ICCV 2019 论文列表

Unlike vision modalities, body-worn sensors or passive sensing can avoid the failure of action understanding in vision related challenges, e.g. occlusion and appearance variation. However, a standard large-scale dataset does not exist, in which different types of modalities across vision and sensors are integrated. To address the disadvantage of vision-based modalities and push towards multi/cross modal action understanding, this paper introduces a new large-scale dataset recorded from 20 distinct subjects with seven different types of modalities: RGB videos, keypoints, acceleration, gyroscope, orientation, Wi-Fi and pressure signal. The dataset consists of more than 36k video clips for 37 action classes covering a wide range of daily life activities such as desktop-related and check-in-based ones in four different distinct scenarios. On the basis of our dataset, we propose a novel multi modality distillation model with attention mechanism to realize an adaptive knowledge transfer from sensor-based modalities to vision-based modalities. The proposed model significantly improves performance of action recognition compared to models trained with only RGB information. The experimental results confirm the effectiveness of our model on cross-subject, -view, -scene and -session evaluation criteria. We believe that this new large-scale multimodal dataset will contribute the community of multimodal based action understanding.

Metric learning seeks to embed images of objects such that class-defined relations are captured by the embedding space. However, variability in images is not just due to different depicted object classes, but also depends on other latent characteristics such as viewpoint or illumination. In addition to these structured properties, random noise further obstructs the visual relations of interest. The common approach to metric learning is to enforce a representation that is invariant under all factors but the ones of interest. In contrast, we propose to explicitly learn the latent characteristics that are shared by and go across object classes. We can then directly explain away structured visual variability, rather than assuming it to be unknown random noise. We propose a novel surrogate task to learn visual characteristics shared across classes with a separate encoder. This encoder is trained jointly with the encoder for class information by reducing their mutual information. On five standard image retrieval benchmarks the approach significantly improves upon the state-of-the-art. Code is available at https://github.com/Confusezius/metric-learning-mining-interclass-characteristics.

Understanding the world in 3D is a critical component of urban autonomous driving. Generally, the combination of expensive LiDAR sensors and stereo RGB imaging has been paramount for successful 3D object detection algorithms, whereas monocular image-only methods experience drastically reduced performance. We propose to reduce the gap by reformulating the monocular 3D detection problem as a standalone 3D region proposal network. We leverage the geometric relationship of 2D and 3D perspectives, allowing 3D boxes to utilize well-known and powerful convolutional features generated in the image-space. To help address the strenuous 3D parameter estimations, we further design depth-aware convolutional layers which enable location specific feature development and in consequence improved 3D scene understanding. Compared to prior work in monocular 3D detection, our method consists of only the proposed 3D region proposal network rather than relying on external networks, data, or multiple stages. M3D-RPN is able to significantly improve the performance of both monocular 3D Object Detection and Bird's Eye View tasks within the KITTI urban autonomous driving dataset, while efficiently using a shared multi-class model.

Facial images in surveillance or mobile scenarios often have large view-point variations in terms of pitch and yaw angles. These jointly occurred angle variations make face recognition challenging. Current public face databases mainly consider the case of yaw variations. In this paper, a new large-scale Multi-yaw Multi-pitch high-quality database is proposed for Facial Pose Analysis (M2FPA), including face frontalization, face rotation, facial pose estimation and pose-invariant face recognition. It contains 397,544 images of 229 subjects with yaw, pitch, attribute, illumination and accessory. M2FPA is the most comprehensive multi-view face database for facial pose analysis. Further, we provide an effective benchmark for face frontalization and pose-invariant face recognition on M2FPA with several state-of-the-art methods, including DR-GAN, TP-GAN and CAPG-GAN. We believe that the new database and benchmark can significantly push forward the advance of facial pose analysis in real-world applications. Moreover, a simple yet effective parsing guided discriminator is introduced to capture the local consistency during GAN optimization. Extensive quantitative and qualitative results on M2FPA and Multi-PIE demonstrate the superiority of our face frontalization method. Baseline results for both face synthesis and face recognition from state-of-the-art methods demonstrate the challenge offered by this new database.

