CVPR 2017 论文列表

In this paper, we propose a novel generative model named Stacked Generative Adversarial Networks (SGAN), which is trained to invert the hierarchical representations of a bottom-up discriminative network. Our model consists of a top-down stack of GANs, each learned to generate lower-level representations conditioned on higher-level representations. A representation discriminator is introduced at each feature hierarchy to encourage the representation manifold of the generator to align with that of the bottom-up discriminative network, leveraging the powerful discriminative representations to guide the generative model. In addition, we introduce a conditional loss that encourages the use of conditional information from the layer above, and a novel entropy loss that maximizes a variational lower bound on the conditional entropy of generator outputs. We first train each stack independently, and then train the whole model end-to-end. Unlike the original GAN that uses a single noise vector to represent all the variations, our SGAN decomposes variations into multiple levels and gradually resolves uncertainties in the top-down generative process. Based on visual inspection, Inception scores and visual Turing test, we demonstrate that SGAN is able to generate images of much higher quality than GANs without stacking.

In this work, we propose a novel way of efficiently localizing a sports field from a single broadcast image of the game. Related work in this area relies on manually annotating a few key frames and extending the localization to similar images, or installing fixed specialized cameras in the stadium from which the layout of the field can be obtained. In contrast, we formulate this problem as a branch and bound inference in a Markov random field where an energy function is defined in terms of semantic cues such as the field surface, lines and circles obtained from a deep semantic segmentation network. Moreover, our approach is fully automatic and depends only on a single image from the broadcast video of the game. We demonstrate the effectiveness of our method by applying it to soccer and hockey.

We propose split-brain autoencoders, a straightforward modification of the traditional autoencoder architecture, for unsupervised representation learning. The method adds a split to the network, resulting in two disjoint sub-networks. Each sub-network is trained to perform a difficult task -- predicting one subset of the data channels from another. Together, the sub-networks extract features from the entire input signal. By forcing the network to solve cross-channel prediction tasks, we induce a representation within the network which transfers well to other, unseen tasks. This method achieves state-of-the-art performance on several large-scale transfer learning benchmarks.

Person re-identification (ReID) is an important task in video surveillance and has various applications. It is non-trivial due to complex background clutters, varying illumination conditions, and uncontrollable camera settings. Moreover, the person body misalignment caused by detectors or pose variations is sometimes too severe for feature matching across images. In this study, we propose a novel Convolutional Neural Network (CNN), called Spindle Net, based on human body region guided multi-stage feature decomposition and tree-structured competitive feature fusion. It is the first time human body structure information is considered in a CNN framework to facilitate feature learning. The proposed Spindle Net brings unique advantages: 1) it separately captures semantic features from different body regions thus the macro- and micro-body features can be well aligned across images, 2) the learned region features from different semantic regions are merged with a competitive scheme and discriminative features can be well preserved. State of the art performance can be achieved on multiple datasets by large margins. We further demonstrate the robustness and effectiveness of the proposed Spindle Net on our proposed dataset SenseReID without fine-tuning.

This paper addresses deep face recognition (FR) problem under open-set protocol, where ideal face features are expected to have smaller maximal intra-class distance than minimal inter-class distance under a suitably chosen metric space. However, few existing algorithms can effectively achieve this criterion. To this end, we propose the angular softmax (A-Softmax) loss that enables convolutional neural networks (CNNs) to learn angularly discriminative features. Geometrically, A-Softmax loss can be viewed as imposing discriminative constraints on a hypersphere manifold, which intrinsically matches the prior that faces also lie on a manifold. Moreover, the size of angular margin can be quantitatively adjusted by a parameter m. We further derive specific m to approximate the ideal feature criterion. Extensive analysis and experiments on Labeled Face in the Wild (LFW), Youtube Faces (YTF) and MegaFace Challenge 1 show the superiority of A-Softmax loss in FR tasks.

