ICCV 2019 论文列表

Appearance information alone is often not sufficient to accurately differentiate between fine-grained visual categories. Human experts make use of additional cues such as where, and when, a given image was taken in order to inform their final decision. This contextual information is readily available in many online image collections but has been underutilized by existing image classifiers that focus solely on making predictions based on the image contents. We propose an efficient spatio-temporal prior, that when conditioned on a geographical location and time, estimates the probability that a given object category occurs at that location. Our prior is trained from presence-only observation data and jointly models object categories, their spatio-temporal distributions, and photographer biases. Experiments performed on multiple challenging image classification datasets show that combining our prior with the predictions from image classifiers results in a large improvement in final classification performance.

Inspired by human neurological structures for action anticipation, we present an action anticipation model that enables the prediction of plausible future actions by forecasting both the visual and temporal future. In contrast to current state-of-the-art methods which first learn a model to predict future video features and then perform action anticipation using these features, the proposed framework jointly learns to perform the two tasks, future visual and temporal representation synthesis, and early action anticipation. The joint learning framework ensures that the predicted future embeddings are informative to the action anticipation task. Furthermore, through extensive experimental evaluations we demonstrate the utility of using both visual and temporal semantics of the scene, and illustrate how this representation synthesis could be achieved through a recurrent Generative Adversarial Network (GAN) framework. Our model outperforms the current state-of-the-art methods on multiple datasets: UCF101, UCF101-24, UT-Interaction and TV Human Interaction.

Given a video of a person in action, we can easily guess the 3D future motion of the person. In this work, we present perhaps the first approach for predicting a future 3D mesh model sequence of a person from past video input. We do this for periodic motions such as walking and also actions like bowling and squatting seen in sports or workout videos. While there has been a surge of future prediction problems in computer vision, most approaches predict 3D future from 3D past or 2D future from 2D past inputs. In this work, we focus on the problem of predicting 3D future motion from past image sequences, which has a plethora of practical applications in autonomous systems that must operate safely around people from visual inputs. Inspired by the success of autoregressive models in language modeling tasks, we learn an intermediate latent space on which we predict the future. This effectively facilitates autoregressive predictions when the input differs from the output domain. Our approach can be trained on video sequences obtained in-the-wild without 3D ground truth labels. The project website with videos can be found at https://jasonyzhang.com/phd.

Persons are often occluded by various obstacles in person retrieval scenarios. Previous person re-identification (re-id) methods, either overlook this issue or resolve it based on an extreme assumption. To alleviate the occlusion problem, we propose to detect the occluded regions, and explicitly exclude those regions during feature generation and matching. In this paper, we introduce a novel method named Pose-Guided Feature Alignment (PGFA), exploiting pose landmarks to disentangle the useful information from the occlusion noise. During the feature constructing stage, our method utilizes human landmarks to generate attention maps. The generated attention maps indicate if a specific body part is occluded and guide our model to attend to the non-occluded regions. During matching, we explicitly partition the global feature into parts and use the pose landmarks to indicate which partial features belonging to the target person. Only the visible regions are utilized for the retrieval. Besides, we construct a large-scale dataset for the Occluded Person Re-ID problem, namely Occluded-DukeMTMC, which is by far the largest dataset for the Occlusion Person Re-ID. Extensive experiments are conducted on our constructed occluded re-id dataset, two partial re-id datasets, and two commonly used holistic re-id datasets. Our method largely outperforms existing person re-id methods on three occlusion datasets, while remains top performance on two holistic datasets.

In this paper we consider the problem of relative pose estimation from two images with per-pixel polarimetric information. Using these additional measurements we derive a simple minimal solver for the essential matrix which only requires two point correspondences. The polarization constraints allow us to pointwise recover the 3D surface normal up to a two-fold ambiguity for the diffuse reflection. Since this ambiguity exists per point, there is a combinatorial explosion of possibilities. However, since our solver only requires two point correspondences, we only need to consider 16 configurations when solving for the relative pose. Once the relative orientation is recovered, we show that it is trivial to resolve the ambiguity for the remaining points. For robustness, we also propose a joint optimization between the relative pose and the refractive index to handle the refractive distortion. In experiments, on both synthetic and real data, we demonstrate that by leveraging the additional information available from polarization cameras, we can improve over classical methods which only rely on the 2D-point locations to estimate the geometry. Finally, we demonstrate the practical applicability of our approach by integrating it into a state-of-the-art global Structure-from-Motion pipeline.