Inferring relational behavior between road users as well as road users and their surrounding physical space is an important step toward effective modeling and prediction of navigation strategies adopted by participants in road scenes. To this end, we propose a relation-aware framework for future trajectory forecast. Our system aims to infer relational information from the interactions of road users with each other and with the environment. The first module involves visual encoding of spatio-temporal features, which captures human-human and human-space interactions over time. The following module explicitly constructs pair-wise relations from spatio-temporal interactions and identifies more descriptive relations that highly influence future motion of the target road user by considering its past trajectory. The resulting relational features are used to forecast future locations of the target, in the form of heatmaps with an additional guidance of spatial dependencies and consideration of the uncertainty. Extensive evaluations on the public benchmark datasets demonstrate the robustness and efficacy of the proposed framework as observed by performances higher than the state-of-the-art methods.

Image inpainting has been substantially improved with deep learning in the past years. Deep inpainting can fill image regions with plausible contents, which are not visually apparent. Although inpainting is originally designed to repair images, it can even be used for malicious manipulations, e.g., removal of specific objects. Therefore, it is necessary to identify the presence of inpainting in an image. This paper presents a method to locate the regions manipulated by deep inpainting. The proposed method employs a fully convolutional network that is based on high-pass filtered image residuals. Firstly, we analyze and observe that the inpainted regions are more distinguishable from the untouched ones in the residual domain. Hence, a high-pass pre-filtering module is designed to get image residuals for enhancing inpainting traces. Then, a feature extraction module, which learns discriminative features from image residuals, is built with four concatenated ResNet blocks. The learned feature maps are finally enlarged by an up-sampling module, so that a pixel-wise inpainting localization map is obtained. The whole network is trained end-to-end with a loss addressing the class imbalance. Extensive experimental results evaluated on both synthetic and realistic images subjected to deep inpainting have shown the effectiveness of the proposed method.

We propose to leverage the local information in a image sequence to support global camera relocalization. In contrast to previous methods that regress global poses from single images, we exploit the spatial-temporal consistency in sequential images to alleviate uncertainty due to visual ambiguities by incorporating a visual odometry (VO) component. Specifically, we introduce two effective steps called content-augmented pose estimation and motion-based refinement. The content-augmentation step focuses on alleviating the uncertainty of pose estimation by augmenting the observation based on the co-visibility in local maps built by the VO stream. Besides, the motion-based refinement is formulated as a pose graph, where the camera poses are further optimized by adopting relative poses provided by the VO component as additional motion constraints. Thus, the global consistency can be guaranteed. Experiments on the public indoor 7-Scenes and outdoor Oxford RobotCar benchmark datasets demonstrate that benefited from local information inherent in the sequence, our approach outperforms state-of-the-art methods, especially in some challenging cases, e.g., insufficient texture, highly repetitive textures, similar appearances, and over-exposure.

The convolution layer has been the dominant feature extractor in computer vision for years. However, the spatial aggregation in convolution is basically a pattern matching process that applies fixed filters which are inefficient at modeling visual elements with varying spatial distributions. This paper presents a new image feature extractor, called the local relation layer, that adaptively determines aggregation weights based on the compositional relationship of local pixel pairs. With this relational approach, it can composite visual elements into higher-level entities in a more efficient manner that benefits semantic inference. A network built with local relation layers, called the Local Relation Network (LR-Net), is found to provide greater modeling capacity than its counterpart built with regular convolution on large-scale recognition tasks such as ImageNet classification.

Unsupervised approaches to learning in neural networks are of substantial interest for furthering artificial intelligence, both because they would enable the training of networks without the need for large numbers of expensive annotations, and because they would be better models of the kind of general-purpose learning deployed by humans. However, unsupervised networks have long lagged behind the performance of their supervised counterparts, especially in the domain of large-scale visual recognition. Recent developments in training deep convolutional embeddings to maximize non-parametric instance separation and clustering objectives have shown promise in closing this gap. Here, we describe a method that trains an embedding function to maximize a metric of local aggregation, causing similar data instances to move together in the embedding space, while allowing dissimilar instances to separate. This aggregation metric is dynamic, allowing soft clusters of different scales to emerge. We evaluate our procedure on several large-scale visual recognition datasets, achieving state-of-the-art unsupervised transfer learning performance on object recognition in ImageNet, scene recognition in Places 205, and object detection in PASCAL VOC.

We propose a method for face de-identification that enables fully automatic video modification at high frame rates. The goal is to maximally decorrelate the identity, while having the perception (pose, illumination and expression) fixed. We achieve this by a novel feed-forward encoder-decoder network architecture that is conditioned on the high-level representation of a person's facial image. The network is global, in the sense that it does not need to be retrained for a given video or for a given identity, and it creates natural looking image sequences with little distortion in time.