We present a method for removing specular highlight reflections in facial images that may contain varying illumination colors. This is accurately achieved through the use of physical and statistical properties of human skin and faces. We employ a melanin and hemoglobin based model to represent the diffuse color variations in facial skin, and utilize this model to constrain the highlight removal solution in a manner that is effective even for partially saturated pixels. The removal of highlights is further facilitated through estimation of directionally variant illumination colors over the face, which is done while taking advantage of a statistically-based approximation of facial geometry. An important practical feature of the proposed method is that the skin color model is utilized in a way that does not require color calibration of the camera. Moreover, this approach does not require assumptions commonly needed in previous highlight removal techniques, such as uniform illumination color or piecewise-constant surface colors. We validate this technique through comparisons to existing methods for removing specular highlights.

Two-stream convolutional networks have shown strong performance in video action recognition tasks. The key idea is to learn spatiotemporal features by fusing convolutional networks spatially and temporally. However, it remains unclear how to model the correlations between the spatial and temporal structures at multiple abstraction levels. First, the spatial stream tends to fail if two videos share similar backgrounds. Second, the temporal stream may be fooled if two actions resemble in short snippets, though appear to be distinct in the long term. We propose a novel spatiotemporal pyramid network to fuse the spatial and temporal features in a pyramid structure such that they can reinforce each other. From the architecture perspective, our network constitutes hierarchical fusion strategies which can be trained as a whole using a unified spatiotemporal loss. A series of ablation experiments support the importance of each fusion strategy. From the technical perspective, we introduce the spatiotemporal compact bilinear operator into video analysis tasks. This operator enables efficient training of bilinear fusion operations which can capture full interactions between the spatial and temporal features. Our final network achieves state-of-the-art results on standard video datasets.

This paper presents a general ConvNet architecture for video action recognition based on multiplicative interactions of spacetime features. Our model combines the appearance and motion pathways of a two-stream architecture by motion gating and is trained end-to-end. We theoretically motivate multiplicative gating functions for residual networks and empirically study their effect on classification accuracy. To capture long-term dependencies we inject identity mapping kernels for learning temporal relationships. Our architecture is fully convolutional in spacetime and able to evaluate a video in a single forward pass. Empirical investigation reveals that our model produces state-of-the-art results on two standard action recognition datasets.

We introduce Spatio-Temporal Vector of Locally Max Pooled Features (ST-VLMPF), a super vector-based encoding method specifically designed for local deep features encoding. The proposed method addresses an important problem of video understanding: how to build a video representation that incorporates the CNN features over the entire video. Feature assignment is carried out at two levels, by using the similarity and spatio-temporal information. For each assignment we build a specific encoding, focused on the nature of deep features, with the goal to capture the highest feature responses from the highest neuron activation of the network. Our ST-VLMPF clearly provides a more reliable video representation than some of the most widely used and powerful encoding approaches (Improved Fisher Vectors and Vector of Locally Aggregated Descriptors), while maintaining a low computational complexity. We conduct experiments on three action recognition datasets: HMDB51, UCF50 and UCF101. Our pipeline obtains state-of-the-art results.

In dynamic object detection, it is challenging to construct an effective model to sufficiently characterize the spatial-temporal properties of the background. This paper proposes a new Spatio-Temporal Self-Organizing Map (STSOM) deep network to detect dynamic objects in complex scenarios. The proposed approach has several contributions: First, a novel STSOM shared by all pixels in a video frame is presented to efficiently model complex background. We exploit the fact that the motions of complex background have the global variation in the space and the local variation in the time, to train STSOM using the whole frames and the sequence of a pixel over time to tackle the variance of complex background. Second, a Bayesian parameter estimation based method is presented to learn thresholds automatically for all pixels to filter out the background. Last, in order to model the complex background more accurately, we extend the single-layer STSOM to the deep network. Then the background is filtered out layer by layer. Experimental results on CDnet 2014 dataset demonstrate that the proposed STSOM deep network outperforms numerous recently proposed methods in the overall performance and in most categories of scenarios.

