CVPR 2017 论文列表

3D models provide a common ground for different representations of human bodies. In turn, robust 2D estimation has proven to be a powerful tool to obtain 3D fits "in-the-wild". However, depending on the level of detail, it can be hard to impossible to acquire labeled data for training 2D estimators on large scale. We propose a hybrid approach to this problem: with an extended version of the recently introduced SMPLify method, we obtain high quality 3D body model fits for multiple human pose datasets. Human annotators solely sort good and bad fits. This procedure leads to an initial dataset, UP-3D, with rich annotations. With a comprehensive set of experiments, we show how this data can be used to train discriminative models that produce results with an unprecedented level of detail: our models predict 31 segments and 91 landmark locations on the body. Using the 91 landmark pose estimator, we present state-of-the art results for 3D human pose and shape estimation using an order of magnitude less training data and without assumptions about gender or pose in the fitting procedure. We show that UP-3D can be enhanced with these improved fits to grow in quantity and quality, which makes the system deployable on large scale. The data, code and models are available for research purposes.

Event detection in unconstrained videos is conceived as a content-based video retrieval with two modalities: textual and visual. Given a text describing a novel event, the goal is to rank related videos accordingly. This task is zero-exemplar, no video examples are given to the novel event. Related works train a bank of concept detectors on external data sources. These detectors predict confidence scores for test videos, which are ranked and retrieved accordingly. In contrast, we learn a joint space in which the visual and textual representations are embedded. The space casts a novel event as a probability of pre-defined events. Also, it learns to measure the distance between an event and its related videos. Our model is trained end-to-end on publicly available EventNet. When applied to TRECVID Multimedia Event Detection dataset, it outperforms the state-of-the-art by a considerable margin.

Understanding traffic density from large-scale web camera (webcam) videos is a challenging problem because such videos have low spatial and temporal resolution, high occlusion and large perspective. To deeply understand traffic density, we explore both optimization based and deep learning based methods. To avoid individual vehicle detection or tracking, both methods map the dense image feature into vehicle density, one based on rank constrained regression and the other based on fully convolutional networks (FCN). The regression based method learns different weights for different blocks of the image to embed road geometry and significantly reduce the error induced by camera perspective. The FCN based method jointly estimates vehicle density and vehicle count with a residual learning framework to perform end-to-end dense prediction, allowing arbitrary image resolution, and adapting to different vehicle scales and perspectives. We analyze and compare both methods, and get insights from optimization based method to improve deep model. Since existing datasets do not cover all the challenges in our work, we collected and labelled a large-scale traffic video dataset, containing 60 million frames from 212 webcams. Both methods are extensively evaluated and compared on different counting tasks and datasets. FCN based method significantly reduces the mean absolute error (MAE) from 10.99 to 5.31 on the public dataset TRANCOS compared with the state-of-the-art baseline.

This paper proposes an algorithm that turns a regular video capturing urban scenes into a high-quality endless animation, known as a Cinemagraph. The creation of a Cinemagraph usually requires a static camera in a carefully configured scene. The task becomes challenging for a regular video with a moving camera and objects. Our approach first warps an input video into the viewpoint of a reference camera. Based on the warped video, we propose effective temporal analysis algorithms to detect regions with static geometry and dynamic appearance, where geometric modeling is reliable and visually attractive animations can be created. Lastly, the algorithm applies a sequence of video processing techniques to produce a Cinemagraph movie. We have tested the proposed approach on numerous challenging real scenes. To our knowledge, this work is the first to automatically generate Cinemagraph animations from regular movies in the wild.