We tackle the human motion imitation, appearance transfer, and novel view synthesis within a unified framework, which means that the model once being trained can be used to handle all these tasks. The existing task-specific methods mainly use 2D keypoints (pose) to estimate the human body structure. However, they only expresses the position information with no abilities to characterize the personalized shape of the individual person and model the limbs rotations. In this paper, we propose to use a 3D body mesh recovery module to disentangle the pose and shape, which can not only model the joint location and rotation but also characterize the personalized body shape. To preserve the source information, such as texture, style, color, and face identity, we propose a Liquid Warping GAN with Liquid Warping Block (LWB) that propagates the source information in both image and feature spaces, and synthesizes an image with respect to the reference. Specifically, the source features are extracted by a denoising convolutional auto-encoder for characterizing the source identity well. Furthermore, our proposed method is able to support a more flexible warping from multiple sources. In addition, we build a new dataset, namely Impersonator (iPER) dataset, for the evaluation of human motion imitation, appearance transfer, and novel view synthesis. Extensive experiments demonstrate the effectiveness of our method in several aspects, such as robustness in occlusion case and preserving face identity, shape consistency and clothes details. All codes and datasets are available on https://svip-lab.github.io/project/impersonator.html.

In recent years, several branch-and-bound (BnB) algorithms have been proposed to globally optimize rigid registration problems. In this paper, we suggest a general framework to improve upon the BnB approach, which we name Quasi BnB. Quasi BnB replaces the linear lower bounds used in BnB algorithms with quadratic quasi-lower bounds which are based on the quadratic behavior of the energy in the vicinity of the global minimum. While quasi-lower bounds are not truly lower bounds, the Quasi-BnB algorithm is globally optimal. In fact we prove that it exhibits linear convergence -- it achieves epsilon accuracy in O(log(1/epsilon)) time while the time complexity of other rigid registration BnB algorithms is polynomial in 1/epsilon. Our experiments verify that Quasi-BnB is significantly more efficient than state-of-the-art BnB algorithms, especially for problems where high accuracy is desired.

We propose a novel image sampling method for differentiable image transformation in deep neural networks. The sampling schemes currently used in deep learning, such as Spatial Transformer Networks, rely on bilinear interpolation, which performs poorly under severe scale changes, and more importantly, results in poor gradient propagation. This is due to their strict reliance on direct neighbors. Instead, we propose to generate random auxiliary samples in the vicinity of each pixel in the sampled image, and create a linear approximation with their intensity values. We then use this approximation as a differentiable formula for the transformed image. We demonstrate that our approach produces more representative gradients with a wider basin of convergence for image alignment, which leads to considerable performance improvements when training networks for registration and classification tasks. This is not only true under large downsampling, but also when there are no scale changes. We compare our approach with multi-scale sampling and show that we outperform it. We then demonstrate that our improvements to the sampler are compatible with other tangential improvements to Spatial Transformer Networks and that it further improves their performance.

Lifelong learning is challenging for deep neural networks due to their susceptibility to catastrophic forgetting. Catastrophic forgetting occurs when a trained network is not able to maintain its ability to accomplish previously learned tasks when it is trained to perform new tasks. We study the problem of lifelong learning for generative models, extending a trained network to new conditional generation tasks without forgetting previous tasks, while assuming access to the training data for the current task only. In contrast to state-of-the-art memory replay based approaches which are limited to label-conditioned image generation tasks, a more generic framework for continual learning of generative models under different conditional image generation settings is proposed in this paper. Lifelong GAN employs knowledge distillation to transfer learned knowledge from previous networks to the new network. This makes it possible to perform image-conditioned generation tasks in a lifelong learning setting. We validate Lifelong GAN for both image-conditioned and label-conditioned generation tasks, and provide qualitative and quantitative results to show the generality and effectiveness of our method.

