and observe that gradients through this proxy model are sufficient for optimizing the background traffic trajectories. Based on this finding, we propose KING, which generates safety-critical driving scenarios with a 20% higher success rate than black-box optimization. By solving the scenarios generated by KING using a privileged rule-based expert algorithm, we obtain training data for an imitation learning policy. After fine-tuning on this new data, we show that the policy becomes better at avoiding collisions. Importantly, our generated data leads to reduced collisions on both held-out scenarios generated via KING as well as traditional hand-crafted scenarios, demonstrating improved robustness.



Paperid:1582
Authors:Tarasha Khurana; Peiyun Hu; Achal Dave; Jason Ziglar; David Held; Deva Ramanan
Title: Differentiable Raycasting for Self-Supervised Occupancy Forecasting
Abstract:
Motion planning for safe autonomous driving requires learning how the environment around an ego-vehicle evolves with time. Ego-centric perception of driveable regions in a scene not only changes with the motion of actors in the environment, but also with the movement of the ego-vehicle itself. Self-supervised representations proposed for large-scale planning, such as ego-centric freespace, confound these two motions, making the representation difficult to use for downstream motion planners. In this paper, we use geometric occupancy as a natural alternative to view-dependent representations such as freespace. Occupancy maps naturally disentagle the motion of the environment from the motion of the ego-vehicle. However, one cannot directly observe the full 3D occupancy of a scene (due to occlusion), making it difficult to use as a signal for learning. Our key insight is to use differentiable raycasting to ""render"" future occupancy predictions into future LiDAR sweep predictions, which can be compared with ground-truth sweeps for self-supervised learning. The use of differentiable raycasting allows occupancy to emerge as an internal representation within the forecasting network. In the absence of groundtruth occupancy, we quantitatively evaluate the forecasting of raycasted LiDAR sweeps and show improvements of upto 15 F1 points. For downstream motion planners, where emergent occupancy can be directly used to guide non-driveable regions, this representation relatively reduces the number of collisions with objects by up to 17% as compared to freespace-centric motion planners.



Paperid:1583
Authors:Taotao Jing; Haifeng Xia; Renran Tian; Haoran Ding; Xiao Luo; Joshua Domeyer; Rini Sherony; Zhengming Ding
Title: InAction: Interpretable Action Decision Making for Autonomous Driving
Abstract:
Autonomous driving has attracted interest for interpretable action decision models that mimic human cognition. Existing interpretable autonomous driving models explore static human explanations, which ignore the implicit visual semantics that are not explicitly annotated or even consistent across annotators. In this paper, we propose a novel Interpretable Action decision making (InAction) model to provide an enriched explanation from both explicit human annotation and implicit visual semantics. First, a proposed visual-semantic module captures the region-based action-inducing components from the visual inputs, which learns the implicit visual semantics to provide a human-understandable explanation in action decision making. Second, an explicit reasoning module is developed by incorporating global visual features and action-inducing visual semantics, which aims to jointly align the human-annotated explanation and action decision making. Experimental results on two autonomous driving benchmarks demonstrate the effectiveness of our InAction model for explaining both implicitly and explicitly by comparing it to existing interpretable autonomous driving models.



Paperid:1584
Authors:Jyh-Jing Hwang; Henrik Kretzschmar; Joshua Manela; Sean Rafferty; Nicholas Armstrong-Crews; Tiffany Chen; Dragomir Anguelov
Title: CramNet: Camera-Radar Fusion with Ray-Constrained Cross-Attention for Robust 3D Object Detection
Abstract:
Robust 3D object detection is critical for safe autonomous driving. Camera and radar sensors are synergistic as they capture complementary information and work well under different environmental conditions. Fusing camera and radar data is challenging, however, as each of the sensors lacks information along a perpendicular axis, that is, depth is unknown to camera and elevation is unknown to radar. We propose the camera-radar matching network CramNet, an efficient approach to fuse the sensor readings from camera and radar in a joint 3D space. To leverage radar range measurements for better camera depth predictions, we propose a novel ray-constrained cross-attention mechanism that resolves the ambiguity in the geometric correspondences between camera features and radar features. Our method supports training with sensor modality dropout, which leads to robust 3D object detection, even when a camera or radar sensor suddenly malfunctions on a vehicle. We demonstrate the effectiveness of our fusion approach through extensive experiments on the RADIATE dataset, one of the few large-scale datasets that provide radar radio frequency imagery. A camera-only variant of our method achieves competitive performance in monocular 3D object detection on the Waymo Open Dataset.



Paperid:1585
Authors:Kaican Li; Kai Chen; Haoyu Wang; Lanqing Hong; Chaoqiang Ye; Jianhua Han; Yukuai Chen; Wei Zhang; Chunjing Xu; Dit-Yan Yeung; Xiaodan Liang; Zhenguo Li; Hang Xu
Title: CODA: A Real-World Road Corner Case Dataset for Object Detection in Autonomous Driving
Abstract:
Contemporary deep-learning object detection methods for autonomous driving usually assume prefixed categories of common traffic participants, such as pedestrians and cars. Most existing detectors are unable to detect uncommon objects and corner cases (e.g., a dog crossing a street), which may lead to severe accidents in some situations, making the timeline for the real-world application of reliable autonomous driving uncertain. One main reason that impedes the development of truly reliably self-driving systems is the lack of public datasets for evaluating the performance of object detectors on corner cases. Hence, we introduce a challenging dataset named CODA that exposes this critical problem of vision-based detectors. The dataset consists of 1500 carefully selected real-world driving scenes, each containing four object-level corner cases (on average), spanning more than 30 object categories. On CODA, the performance of standard object detectors trained on large-scale autonomous driving datasets significantly drops to no more than 12.8% in mAR. Moreover, we experiment with the state-of-the-art open-world object detector and find that it also fails to reliably identify the novel objects in CODA, suggesting that a robust perception system for autonomous driving is probably still far from reach. We expect our CODA dataset to facilitate further research in reliable detection for real-world autonomous driving. Our dataset is available at https://coda-dataset.github.io.



Paperid:1586
Authors:Mahyar Najibi; Jingwei Ji; Yin Zhou; Charles R. Qi; Xinchen Yan; Scott Ettinger; Dragomir Anguelov
Title: Motion Inspired Unsupervised Perception and Prediction in Autonomous Driving
Abstract:
Learning-based perception and prediction modules in modern autonomous driving systems typically rely on expensive human annotation and are designed to perceive only a handful of predefined object categories. This closed-set paradigm is insufficient for the safety-critical autonomous driving task, where the autonomous vehicle needs to process arbitrarily many types of traffic participants and their motion behaviors in a highly dynamic world. To address this difficulty, this paper pioneers a novel and challenging direction, i.e., training perception and prediction models to understand open-set moving objects, with no human supervision. Our proposed framework uses self-learned flow to trigger an automated meta labeling pipeline to achieve automatic supervision. 3D detection experiments on the Waymo Open Dataset show that our method significantly outperforms classical unsupervised approaches and is even competitive to the counterpart with supervised scene flow. We further show that our approach generates highly promising results in open-set 3D detection and trajectory prediction, confirming its potential in closing the safety gap of fully supervised systems.



Paperid:1587
Authors:Adil Kaan Akan; Fatma Güney
Title: StretchBEV: Stretching Future Instance Prediction Spatially and Temporally
Abstract:
In self-driving, predicting future in terms of location and motion of all the agents around the vehicle is a crucial requirement for planning. Recently, a new joint formulation of perception and prediction has emerged by fusing rich sensory information perceived from multiple cameras into a compact bird’s-eye view representation to perform prediction. However, the quality of future predictions degrades over time while extending to longer time horizons due to multiple plausible predictions. In this work, we address this inherent uncertainty in future predictions with a stochastic temporal model. Our model learns temporal dynamics in a latent space through stochastic residual updates at each time step. By sampling from a learned distribution at each time step, we obtain more diverse future predictions that are also more accurate compared to previous work, especially stretching both spatially further regions in the scene and temporally over longer time horizons. Despite separate processing of each time step, our model is still efficient through decoupling of the learning of dynamics and the generation of future predictions.



Paperid:1588
Authors:Shenghua Xu; Xinyue Cai; Bin Zhao; Li Zhang; Hang Xu; Yanwei Fu; Xiangyang Xue
Title: RCLane: Relay Chain Prediction for Lane Detection
Abstract:
Lane detection is an important component of many real-world autonomous systems. Despite a wide variety of lane detection approaches have been proposed, reporting steady benchmark improvements over time, lane detection remains a largely unsolved problem. This is because most of the existing lane detection methods either treat the lane detection as a dense prediction or a detection task, few of them consider the unique topologies (Y-shape, Fork-shape, nearly horizontal lane) of the lane markers, which leads to sub-optimal solution. In this paper, we present a new method for lane detection based on relay chain prediction. Specifically, our model predicts a segmentation map to classify the foreground and background region. For each pixel point in the foreground region, we go through the forward branch and backward branch to recover the whole lane. Each branch decodes a transfer map and a distance map to produce the direction moving to the next point, and how many steps to progressively predict a relay station (next point). As such, our model is able to capture the keypoints along the lanes. Despite its simplicity, our strategy allows us to establish new state-of-the-art on four major benchmarks including TuSimple, CULane, CurveLanes and LLAMAS.



Paperid:1589
Authors:Antonin Vobecky; David Hurych; Oriane Siméoni; Spyros Gidaris; Andrei Bursuc; Patrick Pérez; Josef Sivic
Title: Drive\&Segment: Unsupervised Semantic Segmentation of Urban Scenes via Cross-Modal Distillation
Abstract:
This work investigates learning pixel-wise semantic image segmentation in urban scenes without any manual annotation, just from the raw non-curated data collected by cars which, equipped with cameras and LiDAR sensors, drive around a city. Our contributions are threefold. First, we propose a novel method for cross-modal unsupervised learning of semantic image segmentation by leveraging synchronized LiDAR and image data. The key ingredient of our method is the use of an object proposal module that analyzes the LiDAR point cloud to obtain proposals for spatially consistent objects. Second, we show that these 3D object proposals can be aligned with the input images and reliably clustered into semantically meaningful pseudo-classes. Finally, we develop a cross-modal distillation approach that leverages image data partially annotated with the resulting pseudo-classes to train a transformer-based model for image semantic segmentation. We show the generalization capabilities of our method by testing on four different testing datasets (Cityscapes, Dark Zurich, Nighttime Driving and ACDC) without any finetuning, and demonstrate significant improvements compared to the current state of the art on this problem.