The outcome of standard statistical shape modelling is a vector space representation of objects. Any convex combination of vectors of a set of object class examples generates a real and valid example. In this paper, we propose a Point Auto-Encoder (PointAE) with skip-connection, attention blocks for 3D statistical shape modelling directly on 3D points. The proposed PointAE is able to refine the correspondence with a correspondence refinement block. The data with refined correspondence can be fed to the PointAE again and bootstrap the constructed statistical models. Instead of two seperate models, PointAE can simultaneously model the shape and texture variation. The extensive evaluation in three open-sourced datasets demonstrates that the proposed method achieves better performance in representation ability of the shape variations.

While image synthesis achieves tremendous breakthroughs (e.g., generating realistic faces), video generation is less explored and harder to control, which limits its applications in the real world. For instance, video editing requires temporal coherence across multiple clips and thus poses both start and end constraints within a video sequence. We introduce point-to-point video generation that controls the generation process with two control points: the targeted start- and end-frames. The task is challenging since the model not only generates a smooth transition of frames but also plans ahead to ensure that the generated end-frame conforms to the targeted end-frame for videos of various lengths. We propose to maximize the modified variational lower bound of conditional data likelihood under a skip-frame training strategy. Our model can generate end-frame-consistent sequences without loss of quality and diversity. We evaluate our method through extensive experiments on Stochastic Moving MNIST, Weizmann Action, Human3.6M, and BAIR Robot Pushing under a series of scenarios. The qualitative results showcase the effectiveness and merits of point-to-point generation.

We study the problem of shape generation in 3D mesh representation from a few color images with known camera poses. While many previous works learn to hallucinate the shape directly from priors, we resort to further improving the shape quality by leveraging cross-view information with a graph convolutional network. Instead of building a direct mapping function from images to 3D shape, our model learns to predict series of deformations to improve a coarse shape iteratively. Inspired by traditional multiple view geometry methods, our network samples nearby area around the initial mesh's vertex locations and reasons an optimal deformation using perceptual feature statistics built from multiple input images. Extensive experiments show that our model produces accurate 3D shape that are not only visually plausible from the input perspectives, but also well aligned to arbitrary viewpoints. With the help of physically driven architecture, our model also exhibits generalization capability across different semantic categories, number of input images, and quality of mesh initialization.

Recovering the 3D representation of an object from single-view or multi-view RGB images by deep neural networks has attracted increasing attention in the past few years. Several mainstream works (e.g., 3D-R2N2) use recurrent neural networks (RNNs) to fuse multiple feature maps extracted from input images sequentially. However, when given the same set of input images with different orders, RNN-based approaches are unable to produce consistent reconstruction results. Moreover, due to long-term memory loss, RNNs cannot fully exploit input images to refine reconstruction results. To solve these problems, we propose a novel framework for single-view and multi-view 3D reconstruction, named Pix2Vox. By using a well-designed encoder-decoder, it generates a coarse 3D volume from each input image. Then, a context-aware fusion module is introduced to adaptively select high-quality reconstructions for each part (e.g., table legs) from different coarse 3D volumes to obtain a fused 3D volume. Finally, a refiner further refines the fused 3D volume to generate the final output. Experimental results on the ShapeNet and Pix3D benchmarks indicate that the proposed Pix2Vox outperforms state-of-the-arts by a large margin. Furthermore, the proposed method is 24 times faster than 3D-R2N2 in terms of backward inference time. The experiments on ShapeNet unseen 3D categories have shown the superior generalization abilities of our method.

