ECCV 2020 论文列表

We learn a self-supervised, single-view 3D reconstruction model that predicts the 3D mesh shape, texture and camera pose of a target object with a collection of 2D images and silhouettes. The proposed method does not necessitate 3D supervision, manually annotated keypoints, multi-view images of an object or a prior 3D template. The key insight of our work is that objects can be represented as a collection of deformable parts, and each part is semantically coherent across different instances of the same category (e.g., wings on birds and wheels on cars). Therefore, by leveraging self-supervisedly learned part segmentation of a large collection of category-specific images, we can effectively enforce semantic consistency between the reconstructed meshes and the original images. This significantly reduces ambiguities during joint prediction of shape and camera pose of an object, along with texture. To the best of our knowledge, we are the first to try and solve the single-view reconstruction problem without a category-specific template mesh or semantic keypoints. Thus our model can easily generalize to various object categories without such labels, e.g., horses, penguins, etc. Through a variety of experiments on several categories of deformable and rigid objects, we demonstrate that our unsupervised method performs comparably if not better than existing category-specific reconstruction methods learned with supervision."

Outdoor scene relighting is a challenging problem that requires good understanding of the scene geometry, illumination and albedo. Current techniques are completely supervised, requiring high quality synthetic renderings to train a solution. Such renderings are synthesized using priors learned from limited data. In contrast, we propose a self-supervised approach for relighting. Our approach is trained only on corpora of images collected from the internet without any user-supervision. This virtually endless source of training data allows training a general relighting solution. Our approach first decomposes an image into its albedo, geometry and illumination. A novel relighting is then produced by modifying the illumination parameters. Our solution capture shadow using a dedicated shadow prediction map, and does not rely on accurate geometry estimation. We evaluate our technique subjectively and objectively using a new dataset with ground-truth relighting. Results show the ability of our technique to produce photo-realistic and physically plausible results, that generalizes to unseen scenes. "

Motion representation is key to many computer vision problems but has never been well studied in the literature. Existing works usually rely on the optical flow estimation to assist other tasks such as action recognition, frame prediction, video segmentation, etc. In this paper, we leverage the massive unlabeled video data to learn an accurate explicit motion representation that aligns well with the semantic distribution of the moving objects. Our method subsumes a coarse-to-fine paradigm, which first decodes the low-resolution motion maps from the rich spatial-temporal features of the video, then adaptively upsamples the low-resolution maps to the full-resolution by considering the semantic cues. To achieve this, we propose a novel context guided motion upsampling layer that leverages the spatial context of video objects to learn the upsampling parameters in an efficient way. We prove the effectiveness of our proposed motion representation method on downstream video understanding tasks, e.g., action recognition task. Experimental results show that our method performs favorably against state-of-the-art methods."

Video annotation is expensive and time consuming. Consequently, datasets for multi-person pose estimation and tracking are less diverse and have more sparse annotations compared to large scale image datasets for human pose estimation. This makes it challenging to learn deep learning based models for associating keypoints across frames that are robust to nuisance factors such as motion blur and occlusions for the task of multi-person pose tracking. %It is clear that leveraging large scale human pose annotations from single images benefits frame level human pose estimation in videos. This raises the question can large scale human pose annotations from single images benefit muti-person pose tracking. To address this issue, we propose an approach that relies on keypoint correspondences for associating persons in videos. Instead of training the network for estimating keypoint correspondences on video data, it is trained on a large scale image datasets for human pose estimation using self-supervision. Combined with a top-down framework for human pose estimation, we use keypoints correspondences to (i) recover missed pose detections (ii) associate pose detections across video frames. Our approach achieves state-of-the-art results for temporal pose estimation and multi-person pose tracking on the PosTrack $2017$ and PoseTrack $2018$ data sets. "

In recent years, deep learning emerges as one promising technique for solving many ill-posed inverse problems in image recovery, and most deep-learning-based solutions are based on supervised learning. Motivated by the practical value of reducing the cost and complexity of constructing labeled training datasets, this paper proposed a self-supervised deep learning approach for image recovery, which is dataset-free. Built upon Bayesian deep network, the proposed method trains a network with random weights that predicts the target image for recovery with uncertainty. Such uncertainty enables the prediction of the target image with small mean squared error by averaging multiple predictions. The proposed method is applied for image reconstruction in compressive sensing (CS), i.e., reconstructing an image from few measurements. The experiments showed that the proposed dataset-free deep learning method not only significantly outperforms traditional non-learning methods, but also is very competitive to the state-of-the-art supervised deep learning methods, especially when the measurements are few and noisy."