Paperid:1590
Authors:Zixiang Zhou; Xiangchen Zhao; Yu Wang; Panqu Wang; Hassan Foroosh
Title: CenterFormer: Center-based Transformer for 3D Object Detection
Abstract:
Query-based transformer has shown great potential in constructing long-range attention in many image-domain tasks, but has rarely been considered in LiDAR-based 3D object detection due to the overwhelming size of the point cloud data. In this paper, we propose CenterFormer, a center-based transformer network for 3D object detection. CenterFormer first uses a center heatmap to select center candidates on top of a standard voxel-based point cloud encoder. It then uses the feature of the center candidate as the query embedding in the transformer. To further aggregate features from multiple frames, we design an approach to fuse features through cross-attention. Lastly, regression heads are added to predict the bounding box on the output center feature representation. Our design reduces the convergence difficulty and computational complexity of the transformer structure. The results show significant improvements over the strong baseline of anchor-free object detection networks. CenterFormer achieves state-of-the-art performance for a single model on the Waymo Open Dataset, with 73.7% mAPH on the validation set and 75.6% mAPH on the test set, significantly outperforming all previously published CNN and transformer-based methods. Our code is publicly available at https://github.com/TuSimple/centerformer"



Paperid:1591
Authors:Zhiyuan Cheng; James Liang; Hongjun Choi; Guanhong Tao; Zhiwen Cao; Dongfang Liu; Xiangyu Zhang
Title: Physical Attack on Monocular Depth Estimation with Optimal Adversarial Patches
Abstract:
Deep learning has substantially boosted the performance of Monocular Depth Estimation (MDE), a critical component in fully vision-based autonomous driving (AD) systems (e.g., Tesla and Toyota). In this work, we develop an attack against learning-based MDE. In particular, we use an optimization-based method to systematically generate stealthy physical-object-oriented adversarial patches to attack depth estimation. We balance the stealth and effectiveness of our attack with object-oriented adversarial design, sensitive region localization, and natural style camouflage. Using real-world driving scenarios, we evaluate our attack on concurrent MDE models and a representative downstream task for AD (i.e., 3D object detection). Experimental results show that our method can generate stealthy, effective, and robust adversarial patches for different target objects and models and achieves more than 6 meters mean depth estimation error and 93% attack success rate (ASR) in object detection with a patch of 1/9 of the vehicle’s rear area. Field tests on three different driving routes with a real vehicle indicate that we cause over 6 meters mean depth estimation error and reduce the object detection rate from 90.70% to 5.16% in continuous video frames.



Paperid:1592
Authors:Shengchao Hu; Li Chen; Penghao Wu; Hongyang Li; Junchi Yan; Dacheng Tao
Title: ST-P3: End-to-End Vision-Based Autonomous Driving via Spatial-Temporal Feature Learning
Abstract:
Many existing autonomous driving paradigms involve a multi-stage discrete pipeline of tasks. To better predict the control signals and enhance user safety, an end-to-end approach that benefits from joint spatial-temporal feature learning is desirable. While there are some pioneering works on LiDAR-based input or implicit design, in this paper we formulate the problem in an interpretable vision-based setting. In particular, we propose a spatial-temporal feature learning scheme towards a set of more representative features for perception, prediction and planning tasks simultaneously, which is called ST-P3. Specifically, an egocentric-aligned accumulation technique is proposed to preserve geometry information in 3D space before the bird’s eye view transformation for perception; a dual pathway modeling is devised to take past motion variations into account for future prediction; a temporal-based refinement unit is introduced to compensate for recognizing vision-based elements for planning. To the best of our knowledge, we are the first to systematically investigate each part of an interpretable end-to-end vision-based autonomous driving system. We benchmark our approach against previous state-of-the-arts on both open-loop nuScenes dataset as well as closed-loop CARLA simulation. The results show the effectiveness of our method. Source code, model and protocol details are made publicly available at https://github.com/OpenPerceptionX/ST-P3.



Paperid:1593
Authors:Li Chen; Chonghao Sima; Yang Li; Zehan Zheng; Jiajie Xu; Xiangwei Geng; Hongyang Li; Conghui He; Jianping Shi; Yu Qiao; Junchi Yan
Title: PersFormer: 3D Lane Detection via Perspective Transformer and the OpenLane Benchmark
Abstract:
Methods for 3D lane detection have been recently proposed to address the issue of inaccurate lane layouts in many autonomous driving scenarios (uphill/downhill, bump, etc.). Previous work struggled in complex cases due to their simple designs of the spatial transformation between front view and bird’s eye view (BEV) and the lack of a realistic dataset. Towards these issues, we present PersFormer: an end-to-end monocular 3D lane detector with a novel Transformer-based spatial feature transformation module. Our model generates BEV features by attending to related front-view local regions with camera parameters as a reference. PersFormer adopts a unified 2D/3D anchor design and an auxiliary task to detect 2D/3D lanes simultaneously, enhancing the feature consistency and sharing the benefits of multi-task learning. Moreover, we release one of the first large-scale real-world 3D lane datasets: OpenLane, with high-quality annotation and scenario diversity. OpenLane contains 200,000 frames, over 880,000 instance-level lanes, 14 lane categories, along with scene tags and the closed-in-path object annotations to encourage the development of lane detection and more industrial-related autonomous driving methods. We show that PersFormer significantly outperforms competitive baselines in the 3D lane detection task on our new OpenLane dataset as well as Apollo 3D Lane Synthetic dataset, and is also on par with state-of-the-art algorithms in the 2D task on OpenLane. The project page is available at https://github.com/OpenPerceptionX/PersFormer_3DLane and OpenLane dataset is provided at https://github.com/OpenPerceptionX/OpenLane.



Paperid:1594
Authors:Kwonyoung Kim; Jungin Park; Jiyoung Lee; Dongbo Min; Kwanghoon Sohn
Title: PointFix: Learning to Fix Domain Bias for Robust Online Stereo Adaptation
Abstract:
Online stereo adaptation tackles the domain shift problem, caused by different environments between synthetic (training) and real (test) datasets, to promptly adapt stereo models in dynamic real-world applications such as autonomous driving. However, previous methods often fail to counteract particular regions related to dynamic objects with more severe environmental changes. To mitigate this issue, we propose to incorporate an auxiliary point-selective network into a meta-learning framework, called PointFix, to provide a robust initialization of stereo models for online stereo adaptation. In a nutshell, our auxiliary network learns to fix local variants intensively by effectively back-propagating local information through the meta-gradient for the robust initialization of the baseline model. This network is model-agnostic, so can be used in any kind of architectures in a plug-and-play manner. We conduct extensive experiments to verify the effectiveness of our method under three adaptation settings such as short-, mid-, and long-term sequences. Experimental results show that the proper initialization of the base stereo model by the auxiliary network enables our learning paradigm to achieve state-of-the-art performance at inference.



Paperid:1595
Authors:Wencheng Han; Junbo Yin; Xiaogang Jin; Xiangdong Dai; Jianbing Shen
Title: BRNet: Exploring Comprehensive Features for Monocular Depth Estimation
Abstract:
Self-supervised monocular depth estimation has achieved promising performance recently. A consensus is that high-resolution inputs often yield better results. However, we find that the performance gap between high and low resolutions lies in the inappropriate feature representation of the widely used U-Net backbone. In this paper, we address the comprehensive feature representation problem for self-supervised depth estimation.i.e paying attention to both local and global feature representation. Specifically, we first provide an in-depth analysis of the influence of different input resolutions and find out that the receptive fields play a more crucial role than the information disparity between inputs. To this end, we propose a bilateral depth encoder that can fully exploit detailed and global information. It benefits from more broad receptive fields and thus achieves substantial improvements. Furthermore, we propose a residual decoder to facilitate depth regression as well as save computations by focusing on the information difference between different layers. We named our new depth estimation model Bilateral Residual Depth Network (BRNet). Experimental results show that BRNet achieves new state-of-the-art performance on the KITTI benchmark with three types of self-supervision.



Paperid:1596
Authors:Zhenyao Wu; Xinyi Wu; Xiaoping Zhang; Lili Ju; Song Wang
Title: SiamDoGe: Domain Generalizable Semantic Segmentation Using Siamese Network
Abstract:
Deep learning-based approaches usually suffer from performance drop on out-of-distribution samples, therefore domain generalization is often introduced to improve the robustness of deep models. Domain randomization (DR) is a common strategy to improve the generalization capability of semantic segmentation networks, however, existing DR-based algorithms require collecting auxiliary domain images to stylize the training samples. In this paper, we propose a novel domain generalizable semantic segmentation method, “SiamDoGe”, which builds upon a DR approach without using auxiliary domains and employs a Siamese architecture to learn domain-agnostic features from the training dataset. Particularly, the proposed method takes two augmented versions of each training sample as input and produces the corresponding predictions in parallel. Throughout this process, the features from each branch are randomized by those from the other to enhance the feature diversity of training samples. Then the predictions produced from the two branches are enforced to be consistent conditioned on feature sensitivity. Extensive experiment results demonstrate the proposed method exhibits better generalization ability than existing state-of-the-arts across various unseen target domains.