To improve the robustness to rain, we present a physically-based rain rendering pipeline for realistically inserting rain into clear weather images. Our rendering relies on a physical particle simulator, an estimation of the scene lighting and an accurate rain photometric modeling to augment images with arbitrary amount of realistic rain or fog. We validate our rendering with a user study, proving our rain is judged 40% more realistic that state-of-the-art. Using our generated weather augmented Kitti and Cityscapes dataset, we conduct a thorough evaluation of deep object detection and semantic segmentation algorithms and show that their performance decreases in degraded weather, on the order of 15% for object detection and 60% for semantic segmentation. Furthermore, we show refining existing networks with our augmented images improves the robustness of both object detection and semantic segmentation algorithms. We experiment on nuScenes and measure an improvement of 15% for object detection and 35% for semantic segmentation compared to original rainy performance. Augmented databases and code are available on the project page.

We present a method for creating inconspicuous-looking textures that, when displayed as posters in the physical world, cause visual object tracking systems to become confused. As a target being visually tracked moves in front of such a poster, its adversarial texture makes the tracker lock onto it, thus allowing the target to evade. This adversarial attack evaluates several optimization strategies for fooling seldom-targeted regression models: non-targeted, targeted, and a newly-coined family of guided adversarial losses. Also, while we use the Expectation Over Transformation (EOT) algorithm to generate physical adversaries that fool tracking models when imaged under diverse conditions, we compare the impacts of different scene variables to find practical attack setups with high resulting adversarial strength and convergence speed. We further showcase that textures optimized using simulated scenes can confuse real-world tracking systems for cameras and robots.

Recent style transfer models have provided promising artistic results. However, given a photograph as a reference style, existing methods are limited by spatial distortions or unrealistic artifacts, which should not happen in real photographs. We introduce a theoretically sound correction to the network architecture that remarkably enhances photorealism and faithfully transfers the style. The key ingredient of our method is wavelet transforms that naturally fits in deep networks. We propose a wavelet corrected transfer based on whitening and coloring transforms (WCT2) that allows features to preserve their structural information and statistical properties of VGG feature space during stylization. This is the first and the only end-to-end model that can stylize a 1024x1024 resolution image in 4.7 seconds, giving a pleasing and photorealistic quality without any post-processing. Last but not least, our model provides a stable video stylization without temporal constraints. Our code, generated images, pre-trained models and supplementary documents are all available at https://github.com/ClovaAI/WCT2.

Gaze redirection is the task of changing the gaze to a desired direction for a given monocular eye patch image. Many applications such as videoconferencing, films, games, and generation of training data for gaze estimation require redirecting the gaze, without distorting the appearance of the area surrounding the eye and while producing photo-realistic images. Existing methods lack the ability to generate perceptually plausible images. In this work, we present a novel method to alleviate this problem by leveraging generative adversarial training to synthesize an eye image conditioned on a target gaze direction. Our method ensures perceptual similarity and consistency of synthesized images to the real images. Furthermore, a gaze estimation loss is used to control the gaze direction accurately. To attain high-quality images, we incorporate perceptual and cycle consistency losses into our architecture. In extensive evaluations we show that the proposed method outperforms state-of-the-art approaches in terms of both image quality and redirection precision. Finally, we show that generated images can bring significant improvement for the gaze estimation task if used to augment real training data.

In this paper, we propose a novel perspective-guided convolution (PGC) for convolutional neural network (CNN) based crowd counting (i.e. PGCNet), which aims to overcome the dramatic intra-scene scale variations of people due to the perspective effect. While most state-of-the-arts adopt multi-scale or multi-column architectures to address such issue, they generally fail in modeling continuous scale variations since only discrete representative scales are considered. PGCNet, on the other hand, utilizes perspective information to guide the spatially variant smoothing of feature maps before feeding them to the successive convolutions. An effective perspective estimation branch is also introduced to PGCNet, which can be trained in either supervised setting or weakly-supervised setting when the branch has been pre-trained. Our PGCNet is single-column with moderate increase in computation, and extensive experimental results on four benchmark datasets show the improvements of our method against the state-of-the-arts. Additionally, we also introduce Crowd Surveillance, a large scale dataset for crowd counting that contains 13,000+ high-resolution images with challenging scenarios. Code is available at https://github.com/Zhaoyi-Yan/PGCNet.