Delineation of cancerous regions in gigapixel whole slide images (WSIs) is a crucial diagnostic procedure in digital pathology. This process is time-consuming because of the large search space in the gigapixel WSIs, causing chances of omission and misinterpretation at indistinct tumor lesions. To tackle this, the development of an automated cancerous region segmentation method is imperative. We frame this issue as a modeling problem with partial label WSIs, where some cancerous regions may be misclassified as benign and vice versa, producing patches with noisy labels. To learn from these patches, we propose Self-similarity Student, combining teacher-student model paradigm with similarity learning. Specifically, for each patch, we first sample its similar and dissimilar patches according to spatial distance. A teacher-student model is then introduced, featuring the exponential moving average on both student model weights and teacher predictions ensemble. While our student model takes patches, teacher model takes all their corresponding similar and dissimilar patches for learning robust representation against noisy label patches. Following this similarity learning, our similarity ensemble merges similar patches’ ensembled predictions as the pseudo-label of a given patch to counteract its noisy label. On the CAMELYON16 dataset, our method substantially outperforms state-of-the-art noise-aware learning methods by 5% and the supervised-trained baseline by 10% in various degrees of noise. Moreover, our method is superior to the baseline on our TVGH TURP dataset with 2% improvement, demonstrating the generalizability to more clinical histopathology segmentation tasks."

Estimating the confidence of disparity maps inferred by a stereo algorithm has become a very relevant task in the years, due to the increasing number of applications leveraging such cue. Although self-supervised learning has recently spread across many computer vision tasks, it has been barely considered in the field of confidence estimation. In this paper, we propose a flexible and lightweight solution enabling self-adapting confidence estimation agnostic to the stereo algorithm or network. Our approach relies on the minimum information available in any stereo setup (\ie, the input stereo pair and the output disparity map) to learn an effective confidence measure. This strategy allows us not only a seamless integration with any stereo system, including consumer and industrial devices equipped with undisclosed stereo perception methods, but also, due to its self-adapting capability, for its out-of-the-box deployment in the field. Exhaustive experimental results with different standard datasets support our claims, showing how our solution is the first-ever enabling online learning of accurate confidence estimation for any stereo system and without any requirement for the end-user."

We address the problem of unsupervised procedure learning from instructional videos of multiple tasks using Deep Neural Networks (DNNs). Unlike existing works, we assume that training videos come from multiple tasks without key-step annotations or grammars, and the goals are to classify a test video to the underlying task and to localize its key-steps. Our DNN learns task-dependent attention features from informative regions of each frame without ground-truth bounding boxes and learns to discover and localize key-steps without key-step annotations by using an unsupervised subset selection module as a teacher. It also learns to classify an input video using the discovered key-steps using a learnable key-step feature pooling mechanism that extracts and learns to combine key-step based features for task recognition. By experiments on two instructional video datasets, we show the effectiveness of our method for unsupervised localization of procedure steps and video classification."

Self-supervised monocular depth estimation presents a powerful method to obtain 3D scene information from single camera images, which is trainable on arbitrary image sequences without requiring depth labels, e.g., from a LiDAR sensor. In this work we present a new self-supervised semantically-guided depth estimation (SGDepth) method to deal with moving dynamic-class (DC) objects, such as moving cars and pedestrians, which violate the static-world assumptions typically made during training of such models. Specifically, we propose (i) mutually beneficial cross-domain training of (supervised) semantic segmentation and self-supervised depth estimation with task-specific network heads, (ii) a semantic masking scheme providing guidance to prevent moving DC objects from contaminating the photometric loss, and (iii) a detection method for frames with non-moving DC objects, from which the depth of DC objects can be learned. We demonstrate the performance of our method on several benchmarks, in particular on the Eigen split, where we exceed all baselines without test-time refinement."