Paperid:1597
Authors:Gur-Eyal Sela; Ionel Gog; Justin Wong; Kumar Krishna Agrawal; Xiangxi Mo; Sukrit Kalra; Peter Schafhalter; Eric Leong; Xin Wang; Bharathan Balaji; Joseph Gonzalez; Ion Stoica
Title: Context-Aware Streaming Perception in Dynamic Environments
Abstract:
Efficient vision works maximize accuracy under a latency budget. These works evaluate accuracy offline, one image at a time. However, real-time vision applications like autonomous driving operate in streaming settings, where ground truth changes between inference start and finish. This results in a significant accuracy drop. Therefore, a recent work proposed to maximize accuracy in streaming settings on average. In this paper, we propose to maximize streaming accuracy for every environment context. We posit that scenario difficulty influences the initial (offline) accuracy difference, while obstacle displacement in the scene affects the subsequent accuracy degradation. Our method, Octopus, uses these scenario properties to select configurations that maximize streaming accuracy at test time. Our method improves tracking performance (S-MOTA) by 7.4% over the conventional static approach. Further, performance improvement using our method comes in addition to, and not instead of, advances in offline accuracy.



Paperid:1598
Authors:Colton Stearns; Davis Rempe; Jie Li; Rareș Ambruș; Sergey Zakharov; Vitor Guizilini; Yanchao Yang; Leonidas J. Guibas
Title: SpOT: Spatiotemporal Modeling for 3D Object Tracking
Abstract:
3D multi-object tracking aims to uniquely and consistently identify all mobile entities through time. Despite the rich spatiotemporal information available in this setting, current 3D tracking methods primarily rely on abstracted information and limited history, e.g. single-frame object bounding boxes. In this work, we develop a holistic representation of traffic scenes that leverages both spatial and temporal information of the actors in the scene. Specifically, we reformulate tracking as a spatiotemporal problem by representing tracked objects as sequences of time-stamped points and bounding boxes over a long temporal history. At each timestamp, we improve the location and motion estimates of our tracked objects through learned refinement over the full sequence of object history. By considering time and space jointly, our representation naturally encodes fundamental physical priors such as object permanence and consistency across time. Our spatiotemporal tracking framework achieves state-of-the-art performance on the Waymo and nuScenes benchmarks.



Paperid:1599
Authors:Chang Liu; Xiaoyan Qian; Binxiao Huang; Xiaojuan Qi; Edmund Lam; Siew-Chong Tan; Ngai Wong
Title: Multimodal Transformer for Automatic 3D Annotation and Object Detection
Abstract:
Despite a growing number of datasets being collected for training 3D object detection models, significant human effort is still required to annotate 3D boxes on LiDAR scans. To automate the annotation and facilitate the production of various customized datasets, we propose an end-to-end multimodal transformer (MTrans) autolabeler, which leverages both LiDAR scans and images to generate precise 3D box annotations from weak 2D bounding boxes. To alleviate the pervasive sparsity problem that hinders existing autolabelers, MTrans densifies the sparse point clouds by generating new 3D points based on 2D image information. With a multi-task design, MTrans segments the foreground/background, densifies LiDAR point clouds, and regresses 3D boxes simultaneously. Experimental results verify the effectiveness of the MTrans for improving the quality of the generated labels. By enriching the sparse point clouds, our method achieves 4.48\% and 4.03\% better 3D AP on KITTI moderate and hard samples, respectively, versus the state-of-the-art autolabeler. MTrans can also be extended to improve the accuracy for 3D object detection, resulting in a remarkable 89.45\% AP on KITTI hard samples. Codes are at \url{https://github.com/Cliu2/MTrans}.



Paperid:1600
Authors:Shengyu Huang; Zan Gojcic; Jiahui Huang; Andreas Wieser; Konrad Schindler
Title: Dynamic 3D Scene Analysis by Point Cloud Accumulation
Abstract:
Multi-beam LiDAR sensors, as used on autonomous vehicles and mobile robots, acquire sequences of 3D range scans (""frames""). Each frame covers the scene sparsely, due to limited angular scanning resolution and occlusion. The sparsity restricts the performance of downstream processes like semantic segmentation or surface reconstruction. Luckily, when the sensor moves, frames are captured from a sequence of different viewpoints. This provides complementary information and, when accumulated in a common scene coordinate frame, yields a denser sampling and a more complete coverage of the underlying 3D scene. However, often the scanned scenes contain moving objects. Points on those objects are not correctly aligned by just undoing the scanner’s ego-motion. In the present paper, we explore multi-frame point cloud accumulation as a mid-level representation of 3D scan sequences, and develop a method that exploits inductive biases of outdoor street scenes, including their geometric layout and object-level rigidity. Compared to state-of-the-art scene flow estimators, our proposed approach aims to align all 3D points in a common reference frame correctly accumulating the points on the individual objects. Our approach greatly reduces the alignment errors on several benchmark datasets. Moreover, the accumulated point clouds benefit high-level tasks like surface reconstruction.



Paperid:1601
Authors:Xin Li; Botian Shi; Yuenan Hou; Xingjiao Wu; Tianlong Ma; Yikang Li; Liang He
Title: Homogeneous Multi-modal Feature Fusion and Interaction for 3D Object Detection
Abstract:
Multi-modal 3D object detection has been an active research topic in autonomous driving. Nevertheless, it is non-trivial to explore the cross-modal feature fusion between sparse 3D points and dense 2D pixels. Recent approaches either fuse the image features with the point cloud features that are projected onto the 2D image plane or combine the sparse point cloud with dense image pixels. These fusion approaches often suffer from severe information loss, thus causing sub-optimal performance. To address these problems, we construct the homogeneous structure between the point cloud and images to avoid projective information loss by transforming the camera features into the LiDAR 3D space. In this paper, we propose a homogeneous multi-modal feature fusion and interaction method (HMFI) for 3D object detection. Specifically, we first design an image voxel lifter module (IVLM) to lift 2D image features into the 3D space and generate homogeneous image voxel features. Then, we fuse the voxelized point cloud features with the image features from different regions by introducing the self-attention based query fusion mechanism (QFM). Next, we propose a voxel feature interaction module (VFIM) to enforce the consistency of semantic information from identical objects in the homogeneous point cloud and image voxel representations, which can provide object-level alignment guidance for cross-modal feature fusion and strengthen the discriminative ability in complex backgrounds. We conduct extensive experiments on the KITTI and Waymo Open Dataset, and the proposed HMFI achieves better performance compared with the state-of-the-art multi-modal methods. Particularly, for the 3D detection of cyclist on the KITTI benchmark, HMFI surpasses all the published algorithms by a large margin.



Paperid:1602
Authors:Haimei Zhao; Jing Zhang; Sen Zhang; Dacheng Tao
Title: "JPerceiver: Joint Perception Network for Depth, Pose and Layout Estimation in Driving Scenes"
Abstract:
Depth estimation, visual odometry (VO), and bird’s-eye-view (BEV) scene layout estimation present three critical tasks for driving scene perception, which is fundamental for motion planning and navigation in autonomous driving. Though they are complementary to each other, prior works usually focus on each individual task and rarely deal with all three tasks together. A naive way is to accomplish them independently in a sequential or parallel manner, but there are three drawbacks, i.e., 1) the depth and VO results suffer from the inherent scale ambiguity issue; 2) the BEV layout is usually estimated separately for roads and vehicles, while the explicit overlay-underlay relations between them are ignored; and 3) the BEV layout is directly predicted from the front-view image without using any depth-related information, although the depth map contains useful geometry clues for inferring scene layouts. In this paper, we address these issues by proposing a novel joint perception framework named JPerceiver, which can simultaneously estimate scale-aware depth and VO as well as BEV layout from a monocular video sequence. It exploits the cross-view geometric transformation (CGT) to propagate the absolute scale from the road layout to depth and VO based on a carefully-designed scale loss. Meanwhile, a cross-view and cross-modal transfer (CCT) module is devised to leverage the depth clues for reasoning road and vehicle layout through an attention mechanism. JPerceiver can be trained in an end-to-end multi-task learning way, where the CGT scale loss and CCT module promote inter-task knowledge transfer to benefit feature learning of each task. Experiments on Argoverse, Nuscenes and KITTI show the superiority of JPerceiver over existing methods on all the above three tasks in terms of accuracy, model size, and inference speed. The code and models are available at \href{https://github.com/sunnyHelen/JPerceiver}{https://github.com/sunnyHelen/JPerceiver}.



Paperid:1603
Authors:Junbo Yin; Jin Fang; Dingfu Zhou; Liangjun Zhang; Cheng-Zhong Xu; Jianbing Shen; Wenguan Wang
Title: Semi-Supervised 3D Object Detection with Proficient Teachers
Abstract:
Dominated point cloud-based 3D object detectors in autonomous driving scenarios rely heavily on the huge amount of accurately labeled samples, however, 3D annotation in the point cloud is extremely tedious, expensive and time-consuming. To reduce the dependence on large supervision, semi-supervised learning (SSL) based approaches have been proposed. The Pseudo-Labeling methodology is commonly used for SSL frameworks, however, the low-quality predictions from the teacher model have seriously limited its performance. In this work, we propose a new Pseudo-Labeling framework for semi-supervised 3D object detection, by enhancing the teacher model to a proficient one with several necessary designs. First, to improve the recall of pseudo labels, a Spatialtemporal Ensemble (STE) module is proposed to generate sufficient seed boxes. Second, to improve the precision of recalled boxes, a Clusteringbased Box Voting (CBV) module is designed to get aggregated votes from the clustered seed boxes. This also eliminates the necessity of sophisticated thresholds to select pseudo labels. Furthermore, to reduce the negative influence of wrong pseudo-labeled samples during the training, a soft supervision signal is proposed by considering Box-wise Contrastive Learning (BCL). The effectiveness of our model is verified on both ONCE and Waymo datasets. For example, on ONCE, our approach significantly improves the baseline by 9.51 mAP. Moreover, with half annotations, our model outperforms the oracle model with full annotations on Waymo.