Truncated convex models (TCM) are a special case of pair- wise random fields that have been widely used in computer vision. However, by restricting the order of the potentials to be at most two, they fail to capture useful image statistics. We propose a natural generalization of TCM to high-order random fields, which we call truncated max-of-convex models (TMCM). The energy function of TMCM consists of two types of potentials: (i) unary potential, which has no restriction on its form; and (ii) high-order potential, which is the sum of the truncation of the m largest convex distances over disjoint pairs of random variables in an arbitrary size clique. The use of a convex distance function encourages smoothness, while truncation allows for discontinuities in the labeling. By using m > 1, TMCM provides robustness towards errors in the definition of the cliques. In order to minimize the energy function of a TMCM over all possible labelings, we design an efficient st-mincut based range expansion algorithm. We prove the accuracy of our algorithm by establishing strong multiplicative bounds for several special cases of interest. Using synthetic and standard real datasets, we demonstrate the benefit of our high-order TMCM over pairwise TCM, as well as the benefit of our range expansion algorithm over other st-mincut based approaches.

A human action can be seen as transitions between one's body poses over time, where the transition depicts a temporal relation between two poses. Recognizing actions thus involves learning a classifier sensitive to these pose transitions as well as to static poses. In this paper, we introduce a novel method called transitions forests, an ensemble of decision trees that both learn to discriminate static poses and transitions between pairs of two independent frames. During training, node splitting is driven by alternating two criteria: the standard classification objective that maximizes the discrimination power in individual frames, and the proposed one in pairwise frame transitions. Growing the trees tends to group frames that have similar associated transitions and share same action label incorporating temporal information that was not available otherwise. Unlike conventional decision trees where the best split in a node is determined independently of other nodes, the transition forests try to find the best split of nodes jointly (within a layer) for incorporating distant node transitions. When inferring the class label of a new frame, it is passed down the trees and the prediction is made based on previous frame predictions and the current one in an efficient and online manner. We apply our method on varied skeleton action recognition and online detection datasets showing its suitability over several baselines and state-of-the-art approaches.

We present a transformation-grounded image generation network for novel 3D view synthesis from a single image. Our approach first explicitly infers the parts of the geometry visible both in the input and novel views and then casts the remaining synthesis problem as image completion. Specifically, we both predict a flow to move the pixels from the input to the novel view along with a novel visibility map that helps deal with occulsion/disocculsion. Next, conditioned on those intermediate results, we hallucinate (infer) parts of the object invisible in the input image. In addition to the new network structure, training with a combination of adversarial and perceptual loss results in a reduction in common artifacts of novel view synthesis such as distortions and holes, while successfully generating high frequency details and preserving visual aspects of the input image. We evaluate our approach on a wide range of synthetic and real examples. Both qualitative and quantitative results show our method achieves significantly better results compared to existing methods.

This paper strives to track a target object in a video. Rather than specifying the target in the first frame of a video by a bounding box, we propose to track the object based on a natural language specification of the target, which provides a more natural human-machine interaction as well as a means to improve tracking results. We define three variants of tracking by language specification: one relying on lingual target specification only, one relying on visual target specification based on language, and one leveraging their joint capacity. To show the potential of tracking by natural language specification we extend two popular tracking datasets with lingual descriptions and report experiments. Finally, we also sketch new tracking scenarios in surveillance and other live video streams that become feasible with a lingual specification of the target.

Light fields become a popular representation of three-dimensional scenes, and there is interest in their processing, resampling, and compression. As those operations often result in loss of quality, there is a need to quantify it. In this work, we collect a new dataset of dense reference and distorted light fields as well as the corresponding quality scores which are scaled in perceptual units. The scores were acquired in a subjective experiment using an interactive light-field viewing setup. The dataset contains typical artifacts that occur in light-field processing chain due to light-field reconstruction, multi-view compression, and limitations of automultiscopic displays. We test a number of existing objective quality metrics to determine how well they can predict the quality of light fields. We find that the existing image quality metrics provide good measures of light-field quality, but require dense reference light- fields for optimal performance. For more complex tasks of comparing two distorted light fields, their performance drops significantly, which reveals the need for new, light-field-specific metrics.