Fashion recommendation is the task of suggesting a fashion item that fits well with a given item. In this work, we propose to automatically synthesis new items for recommendation. We jointly consider the two key issues for the task, i.e., compatibility and personalization. We propose a personalized fashion design framework with the help of generative adversarial training. A convolutional network is first used to map the query image into a latent vector representation. This latent representation, together with another vector which characterizes user's style preference, are taken as the input to the generator network to generate the target item image. Two discriminator networks are built to guide the generation process. One is the classic real/fake discriminator. The other is a matching network which simultaneously models the compatibility between fashion items and learns users' preference representations. The performance of the proposed method is evaluated on thousands of outfits composited by online users. The experiments show that the items generated by our model are quite realistic. They have better visual quality and higher matching degree than those generated by alternative methods.

Fine-grained person perception such as body segmentation and pose estimation has been achieved with many 2D and 3D sensors such as RGB/depth cameras, radars (e.g. RF-Pose), and LiDARs. These solutions require 2D images, depth maps or 3D point clouds of person bodies as input. In this paper, we take one step forward to show that fine-grained person perception is possible even with 1D sensors: WiFi antennas. Specifically, we used two sets of WiFi antennas to acquire signals, i.e., one transmitter set and one receiver set. Each set contains three antennas horizontally lined-up as a regular household WiFi router. The WiFi signal generated by a transmitter antenna, penetrates through and reflects on human bodies, furniture, and walls, and then superposes at a receiver antenna as 1D signal samples. We developed a deep learning approach that uses annotations on 2D images, takes the received 1D WiFi signals as input, and performs body segmentation and pose estimation in an end-to-end manner. To our knowledge, our solution is the first work based on off-the-shelf WiFi antennas and standard IEEE 802.11n WiFi signals. Demonstrating comparable results to image-based solutions, our WiFi-based person perception solution is cheaper and more ubiquitous than radars and LiDARs, while invariant to illumination and has little privacy concern comparing to cameras.

Existing person search methods predominantly assume the availability of at least one-shot imagery sample of the queried person. This assumption is limited in circumstances where only a brief textual (or verbal) description of the target person is available. In this work, we present a deep learning method for attribute text description based person search without any query imagery. Whilst conventional cross-modality matching methods, such as global visual-textual embedding based zero-shot learning and local individual attribute recognition, are functionally applicable, they are limited by several assumptions invalid to person search in deployment scale, data quality, and/or category name semantics. We overcome these issues by formulating an Attribute-Image Hierarchical Matching (AIHM) model. It is able to more reliably match text attribute descriptions with noisy surveillance person images by jointly learning global category-level and local attribute-level textual-visual embedding as well as matching. Extensive evaluations demonstrate the superiority of our AIHM model over a wide variety of state-of-the-art methods on three publicly available attribute labelled surveillance person search benchmarks: Market-1501, DukeMTMC, and PA100K.

We consider the problem of comparing the similarity of image sets with variable-quantity, quality and un-ordered heterogeneous images. We use feature restructuring to exploit the correlations of both inner&inter-set images. Specifically, the residual self-attention can effectively restructure the features using the other features within a set to emphasize the discriminative images and eliminate the redundancy. Then, a sparse/collaborative learning-based dependency-guided representation scheme reconstructs the probe features conditional to the gallery features in order to adaptively align the two sets. This enables our framework to be compatible with both verification and open-set identification. We show that the parametric self-attention network and non-parametric dictionary learning can be trained end-to-end by a unified alternative optimization scheme, and that the full framework is permutation-invariant. In the numerical experiments we conducted, our method achieves top performance on competitive image set/video-based face recognition and person re-identification benchmarks.

RGBD images, combining high-resolution color and lower-resolution depth from various types of depth sensors, are increasingly common. One can significantly improve the resolution of depth maps by taking advantage of color information; deep learning methods make combining color and depth information particularly easy. However, fusing these two sources of data may lead to a variety of artifacts. If depth maps are used to reconstruct 3D shapes, e.g., for virtual reality applications, the visual quality of upsampled images is particularly important. The main idea of our approach is to measure the quality of depth map upsampling using renderings of resulting 3D surfaces. We demonstrate that a simple visual appearance-based loss, when used with either a trained CNN or simply a deep prior, yields significantly improved 3D shapes, as measured by a number of existing perceptual metrics. We compare this approach with a number of existing optimization and learning-based techniques.