Paperid:1604
Authors:Zhili Chen; Zian Qian; Sukai Wang; Qifeng Chen
Title: Point Cloud Compression with Sibling Context and Surface Priors
Abstract:
We present a novel octree-based multi-level framework for large-scale point cloud compression, which can organize sparse and unstructured point clouds in a memory-efficient way. In this framework, we propose a new entropy model that explores the hierarchical dependency in an octree using the context of siblings’ children, ancestors, and neighbors to encode the occupancy information of each non-leaf octree node into a bitstream. Moreover, we locally fit quadratic surfaces with a voxel-based geometry-aware module to provide geometric priors in entropy encoding. These strong priors empower our entropy framework to encode the octree into a more compact bitstream. In the decoding stage, we apply a two-step heuristic strategy to restore point clouds with better reconstruction quality. The quantitative evaluation shows that our method outperforms state-of-the-art baselines with a bitrate improvement of 11-16% and 12-14% on the KITTI Odometry and nuScenes datasets, respectively.



Paperid:1605
Authors:Han Zhang; Yunchao Gu; Xinliang Wang; Junjun Pan; Minghui Wang
Title: Lane Detection Transformer Based on Multi-Frame Horizontal and Vertical Attention and Visual Transformer Module
Abstract:
Lane detection requires adequate global information due to the simplicity of lane line features and changeable road scenes. In this paper, we propose a novel lane detection Transformer based on multi-frame input to regress the parameters of lanes under a lane shape modeling. We design a Multi-frame Horizontal and Vertical Attention (MHVA) module to obtain more global features and use Visual Transformer (VT) module to get ""lane tokens"" with interaction information of lane instances. Extensive experiments on two public datasets show that our model can achieve state-of-art results on VIL-100 dataset and comparable performance on Tusimple dataset. In addition, our model runs at 46 fps on multi-frame data while using few parameters, indicating the feasibility and practicability in real-time self-driving applications of our proposed method.



Paperid:1606
Authors:Junbo Yin; Dingfu Zhou; Liangjun Zhang; Jin Fang; Cheng-Zhong Xu; Jianbing Shen; Wenguan Wang
Title: ProposalContrast: Unsupervised Pre-training for LiDAR-Based 3D Object Detection
Abstract:
Existing approaches for unsupervised point cloud pre-training are constrained to either scene-level or point/voxel-level instance discrimination. Scene-level methods tend to lose local details that are crucial for recognizing the road objects, while point/voxel-level methods inherently suffer from limited receptive field that is incapable of perceiving large objects or context environments. Considering region-level representations are more suitable for 3D object detection, we devise a new unsupervised point cloud pre-training framework, called ProposalContrast, that learns robust 3D representations by contrasting region proposals. Specifically, with an exhaustive set of region proposals sampled from each point cloud, geometric point relations within each proposal are modeled for creating expressive proposal representations. To better accommodate 3D detection properties, ProposalContrast optimizes with both inter-cluster and inter-proposal separation, i.e., sharpening the discriminativeness of proposal representations across semantic classes and object instances. The generalizability and transferability of ProposalContrast are verified on various 3D detectors (i.e., PV-RCNN, CenterPoint, PointPillars and PointRCNN) and datasets (i.e., KITTI, Waymo and ONCE).



Paperid:1607
Authors:Chenfeng Xu; Tian Li; Chen Tang; Lingfeng Sun; Kurt Keutzer; Masayoshi Tomizuka; Alireza Fathi; Wei Zhan
Title: PreTraM: Self-Supervised Pre-training via Connecting Trajectory and Map
Abstract:
Deep learning has recently achieved significant progress in trajectory forecasting. However, the scarcity of trajectory data inhibits the data-hungry deep-learning models from learning good representations. While pre-training methods for representation learning exist in computer vision and natural language processing, they still require large-scale data. It is hard to replicate their success in trajectory forecasting due to the inadequate trajectory data (e.g., 34K samples in the nuScenes dataset). To work around the scarcity of trajectory data, we resort to another data modality closely related to trajectories--HD-maps, which are abundantly provided in existing datasets. In this paper, we propose PreTraM, a self-supervised Pre-training scheme via connecting Trajectories and Maps for trajectory forecasting. PreTraM consists of two parts: 1) Trajectory-Map Contrastive Learning, where we project trajectories and maps to a shared embedding space with cross-modal contrastive learning, 2) Map Contrastive Learning, where we enhance map representation with contrastive learning on large quantities of HD-maps. On top of popular baselines such as AgentFormer and Trajectron++, PreTraM reduces their errors by 5.5% and 6.9% relatively on the nuScenes dataset. We show that PreTraM improves data efficiency and scales well with model size.



Paperid:1608
Authors:Nikhil Reddy; Abhinav Singhal; Abhishek Kumar; Mahsa Baktashmotlagh; Chetan Arora
Title: Master of All: Simultaneous Generalization of Urban-Scene Segmentation to All Adverse Weather Conditions
Abstract:
Computer vision systems for autonomous navigation must generalize well in adverse weather and illumination conditions expected in the real world. However, semantic segmentation of images captured in such conditions remains a challenging task for current state-of-the-art (\sota) methods trained on broad daylight images, due to the associated distribution shift. On the other hand, domain adaptation techniques developed for the purpose rely on the availability of the source data, (un)labeled target data and/or its auxiliary information (e.g., \gps). Even then, they typically adapt to a single(specific) target domain(s). To remedy this, we propose a novel, fully test time, adaptation technique, named \textit{Master of ALL} (\mall), for simultaneous generalization to multiple target domains. \mall learns to generalize on unseen adverse weather images from multiple target domains directly at the inference time. More specifically, given a pre-trained model and its parameters, \mall enforces edge consistency prior at the inference stage and updates the model based on (a) a single test sample at a time (\malls), or (b) continuously for the whole test domain (\malld). Not only the target data, \mall also does not need access to the source data and thus, can be used with any pre-trained model. Using a simple model pre-trained on daylight images, \mall outperforms specially designed adverse weather semantic segmentation methods, both in domain generalization and test-time adaptation settings. Our experiments on foggy, snow, night, cloudy, overcast, and rainy conditions demonstrate the target domain-agnostic effectiveness of our approach. We further show that \mall can improve the performance of a model on an adverse weather condition, even when the model is already pre-trained for the specific condition.



Paperid:1609
Authors:Minghua Liu; Yin Zhou; Charles R. Qi; Boqing Gong; Hao Su; Dragomir Anguelov
Title: LESS: Label-Efficient Semantic Segmentation for LiDAR Point Clouds
Abstract:
Semantic segmentation of LiDAR point clouds is an important task in autonomous driving. However, training deep models via conventional supervised methods requires large datasets which are costly to label. It is critical to have label-efficient segmentation approaches to scale up the model to new operational domains or to improve performance on rare cases. While most prior works focus on indoor scenes, we are one of the first to propose a label-efficient semantic segmentation pipeline for outdoor scenes with LiDAR point clouds. Our method co-designs an efficient labeling process with semi/weakly supervised learning and is applicable to nearly any 3D semantic segmentation backbones. Specifically, we leverage geometry patterns in outdoor scenes to have a heuristic pre-segmentation to reduce the manual labeling and jointly design the learning targets with the labeling process. In the learning step, we leverage prototype learning to get more descriptive point embeddings and use multi-scan distillation to exploit richer semantics from temporally aggregated point clouds to boost the performance of single-scan models. Evaluated on the SemanticKITTI and the nuScenes datasets, we show that our proposed method outperforms existing label-efficient methods. With extremely limited human annotations (e.g., 0.1% point labels), our proposed method is even highly competitive compared to the fully supervised counterpart with 100% labels.



Paperid:1610
Authors:Zimin Xia; Olaf Booij; Marco Manfredi; Julian F. P. Kooij
Title: Visual Cross-View Metric Localization with Dense Uncertainty Estimates
Abstract:
This work addresses visual cross-view metric localization for outdoor robotics. Given a ground-level color image and a satellite patch that contains the local surroundings, the task is to identify the location of the ground camera within the satellite patch. Related work addressed this task for range-sensors (LiDAR, Radar), but for vision, only as a secondary regression step after an initial cross-view image retrieval step. Since the local satellite patch could also be retrieved through any rough localization prior (e.g. from GPS/GNSS, temporal filtering), we drop the image retrieval objective and focus on the metric localization only. We devise a novel network architecture with denser satellite descriptors, similarity matching at the bottleneck (rather than at the output as in image retrieval), and a dense spatial distribution as output to capture multi-modal localization ambiguities. We compare against a state-of-the-art regression baseline that uses global image descriptors. Quantitative and qualitative experimental results on the recently proposed VIGOR and the Oxford RobotCar datasets validate our design. The produced probabilities are correlated with localization accuracy, and can even be used to roughly estimate the ground camera’s heading when its orientation is unknown. Overall, our method reduces the median metric localization error by 51%, 37%, and 28% compared to the state-of-the-art when generalizing respectively in the same area, across areas, and across time.



Paperid:1611
Authors:Runsheng Xu; Hao Xiang; Zhengzhong Tu; Xin Xia; Ming-Hsuan Yang; Jiaqi Ma
Title: V2X-ViT: Vehicle-to-Everything Cooperative Perception with Vision Transformer
Abstract:
In this paper, we investigate the application of Vehicle-to-Everything (V2X) communication to improve the perception performance of autonomous vehicles. We present a robust cooperative perception framework with V2X communication using a novel vision Transformer. Specifically, we build a holistic attention model, namely V2X-ViT, to effectively fuse information across on-road agents (i.e., vehicles and infrastructure). V2X-ViT consists of alternating layers of heterogeneous multi-agent self-attention and multi-scale window self-attention, which captures inter-agent interaction and per-agent spatial relationships. These key modules are designed in a unified Transformer architecture to handle common V2X challenges, including asynchronous information sharing, pose errors, and heterogeneity of V2X components. To validate our approach, we create a large-scale V2X perception dataset using CARLA and OpenCDA. Extensive experimental results demonstrate that V2X-ViT sets new state-of-the-art performance for 3D object detection and achieves robust performance even under harsh noisy environments. The code is available at https://github.com/DerrickXuNu/v2x-vit.