We propose a method for multi-person detection and 2-D pose estimation that achieves state-of-art results on the challenging COCO keypoints task. It is a simple, yet powerful, top-down approach consisting of two stages. In the first stage, we predict the location and scale of boxes which are likely to contain people; for this we use the Faster RCNN detector. In the second stage, we estimate the keypoints of the person potentially contained in each proposed bounding box. For each keypoint type we predict dense heatmaps and offsets using a fully convolutional ResNet. To combine these outputs we introduce a novel aggregation procedure to obtain highly localized keypoint predictions. We also use a novel form of keypoint-based Non-Maximum-Suppression (NMS), instead of the cruder box-level NMS, and a novel form of keypoint-based confidence score estimation, instead of box-level scoring. Trained on COCO data alone, our final system achieves average precision of 0.649 on the COCO test-dev set and the 0.643 test-standard sets, outperforming the winner of the 2016 COCO keypoints challenge and other recent state-of-art. Further, by using additional in-house labeled data we obtain an even higher average precision of 0.685 on the test-dev set and 0.673 on the test-standard set, more than 5% absolute improvement compared to the previous best performing method on the same dataset.

Localizing a query image against a 3D model at large scale is a hard problem, since 2D-3D matches become more and more ambiguous as the model size increases. This creates a need for pose estimation strategies that can handle very low inlier ratios. In this paper, we draw new insights on the geometric information available from the 2D-3D matching process. As modern descriptors are not invariant against large variations in viewpoint, we are able to find the rays in space used to triangulate a given point that are closest to a query descriptor. It is well known that two correspondences constrain the camera to lie on the surface of a torus. Adding the knowledge of direction of triangulation, we are able to approximate the position of the camera from two matches alone. We derive a geometric solver that can compute this position in under 1 microsecond. Using this solver, we propose a simple yet powerful outlier filter which scales quadratically in the number of matches. We validate the accuracy of our solver and demonstrate the usefulness of our method in real world settings.

Neural image/video captioning models can generate accurate descriptions, but their internal process of mapping regions to words is a black box and therefore difficult to explain. Top-down neural saliency methods can find important regions given a high-level semantic task such as object classification, but cannot use a natural language sentence as the top-down input for the task. In this paper, we propose Caption-Guided Visual Saliency to expose the region-to-word mapping in modern encoder-decoder networks and demonstrate that it is learned implicitly from caption training data, without any pixel-level annotations. Our approach can produce spatial or spatiotemporal heatmaps for both predicted captions, and for arbitrary query sentences. It recovers saliency without the overhead of introducing explicit attention layers, and can be used to analyze a variety of existing model architectures and improve their design. Evaluation on large-scale video and image datasets demonstrates that our approach achieves comparable captioning performance with existing methods while providing more accurate saliency heatmaps. Our code is available at visionlearninggroup.github.io/caption-guided-saliency/

The notion of a Fast Moving Object (FMO), i.e. an object that moves over a distance exceeding its size within the exposure time, is introduced. FMOs may, and typically do, rotate with high angular speed. FMOs are very common in sports videos, but are not rare elsewhere. In a single frame, such objects are often barely visible and appear as semitransparent streaks. A method for the detection and tracking of FMOs is proposed. The method consists of three distinct algorithms, which form an efficient localization pipeline that operates successfully in a broad range of conditions. We show that it is possible to recover the appearance of the object and its axis of rotation, despite its blurred appearance. The proposed method is evaluated on a new annotated dataset. The results show that existing trackers are inadequate for the problem of FMO localization and a new approach is required. Two applications of localization, temporal superresolution and highlighting, are presented.

One of the most intriguing features of the Visual Question Answering (VQA) challenge is the unpredictability of the questions. Extracting the information required to answer them demands a variety of image operations from detection and counting, to segmentation and reconstruction. To train a method to perform even one of these operations accurately from image,question,answer tuples would be challenging, but to aim to achieve them all with a limited set of such training data seems ambitious at best. Our method thus learns how to exploit a set of external off-the-shelf algorithms to achieve its goal, an approach that has something in common with the Neural Turing Machine. The core of our proposed method is a new co-attention model. In addition, the proposed approach generates human-readable reasons for its decision, and can still be trained end-to-end without ground truth reasons being given. We demonstrate the effectiveness on two publicly available datasets, Visual Genome and VQA, and show that it produces the state-of-the-art results in both cases.