Single-frame object detectors perform well on videos sometimes, even without temporal context. However, challenges such as occlusion, motion blur, and rare poses of objects are hard to resolve without temporal awareness. Thus, there is a strong need to improve video object detection by considering long-range temporal dependencies. In this paper, we present a light-weight modification to a single-frame detector that accounts for arbitrary long dependencies in a video. It improves the accuracy of a single-frame detector significantly with negligible compute overhead. The key component of our approach is a novel temporal relation module, operating on object proposals, that learns the similarities between proposals from different frames and selects proposals from past and/or future to support current proposals. Our final "causal" model, without any offline post-processing steps, runs at a similar speed as a single-frame detector and achieves state-of-the-art video object detection on ImageNet VID dataset.

Video surveillance systems have wide range of utilities, yet easily suffer from great quality degeneration under dim light circumstances. Industrial solutions mainly use extra near-infrared illuminations, even though it doesn't preserve color and texture information. A variety of researches enhanced low-light videos shot by visible light cameras, while they either relied on task specific preconditions or trained with synthetic datasets. We propose a novel optical system to capture bright and dark videos of the exact same scenes, generating training and groud truth pairs for authentic low-light video dataset. A fully convolutional network with 3D and 2D miscellaneous operations is utilized to learn an enhancement mapping with proper spatial-temporal transformation from raw camera sensor data to bright RGB videos. Experiments show promising results by our method, and it outperforms state-of-the-art low-light image/video enhancement algorithms.

From a single view of an urban environment, we propose a method to effectively exploit the global structural regularities for obtaining a compact, accurate, and intuitive 3D wireframe representation. Our method trains a single convolutional neural network to simultaneously detect salient junctions and straight lines, as well as predict their 3D depth and vanishing points. Compared with state-of-the-art learning-based wireframe detection methods, our network is much simpler and more unified, leading to better 2D wireframe detection. With a global structural prior (such as Manhattan assumption), our method further reconstructs a full 3D wireframe model, a compact vector representation suitable for a variety of high-level vision tasks such as AR and CAD. We conduct extensive evaluations of our method on a large new synthetic dataset of urban scenes as well as real images. Our code and datasets will be published along with the paper.

Model-based human pose estimation is currently approached through two different paradigms. Optimization-based methods fit a parametric body model to 2D observations in an iterative manner, leading to accurate image-model alignments, but are often slow and sensitive to the initialization. In contrast, regression-based methods, that use a deep network to directly estimate the model parameters from pixels, tend to provide reasonable, but not pixel accurate, results while requiring huge amounts of supervision. In this work, instead of investigating which approach is better, our key insight is that the two paradigms can form a strong collaboration. A reasonable, directly regressed estimate from the network can initialize the iterative optimization making the fitting faster and more accurate. Similarly, a pixel accurate fit from iterative optimization can act as strong supervision for the network. This is the core of our proposed approach SPIN (SMPL oPtimization IN the loop). The deep network initializes an iterative optimization routine that fits the body model to 2D joints within the training loop, and the fitted estimate is subsequently used to supervise the network. Our approach is self-improving by nature, since better network estimates can lead the optimization to better solutions, while more accurate optimization fits provide better supervision for the network. We demonstrate the effectiveness of our approach in different settings, where 3D ground truth is scarce, or not available, and we consistently outperform the state-of-the-art model-based pose estimation approaches by significant margins. The project website with videos, results, and code can be found at https://seas.upenn.edu/ nkolot/projects/spin.

Non-Maximum Suppression (NMS) is an essential step of modern object detection models for removing duplicated candidates. The efficacy of NMS heavily affects the final detection results. Prior works exploit suppression criterions relying on either the objectiveness derived from classification or the overlapness produced by regression, both of which are heuristically designed and fail to explicitly link with the suppression rank. To address this issue, in this paper, we propose a novel Learning-to-Rank (LTR) model to produce the suppression rank via a learning procedure, thus facilitating the candidate generation and lifting the detection performance. In particular, we define a ranking score based on IoU to indicate the ranks of candidates during the NMS step, where candidates with high ranking score will be reserved and the ones with low ranking score will be eliminated. We design a lightweight network to predict the ranking score. We introduce a ranking loss to supervise the generation of these ranking scores, which encourages candidates with IoU to the ground-truth to rank higher. To facilitate the training procedure, we design a novel sampling strategy via dividing candidates into different levels and select hard pairs to adopt in the training. During the inference phase, this module can be exploited as a plugin to the current object detector. The training and inference of the overall framework is end-to-end. Comprehensive experiments on benchmarks PASCAL VOC and MS COCO demonstrate the generality and effectiveness of our model for facilitating existing object detectors to state-of-the-art accuracy.