Paperid:1612
Authors:Kaichen Zhou; Lanqing Hong; Changhao Chen; Hang Xu; Chaoqiang Ye; Qingyong Hu; Zhenguo Li
Title: DevNet: Self-Supervised Monocular Depth Learning via Density Volume Construction
Abstract:
Self-supervised depth learning from monocular images normally relies on the 2D pixel-wise photometric relation between temporally adjacent image frames. However, they neither fully exploit the 3D point-wise geometric correspondences, nor effectively tackle the ambiguities in the photometric warping caused by occlusions or illumination inconsistency. To address these problems, this work proposes Density Volume Construction Network (DevNet), a novel self-supervised monocular depth learning framework, that can consider 3D spatial information, and exploit stronger geometric constraints among adjacent camera frustums. Instead of directly regressing the pixel value from a single image, our DevNet divides the camera frustum into multiple parallel planes and predicts the pointwise occlusion probability density on each plane. The final depth map is generated by integrating the density along corresponding rays. During the training process, novel regularization strategies and loss functions are introduced to mitigate photometric ambiguities and overfitting. Without obviously enlarging model parameters size or running time, DevNet outperforms several representative baselines on both the KITTI-2015 outdoor dataset and NYU-V2 indoor dataset. In particular, the root-mean-square-deviation is reduced by around 4% with DevNet on both KITTI-2015 and NYU-V2 in the task of depth estimation. Code is available at https://github.com/gitkaichenzhou/DevNet.



Paperid:1613
Authors:Marah Halawa; Olaf Hellwich; Pia Bideau
Title: Action-Based Contrastive Learning for Trajectory Prediction
Abstract:
Trajectory prediction is an essential task for successful human robot interaction, such as in autonomous driving. In this work, we address the problem of predicting future pedestrian trajectories in a first person view setting with a moving camera. To that end, we propose a novel action-based contrastive learning loss, that utilizes pedestrian action information to improve the learned trajectory embeddings. The fundamental idea behind this new loss is that trajectories of pedestrians performing the same action should be closer to each other in the feature space than the trajectories of pedestrians with significantly different actions. In other words, we argue that behavioral information about pedestrian action influences their future trajectory. Furthermore, we introduce a novel sampling strategy for trajectories that is able to effectively increase negative and positive contrastive samples. Additional synthetic trajectory samples are generated using a trained Conditional Variational Autoencoder (CVAE), which is at the core of several models developed for trajectory prediction. Results show that our proposed contrastive framework employs contextual information about pedestrian behavior, i.e. action, effectively, and it learns a better trajectory representation. Thus, integrating the proposed contrastive framework within a trajectory prediction model improves its results and outperforms state-of-the-art methods on three trajectory prediction benchmarks.



Paperid:1614
Authors:Sohrab Madani; Jayden Guan; Waleed Ahmed; Saurabh Gupta; Haitham Hassanieh
Title: Radatron: Accurate Detection Using Multi-Resolution Cascaded MIMO Radar
Abstract:
Millimeter wave (mmWave) radars are becoming a more popular sensing modality in self-driving cars due to their favorable characteristics in adverse weather. Yet, they currently lack sufficient spatial resolution for semantic scene understanding. In this paper, we present Radatron, a system capable of accurate object detection using mmWave radar as a stand-alone sensor. To enable Radatron, we introduce a first-of-its-kind, high-resolution automotive radar dataset collected with a cascaded MIMO (Multiple Input Multiple Output) radar. Our radar achieves 5 cm range resolution and 1.2-degree angular resolution, 10× finer than other publicly available datasets. We also develop a novel hybrid radar processing and deep learning approach to achieve high vehicle detection accuracy. We train and extensively evaluate Radatron to show it achieves 92.6% AP50 and 56.3% AP75 accuracy in 2D bounding box detection, an 8% and 15.9% improvement over prior art respectively. Code and dataset are available on https://jguan.page/Radatron/.



Paperid:1615
Authors:Yi Wei; Zibu Wei; Yongming Rao; Jiaxin Li; Jie Zhou; Jiwen Lu
Title: LiDAR Distillation: Bridging the Beam-Induced Domain Gap for 3D Object Detection
Abstract:
In this paper, we propose the LiDAR Distillation to bridge the domain gap induced by different LiDAR beams for 3D object detection. In many real-world applications, the LiDAR points used by mass-produced robots and vehicles usually have fewer beams than that in large-scale public datasets. Moreover, as the LiDARs are upgraded to other product models with different beam amount, it becomes challenging to utilize the labeled data captured by previous versions’ high-resolution sensors. Despite the recent progress on domain adaptive 3D detection, most methods struggle to eliminate the beam-induced domain gap. We find that it is essential to align the point cloud density of the source domain with that of the target domain during the training process. Inspired by this discovery, we propose a progressive framework to mitigate the beam-induced domain shift. In each iteration, we first generate low-beam pseudo LiDAR by downsampling the high-beam point clouds. Then the teacher-student framework is employed to distill rich information from the data with more beams. Extensive experiments on Waymo, nuScenes and KITTI datasets with three different LiDAR-based detectors demonstrate the effectiveness of our LiDAR Distillation. Notably, our approach does not increase any additional computation cost for inference. Code is available at https://github.com/weiyithu/LiDAR-Distillation.



Paperid:1616
Authors:Maosheng Ye; Rui Wan; Shuangjie Xu; Tongyi Cao; Qifeng Chen
Title: Efficient Point Cloud Segmentation with Geometry-Aware Sparse Networks
Abstract:
In point cloud learning, sparsity and geometry are two core properties. Recently, many approaches have been proposed through single or multiple representations to improve the performance of point cloud semantic segmentation. However, these works fail to maintain the balance among performance, efficiency, and memory consumption, showing incapability to integrate sparsity and geometry appropriately. To address these issues, we propose the Geometry-aware Sparse Networks (GASN) by utilizing the sparsity and geometry of a point cloud in a single voxel representation. GASN mainly consists of two modules, namely Sparse Feature Encoder and Sparse Geometry Feature Enhancement. The Sparse Feature Encoder extracts the local context information, and the Sparse Geometry Feature Enhancement enhances the geometric properties of a sparse point cloud to improve both efficiency and performance. In addition, we propose deep sparse supervision in the training phase to help convergence and alleviate the memory consumption problem. Our GASN achieves state-of-the-art performance on both SemanticKITTI and Nuscenes datasets while running significantly faster and consuming less memory.



Paperid:1617
Authors:Lihe Ding; Shaocong Dong; Tingfa Xu; Xinli Xu; Jie Wang; Jianan Li
Title: FH-Net: A Fast Hierarchical Network for Scene Flow Estimation on Real-World Point Clouds
Abstract:
Estimating scene flow from real-world point clouds is a fundamental task for practical 3D vision. Previous methods often rely on deep models to first extract expensive per-point features at full resolution, and then get the flow either from complex matching mechanism or feature decoding, suffering high computational cost and latency. In this work, we propose a fast hierarchical network, FH-Net, which directly gets the key points flow through a lightweight Trans-flow layer utilizing the reliable local geometry prior, and optionally back-propagates the computed sparse flows through an inverse Trans-up layer to obtain hierarchical flows at different resolutions. To focus more on challenging dynamic objects, we also provide a new copy-and-paste data augmentation technique based on dynamic object pairs generation. Moreover, to alleviate the chronic shortage of real-world training data, we establish two new large-scale datasets to this field by collecting lidar-scanned point clouds from public autonomous driving datasets and annotating the collected data through novel pseudo-labeling. Extensive experiments on both public and proposed datasets show that our method outperforms prior state-of-the-arts while running at least 7× faster at 113 FPS. Code and data are released at https://github.com/pigtigger/FH-Net.



Paperid:1618
Authors:Simon Doll; Richard Schulz; Lukas Schneider; Viviane Benzin; Markus Enzweiler; Hendrik P.A. Lensch
Title: SpatialDETR: Robust Scalable Transformer-Based 3D Object Detection from Multi-View Camera Images with Global Cross-Sensor Attention
Abstract:
Based on the key idea of DETR this paper introduces an object-centric 3D object detection framework that operates on a limited number of 3D object queries instead of dense bounding box proposals followed by non-maximum suppression. After image feature extraction a decoder-only transformer architecture is trained on a set-based loss. SpatialDETR infers the classification and bounding box estimates based on attention both spatially within each image and across the different views. To fuse the multi-view information in the attention block we introduce a novel geometric positional encoding that incorporates the view ray geometry to explicitly consider the extrinsic and intrinsic camera setup. This way, the spatially-aware cross-view attention exploits arbitrary receptive fields to integrate cross-sensor data and therefore global context. Extensive experiments on the nuScenes benchmark demonstrate the potential of global attention and result in state-of-the-art performance. Code available at https://github.com/cgtuebingen/SpatialDETR.



Paperid:1619
Authors:Yu Tian; Yuyuan Liu; Guansong Pang; Fengbei Liu; Yuanhong Chen; Gustavo Carneiro
Title: Pixel-Wise Energy-Biased Abstention Learning for Anomaly Segmentation on Complex Urban Driving Scenes
Abstract:
State-of-the-art (SOTA) anomaly segmentation approaches on complex urban driving scenes explore pixel-wise classification uncertainty learned from outlier exposure, or external reconstruction models. However, previous uncertainty approaches that directly associate high uncertainty to anomaly may sometimes lead to incorrect anomaly predictions, and external reconstruction models tend to be too inefficient for real-time self-driving embedded systems. In this paper, we propose a new anomaly segmentation method, named pixel-wise energy-biased abstention learning (PEBAL), that explores pixel-wise abstention learning (AL) with a model that learns an adaptive pixel-level anomaly class, and an energy-based model (EBM) that learns inlier pixel distribution. More specifically, PEBAL is based on a non-trivial joint training of EBM and AL, where EBM is trained to output high-energy for anomaly pixels (from outlier exposure) and AL is trained such that these high-energy pixels receive adaptive low penalty for being included to the anomaly class. We extensively evaluate PEBAL against the SOTA and show that it achieves the best performance across four benchmarks.