Recent learning-based approaches, in which models are trained by single-view images have shown promising results for monocular 3D face reconstruction, but they suffer from the ill-posed face pose and depth ambiguity issue. In contrast to previous works that only enforce 2D feature constraints, we propose a self-supervised training architecture by leveraging the multi-view geometry consistency, which provides reliable constraints on face pose and depth estimation. We first propose an occlusion-aware view synthesis method to apply multi-view geometry consistency to self-supervised learning. Then we design three novel loss functions for multi-view consistency, including the pixel consistency loss, the depth consistency loss, and the facial landmark-based epipolar loss. Our method is accurate and robust, especially under large variations of expressions, poses, and illumination conditions. Comprehensive experiments on the face alignment and 3D face reconstruction benchmarks have demonstrated superiority over state-of-the-art methods."

Our objective is to transform a video into a set of discrete audio-visual objects using self-supervised learning. To this end we introduce a model that uses attention to localize and group sound sources, and optical flow to aggregate information over time. We demonstrate the effectiveness of the audio-visual object embeddings that our model learns by using them for four downstream speech-oriented tasks: (a) multi-speaker sound source separation, (b) localizing and tracking speakers, (c) correcting misaligned audio-visual data, and (d) active speaker detection. Using our representation, these tasks can be solved entirely by training on unlabeled video, without the aid of object detectors. We also demonstrate the generality of our method by applying it to non-human speakers, including cartoons and puppets. Our model significantly outperforms other self-supervised approaches, and obtains performance competitive with methods that use supervised face detection."

Recent generative adversarial network (GAN) based methods (e.g., CycleGAN) are prone to fail at preserving image-objects in image-to-image translation, which reduces their practicality on tasks such as domain adaptation. Some frameworks have been proposed to adopt a segmentation network as the auxiliary regularization to prevent the content distortion. However, all of them require extra pixel-wise annotations, which is difficult to fulfill in practical applications. In this paper, we propose a novel GAN (namely OP-GAN) to address the problem, which involves a self-supervised module to enforce the image content consistency during image-to-image translations without any extra annotations. We evaluate the proposed OP-GAN on three publicly available datasets. The experimental results demonstrate that our OP-GAN can yield visually plausible translated images and significantly improve the semantic segmentation accuracy in different domain adaptation scenarios with off-the-shelf deep learning networks such as PSPNet and U-Net."

We develop a novel learning scheme named Self-Prediction for 3D instance and semantic segmentation of point clouds. Distinct from most existing methods that focus on designing convolutional operators, our method designs a new learning scheme to enhance point relation exploring for better segmentation. More specifically, we divide a point cloud sample into two subsets and construct a complete graph based on their representations. Then we use label propagation algorithm to predict labels of one subset when given labels of the other subset. By training with this Self-Prediction task, the backbone network is constrained to fully explore relational context/geometric/shape information and learn more discriminative features for segmentation. Moreover, a general associated framework equipped with our Self-Prediction scheme is designed for enhancing instance and semantic segmentation simultaneously, where instance and semantic representations are combined to perform Self-Prediction. Through this way, instance and semantic segmentation are collaborated and mutually reinforced. Significant performance improvements on instance and semantic segmentation compared with baseline are achieved on S3DIS and ShapeNet. Our method achieves state-of-the-art instance segmentation results on S3DIS and comparable semantic segmentation results compared with state-of-the-arts on S3DIS and ShapeNet when we only take PointNet++ as the backbone network."

Deep discriminative models (e.g.deep regression forests, deep neural decision forests) have achieved remarkable success recently to solve problems such as facial age estimation and head pose estimation. Most existing methods pursue robust and unbiased solutions either through learning discriminative features, or reweighting samples. We argue what is more desirable is learning gradually to discriminate like our human beings, and hence we resort to self-paced learning (SPL). Then, a natural question arises: can self-paced regime lead deep discriminative models to achieve more robust and less biased solutions? To this end, this paper proposes a new deep discriminative model—self-paceddeep regression forests with consideration on underrepresented examples (SPUDRFs). It tackles the fundamental ranking and selecting problem in SPL from a new perspective: fairness. This paradigm is fundamental and could be easily combined with a variety of deep discriminative models (DDMs). Extensive experiments on two computer vision tasks, i.e., facial age estimation and head pose estimation, demonstrate the efficacy of SPUDRFs, where state-of-the-art performances are achieved."