The three-dimensional reconstruction of scenes from multiple views has made impressive strides in recent years, chiefly by methods correlating isolated feature points, intensities, or curvilinear structure. In the general setting, i.e., without requiring controlled acquisition, limited number of objects, abundant patterns on objects, or object curves to follow particular models, the majority of these methods produce unorganized point clouds, meshes, or voxel representations of the reconstructed scene, with some exceptions producing 3D drawings as networks of curves. Many applications, e.g., robotics, urban planning, industrial design, and hard surface modeling, however, require structured representations which make explicit 3D curves, surfaces, and their spatial relationships. Reconstructing surface representations can now be constrained by the 3D drawing acting like a scaffold to hang on the computed representations, leading to increased robustness and quality of reconstruction. This paper presents one way of completing such 3D drawings with surface reconstructions, by exploring occlusion reasoning through lofting algorithms.

There is a significant interest in scene reconstruction from underwater images given its utility for oceanic research and for recreational image manipulation. In this paper we propose a novel algorithm for two view camera motion estimation for underwater imagery. Our method leverages the constraints provided by the attenuation properties of water and its effects on the appearance of the color to determine the depth difference of a point with respect to the two observing views of the underwater cameras. Additionally, we propose an algorithm, leveraging the depth differences of three such observed points, to estimate the relative pose of the cameras. Given the unknown underwater attenuation coefficients, our method estimates the relative motion up to scale. The results are represented as a generalized camera. We evaluate our method on both real data and simulated data.

Multiresolution analysis and matrix factorization are foundational tools in computer vision. In this work, we study the interface between these two distinct topics and obtain techniques to uncover hierarchical block structure in symmetric matrices -- an important aspect in the success of many vision problems. Our new algorithm, the incremental multiresolution matrix factorization, uncovers such structure one feature at a time, and hence scales well to large matrices. We describe how this multiscale analysis goes much farther than what a direct "global" factorization of the data can identify. We evaluate the efficacy of the resulting factorizations for relative leveraging within regression tasks using medical imaging data. We also use the factorization on representations learned by popular deep networks, providing evidence of their ability to infer semantic relationships even when they are not explicitly trained to do so. We show that this algorithm can be used as an exploratory tool to improve the network architecture, and within numerous other settings in vision.

In computer vision, selection of the most informative samples from a huge pool of training data in order to learn a good recognition model is an active research problem. Furthermore, it is also useful to reduce the annotation cost, as it is time consuming to annotate unlabeled samples. In this paper, motivated by the theories in data compression, we propose a novel sample selection strategy which exploits the concept of typicality from the domain of information theory. Typicality is a simple and powerful technique which can be applied to compress the training data to learn a good classification model. In this work, typicality is used to identify a subset of the most informative samples for labeling, which is then used to update the model using active learning. The proposed model can take advantage of the inter-relationships between data samples. Our approach leads to a significant reduction of manual labeling cost while achieving similar or better recognition performance compared to a model trained with entire training set. This is demonstrated through rigorous experimentation on five datasets.

This paper combines three contributions to establish a new state-of-the-art in dynamic scene recognition. First, we present a novel ConvNet architecture based on temporal residual units that is fully convolutional in spacetime. Our model augments spatial ResNets with convolutions across time to hierarchically add temporal residuals as the depth of the network increases. Second, existing approaches to video-based recognition are categorized and a baseline of seven previously top performing algorithms is selected for comparative evaluation on dynamic scenes. Third, we introduce a new and challenging video database of dynamic scenes that more than doubles the size of those previously available. This dataset is explicitly split into two subsets of equal size that contain videos with and without camera motion to allow for systematic study of how this variable interacts with the defining dynamics of the scene per se. Our evaluations verify the particular strengths and weaknesses of the baseline algorithms with respect to various scene classes and camera motion parameters. Finally, our temporal ResNet boosts recognition performance and establishes a new state-of-the-art on dynamic scene recognition, as well as on the complementary task of action recognition.