Paperid:1620
Authors:Chris Zhang; Runsheng Guo; Wenyuan Zeng; Yuwen Xiong; Binbin Dai; Rui Hu; Mengye Ren; Raquel Urtasun
Title: Rethinking Closed-Loop Training for Autonomous Driving
Abstract:
Recent advances in high-fidelity simulators [22,82,44] have enabled closed-loop training of autonomous driving agents, potentially solving the distribution shift in training v.s. deployment and allowing training to be scaled both safely and cheaply. However, there is a lack of understanding of how to build effective training benchmarks for closed-loop training. In this work, we present the first empirical study which analyzes the effects of different training benchmark designs on the success of learning agents, such as how to design traffic scenarios and scale training environments. Furthermore, we show that many popular RL algorithms cannot achieve satisfactory performance in the context of autonomous driving, as they lack long-term planning and take an extremely long time to train. To address these issues, we propose trajectory value learning (TRAVL), an RL-based driving agent approach that performs planning with multistep look-ahead and exploits cheaply generated imagined data for efficient learning. Our experiments show that TRAVL can learn much faster and produce safer maneuvers compared to all the baselines.



Paperid:1621
Authors:Gwangtak Bae; Byungjun Kim; Seongyong Ahn; Jihong Min; Inwook Shim
Title: SLiDE: Self-Supervised LiDAR De-Snowing through Reconstruction Difficulty
Abstract:
LiDAR is widely used to capture accurate 3D outdoor scene structures. However, LiDAR produces many undesirable noise points in snowy weather, which hamper analyzing meaningful 3D scene structures. Semantic segmentation with snow labels would be a straightforward solution for removing them, but it requires laborious point-wise annotation. To address this problem, we propose a novel self-supervised learning framework for snow points removal in LiDAR point clouds. Our method exploits the structural characteristic of the noise points: low spatial correlation with their neighbors. Our method consists of two deep neural networks: Point Reconstruction Network (PR-Net) reconstructs each point from its neighbors; Reconstruction Difficulty Network (RD-Net) predicts point-wise difficulty of the reconstruction by PR-Net, which we call reconstruction difficulty. With simple post-processing, our method effectively detects snow points without any label. Our method achieves the state-of-the-art performance among label-free approaches and is comparable to the fully-supervised method. Moreover, we demonstrate that our method can be exploited as a pretext task to improve label-efficiency of supervised training of de-snowing.



Paperid:1622
Authors:Sixian Zhang; Weijie Li; Xinhang Song; Yubing Bai; Shuqiang Jiang
Title: Generative Meta-Adversarial Network for Unseen Object Navigation
Abstract:
Object navigation is a task to let the agent navigate to a target object. Prevailing works attempt to expand navigation ability in new environments and achieve reasonable performance on the seen object categories that have been observed in training environments. However, this setting is somewhat limited in real world scenario, where navigating to unseen object categories is generally unavoidable. In this paper, we focus on the problem of navigating to unseen objects in new environments only based on limited training knowledge. Same as the common ObjectNav tasks, our agent still gets the egocentric observation and target object category as the input and does not require any extra inputs. Our solution is to let the agent “imagine"" the unseen object by synthesizing features of the target object. We propose a generative meta-adversarial network (GMAN), which is mainly composed of a feature generator and an environmental meta discriminator, aiming to generate features for unseen objects and new environments in two steps. The former generates the initial features of the unseen objects based on the semantic embedding of the object category. The latter enables the generator to further learn the background characteristics of the new environment, progressively adapting the generated features to approximate the real features of the target object. The adapted features serve as a more specific representation of the target to guide the agent. Moreover, to fast update the generator with a few observations, the entire adversarial framework is learned in the gradient-based meta-learning manner. The experimental results on AI2THOR and RoboTHOR simulators demonstrate the effectiveness of the proposed method in navigating to unseen object categories.



Paperid:1623
Authors:Kiana Ehsani; Ali Farhadi; Aniruddha Kembhavi; Roozbeh Mottaghi
Title: Object Manipulation via Visual Target Localization
Abstract:
Object manipulation is a critical skill required for Embodied AI agents interacting with the world around them. Training agents to manipulate objects, poses many challenges. These include occlusion of the target object by the agent’s arm, noisy object detection and localization, and the target frequently going out of view as the agent moves around in the scene. We propose Manipulation via Visual Object Location Estimation (m-VOLE), an approach that explores the environment in search for target objects, computes their 3D coordinates once they are located, and then continues to estimate their 3D locations even when the objects are not visible, thus robustly aiding the task of manipulating these objects throughout the episode. Our evaluations show a massive 3x improvement in success rate over a model that has access to the same sensory suite but is trained without the object location estimator, and our analysis shows that our agent is robust to noise in depth perception and agent localization. Importantly, our proposed approach relaxes several assumptions about idealized localization and perception that are commonly employed by recent works in navigation and manipulation -- an important step towards training agents for object manipulation in the real world. Our code and data are available at https://prior.allenai.org/projects/m-vole}{prior.allenai.org/projects/m-vole.



Paperid:1624
Authors:Eun Sun Lee; Junho Kim; SangWon Park; Young Min Kim
Title: MoDA: Map Style Transfer for Self-Supervised Domain Adaptation of Embodied Agents
Abstract:
We propose a domain adaptation method, MoDA, which adapts a pretrained embodied agent to a new, noisy environment without ground-truth supervision. Map-based memory provides important contextual information for visual navigation, and exhibits unique spatial structure mainly composed of flat walls and rectangular obstacles. Our adaptation approach encourages the inherent regularities on the estimated maps to guide the agent to overcome the prevalent domain discrepancy in a novel environment. Specifically, we propose an efficient learning curriculum to handle the visual and dynamics corruptions in an online manner, self-supervised with pseudo clean maps generated by style transfer networks. Because the map-based representation provides spatial knowledge for the agent’s policy, our formulation can deploy the pretrained policy networks from simulators in a new setting. We evaluate MoDA in various practical scenarios and show that our proposed method quickly enhances the agent’s performance in downstream tasks including localization, mapping, exploration, and point-goal navigation.



Paperid:1625
Authors:Yash Kant; Arun Ramachandran; Sriram Yenamandra; Igor Gilitschenski; Dhruv Batra; Andrew Szot; Harsh Agrawal
Title: Housekeep: Tidying Virtual Households Using Commonsense Reasoning
Abstract:
We introduce Housekeep, a benchmark to evaluate commonsense reasoning in the home for embodied AI. In Housekeep, an embodied agent must tidy a house by rearranging misplaced objects without explicit instructions specifying which objects need to be rearranged. Instead, the agent must learn from and is evaluated against human preferences of which objects belong where in a tidy house. Specifically, we collect a dataset of where humans typically place objects in tidy and untidy houses constituting 1799 objects, 268 object categories, 585 placements, and 105 rooms. Next, we propose a modular baseline approach for Housekeep that integrates planning, exploration, and navigation. It leverages a fine-tuned large language model (LLM) trained on an internet text corpus for effective planning. We show that our baseline generalizes to rearranging unseen objects in unknown environments.



Paperid:1626
Authors:Qiyu Dai; Jiyao Zhang; Qiwei Li; Tianhao Wu; Hao Dong; Ziyuan Liu; Ping Tan; He Wang
Title: Domain Randomization-Enhanced Depth Simulation and Restoration for Perceiving and Grasping Specular and Transparent Objects
Abstract:
Commercial depth sensors usually generate noisy and missing depths, especially on specular and transparent objects, which poses critical issues to downstream depth or point cloud-based tasks. To mitigate this problem, we propose a powerful RGBD fusion network, SwinDRNet, for depth restoration. We further propose Domain Randomization-Enhanced Depth Simulation (DREDS) approach to simulate an active stereo depth system using physically based rendering and generate a large-scale synthetic dataset that contains 130K photorealistic RGB images along with their simulated depths carrying realistic sensor noises. To evaluate depth restoration methods, we also curate a real-world dataset, namely STD, that captures 30 cluttered scenes composed of 50 objects with different materials from specular, transparent, to diffuse. Experiments demonstrate that the proposed DREDS dataset bridges the sim-to-real domain gap such that, trained on DREDS, our SwinDRNet can seamlessly generalize to other real depth datasets, e.g. ClearGrasp, and outperform the competing methods on depth restoration. We further show that our depth restoration effectively boosts the performance of downstream tasks, including category-level pose estimation and grasping tasks. Our data and code are available at https://github.com/PKU-EPIC/DREDS.



Paperid:1627
Authors:Chia-Chi Chuang; Donglin Yang; Chuan Wen; Yang Gao
Title: Resolving Copycat Problems in Visual Imitation Learning via Residual Action Prediction
Abstract:
Imitation learning is a widely used policy learning method that enables intelligent agents to acquire complex skills from expert demonstrations. The input to the imitation learning algorithm is usually composed of both the current observation and historical observations since the most recent observation might not contain enough information. This is especially the case with image observations, where a single image only includes one view of the scene, and it suffers from a lack of motion information and object occlusions. In theory, providing multiple observations to the imitation learning agent will lead to better performance. However, surprisingly people find that sometimes imitation from observation histories performs worse than imitation from the most recent observation. In this paper, we explain this phenomenon from the information flow within the neural network perspective. We also propose a novel imitation learning neural network architecture that does not suffer from this issue by design. Furthermore, our method scales to high-dimensional image observations. Finally, we benchmark our approach on two widely used simulators, CARLA and MuJoCo, and it successfully alleviates the copycat problem and surpasses the existing solutions.



Paperid:1628
Authors:Hanxiao Jiang; Yongsen Mao; Manolis Savva; Angel X. Chang
Title: OPD: Single-View 3D Openable Part Detection
Abstract:
We address the task of predicting what parts of an object can open and how they move when they do so. The input is a single image of an object, and as output we detect what parts of the object can open, and the motion parameters describing the articulation of each openable part. To tackle this task, we create two datasets of 3D objects: OPDSynth based on existing synthetic objects, and OPDReal based on RGBD reconstructions of real objects. We then design OPDRCNN, a neural architecture that detects openable parts and predicts their motion parameters. Our experiments show that this is a challenging task especially when considering generalization across object categories, and the limited amount of information in a single image. Our architecture outperforms baselines and prior work especially for RGB image inputs.