Convolutional Neural Networks (CNN) conduct image classification by activating dominant features that correlated with labels. When the training and testing data are under similar distributions, their dominant features are similar, leading to decent test performance. The performance is nonetheless unmet when tested with different distributions, leading to the challenges in cross-domain image classification. We introduce a simple training heuristic, Representation Self-Challenging (RSC), that significantly improves the generalization of CNN to the out-of-domain data. RSC iteratively challenges (discards) the dominant features activated on the training data, and forces the network to activate remaining features that correlate with labels. This process appears to activate feature representations applicable to out-of-domain data without prior knowledge of the new domain and without learning extra network parameters. We present the theoretical properties and conditions of RSC for improving cross-domain generalization. The experiments endorse the simple, effective, and architecture-agnostic nature of our RSC method."

Popular approaches for few-shot classification consist of first learning a generic data representation based on a large annotated dataset, before adapting the representation to new classes given only a few labeled samples. In this work, we propose a new strategy based on feature selection, which is both simpler and more effective than previous feature adaptation approaches. First, we obtain a multi-domain representation by training a set of semantically different feature extractors. Then, given a few-shot learning task, we use our multi-domain feature bank to automatically select the most relevant representations. We show that a simple non-parametric classifier built on top of such features produces high accuracy and generalizes to domains never seen during training, which leads to state-of-the-art results on MetaDataset and improved accuracy on mini-ImageNet."

Transparent objects such as windows and bottles made by glass widely exist in the real world. Segmenting transparent objects is challenging because these objects have diverse appearances inherited from the image background, making them had similar appearance with their surroundings. Besides the technical diculty of this task, only a few previous datasets were specially designed and collected to explore this task and most of the existing datasets have major drawbacks. They either possess limited sample size such as merely a thousand of images without manual annotations, or they generate all images by using computer graphics method (i.e. not real image). To address this important problem, this work proposes a large-scale dataset for transparent object segmentation, named Trans10K, consisting of 10,428 images of real scenarios with carefully manual annotations, which are 10 times larger than the existing datasets. The transparent objects in Trans10K are extremely challenging due to high diversity in scale, viewpoint and occlusion as shown in Fig. 1. To evaluate theeffectiveness of Trans10K, we propose a novel boundary-aware segmentation method, termed TransLab, which exploits boundary as the clue to improve segmentation of transparent objects. Extensive experiments and ablation studies demonstrate the effectiveness of Trans10K and validate the practicality of learning object boundary in TransLab. For example, TransLab significantly outperforms 20 recent object segmentation methods based on deep learning, showing that this task is largely unsolved. We believe that both Trans10K and TransLab have important contributions to both the academia and industry, facilitating future researches and applications."

Today's success of state of the art methods for semantic segmentation is driven by large datasets. Data is considered an important asset that needs to be protected, as the collection and annotation of such datasets comes at significant efforts and associated costs. In addition, visual data might contain private or sensitive information, that makes it equally unsuited for public release. Unfortunately, recent work on membership inference in the broader area of adversarial machine learning and inference attacks on machine learning models has shown that even black box classifiers leak information on the dataset that they were trained on. We present the first attacks and defenses for complex, state of the art models for semantic segmentation. In order to mitigate the associated risks, we also study a series of defenses against such membership inference attacks and find effective counter measures against the existing risks. Finally, we extensively evaluate our attacks and defenses on a range of relevant real-world datasets: Cityscapes, BDD100K, and Mapillary Vistas."

Current multi-object tracking and segmentation (MOTS) methods follow the tracking-by-detection paradigm and adopt convolutions for feature extraction. However, as affected by the inherent receptive field, convolution based feature extraction inevitably mixes up the foreground features and the background features, resulting in ambiguities in the subsequent instance association. In this paper, we propose a highly effective method for learning instance embeddings based on segments by converting the compact image representation to un-ordered 2D point cloud representation. Our method generates a new tracking-by-points paradigm where discriminative instance embeddings are learned from randomly selected points rather than images. Furthermore, multiple informative data modalities are converted into point-wise representations to enrich point-wise features. The resulting online MOTS framework, named PointTrack, surpasses all the state-of-the-art methods including 3D tracking methods by large margins (5.4\% higher MOTSA and 18 times faster over MOTSFusion) with the near real-time speed (22 FPS). Evaluations across three datasets demonstrate both the effectiveness and efficiency of our method. Moreover, based on the observation that current MOTS datasets lack crowded scenes, we build a more challenging MOTS dataset named APOLLO MOTS with higher instance density. Both APOLLO MOTS and our codes are publicly available at https://github.com/detectRecog/PointTrack."