Paperid:1629
Authors:Bowen Li; Chen Wang; Pranay Reddy; Seungchan Kim; Sebastian Scherer
Title: AirDet: Few-Shot Detection without Fine-Tuning for Autonomous Exploration
Abstract:
Few-shot object detection has attracted increasing attention and rapidly progressed in recent years. However, the requirement of an exhaustive offline fine-tuning stage in existing methods is time-consuming and significantly hinders their usage in online applications such as autonomous exploration of low-power robots. We find that their major limitation is that the little but valuable information from a few support images is not fully exploited. To solve this problem, we propose a brand new architecture, AirDet, and surprisingly find that, by learning class-agnostic relation with the support images in all modules, including cross-scale object proposal network, shots aggregation module, and localization network, AirDet without fine-tuning achieves comparable or even better results than many fine-tuned methods, reaching up to 30-40% improvements. We also present solid results of onboard tests on real-world exploration data from the DARPA Subterranean Challenge, which strongly validate the feasibility of AirDet in robotics. To the best of our knowledge, AirDet is the first feasible few-shot detection method for autonomous exploration of low-power robots. The code and pre-trained models are released at https://github.com/Jaraxxus-Me/AirDet.



Paperid:1630
Authors:Hongtao Wen; Jianhang Yan; Wanli Peng; Yi Sun
Title: TransGrasp: Grasp Pose Estimation of a Category of Objects by Transferring Grasps from Only One Labeled Instance
Abstract:
Grasp pose estimation is an important issue for robots to interact with the real world. However, most of existing methods require exact 3D object models available beforehand or a large amount of grasp annotations for training. To avoid these problems, we propose TransGrasp, a category-level grasp pose estimation method that predicts grasp poses of a category of objects by labeling only one object instance. Specifically, we perform grasp pose transfer across a category of objects based on their shape correspondences and propose a grasp pose refinement module to further fine-tune grasp pose of grippers so as to ensure successful grasps. Experiments demonstrate the effectiveness of our method on achieving high-quality grasps with the transferred grasp poses. Our code is available at https://github.com/yanjh97/TransGrasp.



Paperid:1631
Authors:Jinghuan Shang; Kumara Kahatapitiya; Xiang Li; Michael S. Ryoo
Title: StARformer: Transformer with State-Action-Reward Representations for Visual Reinforcement Learning
Abstract:
Reinforcement Learning (RL) can be considered as a sequence modeling task: given a sequence of past state-action-reward experiences, an agent predicts a sequence of next actions. In this work, we propose State-Action-Reward Transformer (StARformer) for visual RL, which explicitly models short-term state-action-reward representations (StAR-representations), essentially introducing a Markovian-like inductive bias to improve long-term modeling. Our approach first extracts StAR-representations by self-attending image state patches, action, and reward tokens within a short temporal window. These are then combined with pure image state representations --- extracted as convolutional features, to perform self-attention over the whole sequence. Our experiments show that StARformer outperforms the state-of-the-art Transformer-based method on image-based Atari and DeepMind Control Suite benchmarks, in both offline-RL and imitation learning settings. StARformer is also more compliant with longer sequences of inputs. Our code is available at https://github.com/elicassion/StARformer.



Paperid:1632
Authors:Gabriel Sarch; Zhaoyuan Fang; Adam W. Harley; Paul Schydlo; Michael J. Tarr; Saurabh Gupta; Katerina Fragkiadaki
Title: TIDEE: Tidying Up Novel Rooms Using Visuo-Semantic Commonsense Priors
Abstract:
We introduce TIDEE, an embodied agent that tidies up a disordered scene based on learned commonsense object placement and room arrangement priors. TIDEE explores a home environment, detects objects that are out of their natural place, infers plausible object contexts for them, localizes such contexts in the current scene, and repositions the objects. Commonsense priors are encoded in three modules: i) visuo-semantic detectors that detect out-of-place objects, ii) an associative neural graph memory of objects and spatial relations that proposes plausible semantic receptacles and surfaces for object repositions, and iii) a visual search network that guides the agent’s exploration for efficiently localizing the receptacle-of-interest in the current scene to reposition the object. We test TIDEE on tidying up disorganized scenes in the AI2THOR simulation environment. TIDEE carries out the task directly from pixel and raw depth input without ever having observed the same room beforehand, relying only on priors learned from a separate set of training houses. Human evaluations on the resulting room reorganizations show TIDEE outperforms ablative versions of the model that do not use one or more of the commonsense priors. On a related room rearrangement benchmark that allows the agent to view the goal state prior to rearrangement, a simplified version of our model significantly outperforms a top-performing method by a large margin. Code and data are available at the project website: https://tidee-agent.github.io/.



Paperid:1633
Authors:Chao Yu; Xinyi Yang; Jiaxuan Gao; Huazhong Yang; Yu Wang; Yi Wu
Title: Learning Efficient Multi-agent Cooperative Visual Exploration
Abstract:
We tackle the problem of cooperative visual exploration where multiple agents need to jointly explore unseen regions as fast as possible based on visual signals. Classical planning-based methods often suffer from expensive computation overhead at each step and a limited expressiveness of complex cooperation strategy. By contrast, reinforcement learning (RL) has recently become a popular paradigm for tackling this challenge due to its modeling capability of arbitrarily complex strategies and minimal inference overhead. In this paper, we propose a novel RL-based multi-agent planning module, Multi-agent Spatial Planner (MSP). MSP leverages a transformer-based architecture, Spatial-TeamFormer, which effectively captures spatial relations and intra-agent interactions via hierarchical spatial self-attentions. In addition, we also implement a few multi-agent enhancements to process local information from each agent for an aligned spatial representation and more precise planning. Finally, we perform policy distillation to extract a meta policy to significantly improve the generalization capability of final policy. We call this overall solution, Multi-Agent Active Neural SLAM (MAANS). MAANS substantially outperforms classical planning-based baselines for the first time in a photo-realistic 3D simulator, Habitat. Code and videos can be found at https://sites.google.com/view/maans.



Paperid:1634
Authors:Walter Goodwin; Sagar Vaze; Ioannis Havoutis; Ingmar Posner
Title: Zero-Shot Category-Level Object Pose Estimation
Abstract:
Object pose estimation is an important component of most vision pipelines for embodied agents, as well as in 3D vision more generally. In this paper we tackle the problem of estimating the pose of novel object categories in a zero-shot manner. This extends much of the existing literature by removing the need for pose-labelled datasets or category-specific CAD models for training or inference. Specifically, we make the following contributions. First, we formalise the zero-shot, category-level pose estimation problem and frame it in a way that is most applicable to real-world embodied agents. Secondly, we propose a novel method based on semantic correspondences from a self-supervised vision transformer to solve the pose estimation problem. We further re-purpose the recent CO3D dataset to present a controlled and realistic test setting. Finally, we demonstrate that all baselines for our proposed task perform poorly, and show that our method provides a six-fold improvement in average rotation accuracy at 30 degrees. Our code is available at https://github.com/applied-ai-lab/zero-shot-pose.



Paperid:1635
Authors:Kai Chen; Rui Cao; Stephen James; Yichuan Li; Yun-Hui Liu; Pieter Abbeel; Qi Dou
Title: Sim-to-Real 6D Object Pose Estimation via Iterative Self-Training for Robotic Bin Picking
Abstract:
6D object pose estimation is important for robotic bin-picking, and serves as a prerequisite for many downstream industrial applications. However, it is burdensome to annotate a customized dataset associated with each specific bin-picking scenario for training pose estimation models. In this paper, we propose an iterative self-training framework for sim-to-real 6D object pose estimation to facilitate cost-effective robotic grasping. Given a bin-picking scenario, we establish a photo-realistic simulator to synthesize abundant virtual data, and use this to train an initial pose estimation network. This network then takes the role of a teacher model, which generates pose predictions for unlabeled real data. With these predictions, we further design a comprehensive adaptive selection scheme to distinguish reliable results, and leverage them as pseudo labels to update a student model for pose estimation on real data. To continuously improve the quality of pseudo labels, we iterate the above steps by taking the trained student model as a new teacher and re-label real data using the refined teacher model. We evaluate our method on a public benchmark and our newly-released dataset, achieving an ADD(-S) improvement of 11.49% and 22.62% respectively. Our method is also able to improve robotic bin-picking success by 19.54%, demonstrating the potential of iterative sim-to-real solutions for robotic applications. Project homepage: www.cse.cuhk.edu.hk/ kaichen/sim2real_pose.html.



Paperid:1636
Authors:Sagnik Majumder; Kristen Grauman
Title: Active Audio-Visual Separation of Dynamic Sound Sources
Abstract:
We explore active audio-visual separation for dynamic sound sources, where an embodied agent moves intelligently in a 3D environment to continuously isolate the time-varying audio stream being emitted by an object of interest. The agent hears a mixed stream of multiple audio sources (e.g., multiple people conversing and a band playing music at a noisy party). Given a limited time budget, it needs to extract the target sound accurately at every step using egocentric audio-visual observations. We propose a reinforcement learning agent equipped with a novel transformer memory that learns motion policies to control its camera and microphone to recover the dynamic target audio, using self-attention to make high-quality estimates for current timesteps and also simultaneously improve its past estimates. Using highly realistic acoustic SoundSpaces simulations in real-world scanned Matterport3D environments, we show that our model is able to learn efficient behavior to carry out continuous separation of a dynamic audio target. Project: https://vision.cs.utexas.edu/projects/active-av-dynamic-separation/.



Paperid:1637
Authors:Yuzhe Qin; Yueh-Hua Wu; Shaowei Liu; Hanwen Jiang; Ruihan Yang; Yang Fu; Xiaolong Wang
Title: DexMV: Imitation Learning for Dexterous Manipulation from Human Videos
Abstract:
While in computer vision we have made significant progress on understanding hand-object interactions, it is still very challenging for robots to perform complex dexterous manipulation. In this paper, we propose a new platform and pipeline, DexMV (Dexterous Manipulation from Videos), for imitation learning to bridge the gap between computer vision and robot learning. We design a platform with: (i) a simulation system for complex dexterous manipulation tasks with a multi-finger robot hand and (ii) a computer vision system to record large-scale demonstrations of a human hand conducting the same tasks. In the DexMV pipeline, we couple 3D hand and object pose estimation on the videos with hand motion retargeting algorithm, to extract the hand-object state trajectories. We compare multiple imitation learning and reinforcement learning (RL) algorithms on the manipulation tasks in the simulation. We show that the demonstrations can indeed improve robot learning by a large margin and solve the complex tasks which RL alone cannot solve.



Paperid:1638
Authors:Jacob Krantz; Stefan Lee
Title: Sim-2-Sim Transfer for Vision-and-Language Navigation in Continuous Environments
Abstract:
Recent work in Vision-and-Language Navigation (VLN) has presented two environmental paradigms with differing realism -- the standard VLN setting built on topological environments where navigation is abstracted away, and the VLN-CE setting where agents must navigate continuous 3D environments using low-level actions. Despite sharing the high-level task and even the underlying instruction-path data, performance on VLN-CE lags behind VLN significantly. In this work, we explore this gap by transferring an agent from the abstract environment of VLN to the continuous environment of VLN-CE. We find that this sim-2-sim transfer is highly effective, improving over the prior state of the art in VLN-CE by +12% success rate. While this demonstrates the potential for this direction, the transfer does not fully retain the original performance of the agent in the abstract setting. We present a sequence of experiments to identify what differences result in performance degradation, providing clear directions for further improvement.



Paperid:1639
Authors:Juyong Lee; Seokjun Ahn; Jaesik Park
Title: Style-Agnostic Reinforcement Learning
Abstract:
We present a novel method of learning style-agnostic representation using both style transfer and adversarial learning in the reinforcement learning framework. The style, here, refers to task-irrelevant details such as the color of the background in the images, where generalizing the learned policy across environments with different styles is still a challenge. Focusing on learning style-agnostic representations, our method trains the actor with diverse image styles generated from an inherent adversarial style perturbation generator, which plays a min-max game between the actor and the generator, without demanding expert knowledge for data augmentation or additional class labels for adversarial training. We verify that our method achieves competitive or better performances than the state-of-the-art approaches on Procgen and Distracting Control Suite benchmarks, and further investigate the features extracted from our model, showing that the model better captures the invariants and is less distracted by the shifted style. The code is available at https://github.com/POSTECH-CVLab/style-agnostic-RL.



Paperid:1640
Authors:Houjian Yu; Changhyun Choi
Title: Self-Supervised Interactive Object Segmentation through a Singulation-and-Grasping Approach
Abstract:
Instance segmentation with unseen objects is a challenging problem in unstructured environments. To solve this problem, we propose a robot learning approach to actively interact with novel objects and collect each object’s training label for further fine-tuning to improve the segmentation model performance, while avoiding the time-consuming process of manually labeling a dataset. Given a cluttered pile of objects, our approach chooses pushing and grasping motions to break the clutter and conducts object-agnostic grasping for which the Singulation-and-Grasping (SaG) policy takes as input the visual observations and imperfect segmentation. We decompose the problem into three subtasks: (1) the object singulation subtask aims to separate the objects from each other, which creates more space that alleviates the difficulty of (2) the collision-free grasping subtask; (3) the mask generation subtask obtains the self-labeled ground truth masks by using an optical flow-based binary classifier and motion cue post-processing for transfer learning. Our system achieves 70% singulation success rate in simulated cluttered scenes. The interactive segmentation of our system achieves 87.8%, 73.9%, and 69.3% average precision for toy blocks, YCB objects in simulation, and real-world novel objects, respectively, which outperforms the compared baselines.



Paperid:1641
Authors:Shizhe Chen; Pierre-Louis Guhur; Makarand Tapaswi; Cordelia Schmid; Ivan Laptev
Title: Learning from Unlabeled 3D Environments for Vision-and-Language Navigation
Abstract:
In vision-and-language navigation (VLN), an embodied agent is required to navigate in realistic 3D environments following natural language instructions. One major bottleneck for existing VLN approaches is the lack of sufficient training data, resulting in unsatisfactory generalization to unseen environments. While VLN data is typically collected manually, such an approach is expensive and prevents scalability. In this work, we address the data scarcity issue by proposing to automatically create a large-scale VLN dataset from 900 unlabeled 3D buildings from HM3D. We generate a navigation graph for each building and transfer object predictions from 2D to generate pseudo 3D object labels by cross-view consistency. We then fine-tune a pretrained language model using pseudo object labels as prompts to alleviate the cross-modal gap in instruction generation. Our resulting HM3D-AutoVLN dataset is an order of magnitude larger than existing VLN datasets in terms of navigation environments and instructions. We experimentally demonstrate that HM3D-AutoVLN significantly increases the generalization ability of resulting VLN models. On the SPL metric, our approach improves over state of the art by 7.1% and 8.1% on the unseen validation splits of REVERIE and SOON datasets respectively.



Paperid:1642
Authors:Dorian F. Henning; Tristan Laidlow; Stefan Leutenegger
Title: "BodySLAM: Joint Camera Localisation, Mapping, and Human Motion Tracking"
Abstract:
Estimating human motion from video is an active research area due to its many potential applications. Most state-of-the-art methods predict human shape and posture estimates for individual images and do not leverage the temporal information available in video. Many ""in the wild"" sequences of human motion are captured by a moving camera, which adds the complication of conflated camera and human motion to the estimation. We therefore present BodySLAM, a monocular SLAM system that jointly estimates the position, shape, and posture of human bodies, as well as the camera trajectory. We also introduce a novel human motion model to constrain sequential body postures and observe the scale of the scene. Through a series of experiments on video sequences of human motion captured by a moving monocular camera, we demonstrate that BodySLAM improves estimates of all human body parameters and camera poses compared to estimating these separately.



Paperid:1643
Authors:Fabian Duffhauss; Ngo Anh Vien; Hanna Ziesche; Gerhard Neumann
Title: FusionVAE: A Deep Hierarchical Variational Autoencoder for RGB Image Fusion
Abstract:
Sensor fusion can significantly improve the performance of many computer vision tasks. However, traditional fusion approaches are either not data-driven and cannot exploit prior knowledge nor find regularities in a given dataset or they are restricted to a single application. We overcome this shortcoming by presenting a novel deep hierarchical variational autoencoder called FusionVAE that can serve as a basis for many fusion tasks. Our approach is able to generate diverse image samples that are conditioned on multiple noisy, occluded, or only partially visible input images. We derive and optimize a variational lower bound for the conditional log-likelihood of FusionVAE. In order to assess the fusion capabilities of our model thoroughly, we created three novel datasets for image fusion based on popular computer vision datasets. In our experiments, we show that FusionVAE learns a representation of aggregated information that is relevant to fusion tasks. The results demonstrate that our approach outperforms traditional methods significantly. Furthermore, we present the advantages and disadvantages of different design choices.



Paperid:1644
Authors:Chi Zhang; Sirui Xie; Baoxiong Jia; Ying Nian Wu; Song-Chun Zhu; Yixin Zhu
Title: Learning Algebraic Representation for Systematic Generalization in Abstract Reasoning
Abstract:
Is intelligence realized by connectionist or classicist? While connectionist approaches have achieved superhuman performance, there has been growing evidence that such task-specific superiority is particularly fragile in systematic generalization. This observation lies in the central debate between connectionist and classicist, wherein the latter continually advocates an algebraic treatment in cognitive architectures. In this work, we follow the classicist’s call and propose a hybrid approach to improve systematic generalization in reasoning. Specifically, we showcase a prototype with algebraic representation for the abstract spatial-temporal reasoning task of Raven’s Progressive Matrices (RPM) and present the ALgebra-Aware Neuro-Semi-Symbolic (ALANS) learner. The ALANS learner is motivated by abstract algebra and the representation theory. It consists of a neural visual perception frontend and an algebraic abstract reasoning backend: the frontend summarizes the visual information from object-based representation, while the backend transforms it into an algebraic structure and induces the hidden operator on the fly. The induced operator is later executed to predict the answer’s representation, and the choice most similar to the prediction is selected as the solution. Extensive experiments show that by incorporating an algebraic treatment, the ALANS learner outperforms various pure connectionist models in domains requiring systematic generalization. We further show the generative nature of the learned algebraic representation; it can be decoded by isomorphism to generate an answer.



Paperid:1645
Authors:Hoang-Anh Pham; Thao Minh Le; Vuong Le; Tu Minh Phuong; Truyen Tran
Title: Video Dialog As Conversation about Objects Living in Space-Time
Abstract:
It would be a technological feat to be able to create a system that can hold a meaningful conversation with humans about what they watch. A setup toward that goal is presented as a video dialog task, where the system is asked to generate natural utterances in response to a question in an ongoing dialog. The task poses great visual, linguistic, and reasoning challenges that cannot be easily overcome without an appropriate representation scheme over video and dialog that supports high-level reasoning. To tackle these challenges we present a new object-centric framework for video dialog that supports neural reasoning dubbed COST - which stands for Conversation about Objects in Space-Time. Here dynamic space-time visual content in videos is first parsed into object trajectories. Given this video abstraction, COST maintains and tracks object-associated dialog states, which are updated upon receiving new questions. Object interactions are dynamically and conditionally inferred for each question, and these serve as the basis for relational reasoning among them. COST also maintains a history of previous answers, and this allows retrieval of relevant object-centric information to enrich the answer forming process. Language production then proceeds in a step-wise manner, taking into the context of the current utterance, the existing dialog, the current question. We evaluate COST on the DSTC7 and DSTC8 benchmarks, demonstrating its competitiveness against state-of-the-arts."