Motivated by previous success of using non-parametric methods to recognize objects, e.g., NBNN, we extend it to recognize actions using skeletons. Each 3D action is presented by a sequence of 3D poses. Similar to NBNN, our proposed Spatio-Temporal-NBNN applies stage-to-class distance to classify actions. However, ST-NBNN takes the spatio-temporal structure of 3D actions into consideration and relaxes the Naive Bayes assumption of NBNN. Specifically, ST-NBNN adopts bilinear classifiers to identify both key temporal stages as well as spatial joints for action classification. Although only using a linear classifier, experiments on three benchmark datasets show that by combining the strength of both non-parametric and parametric models, ST-NBNN can achieve competitive performance compared with state-of-the-art results using sophisticated models such as deep learning. Moreover, by identifying key skeleton joints and temporal stages for each action class, our ST-NBNN can capture the essential spatio-temporal patterns that play key roles of recognizing actions, which is not always achievable by using end-to-end models.

This paper addresses the problem of spatio-temporal alignment of multiple video sequences. We identify and tackle a novel scenario of this problem referred to as Nonoverlapping Sequences (NOS). NOS are captured by multiple freely panning handheld cameras whose field of views (FOV) might have no direct spatial overlap. With the popularity of mobile sensors, NOS rise when multiple cooperative users capture a public event to create a panoramic video, or when consolidating multiple footages of an incident into a single video. To tackle this novel scenario, we first spatially align the sequences by reconstructing the background of each sequence and registering these backgrounds, even if the backgrounds are not overlapping. Given the spatial alignment, we temporally synchronize the sequences, such that the trajectories of moving objects (e.g., cars or pedestrians) are consistent across sequences. Experimental results demonstrate the performance of our algorithm in this novel and challenging scenario, quantitatively and qualitatively.

The detection of spatially-varying blur without having any information about the blur type is a challenging task. In this paper, we propose a novel effective approach to address this blur detection problem from a single image without requiring any knowledge about the blur type, level, or camera settings. Our approach computes blur detection maps based on a novel High-frequency multiscale Fusion and Sort Transform (HiFST) of gradient magnitudes. The evaluations of the proposed approach on a diverse set of blurry images with different blur types, levels, and contents demonstrate that the proposed algorithm performs favorably against the state-of-the-art methods qualitatively and quantitatively.

This paper proposes a deep learning architecture based on Residual Network that dynamically adjusts the number of executed layers for the regions of the image. This architecture is end-to-end trainable, deterministic and problem-agnostic. It is therefore applicable without any modifications to a wide range of computer vision problems such as image classification, object detection and image segmentation. We present experimental results showing that this model improves the computational efficiency of Residual Networks on the challenging ImageNet classification and COCO object detection datasets. Additionally, we evaluate the computation time maps on the visual saliency dataset cat2000 and find that they correlate surprisingly well with human eye fixation positions.

Nonlinear regression is a common statistical tool to solve many computer vision problems (e.g., age estimation, pose estimation). Existing approaches to nonlinear regression fall into two main categories: (1) The universal approach provides an implicit or explicit homogeneous feature mapping (e.g., kernel ridge regression, Gaussian process regression, neural networks). These approaches may fail when data is heterogeneous or discontinuous. (2) Divide-and-conquer approaches partition a heterogeneous input feature space and learn multiple local regressors. However, existing divide-and-conquer approaches fail to deal with discontinuities between partitions (e.g., Gaussian mixture of regressions) and they cannot guarantee that the partitioned input space will be homogeneously modeled by local regressors (e.g., ordinal regression). To address these issues, this paper proposes Soft-Margin Mixture of Regressions (SMMR), a method that directly learns homogeneous partitions of the input space and is able to deal with discontinuities. SMMR outperforms the state-of-the-art methods on three popular computer vision tasks: age estimation, crowd counting and viewpoint estimation from images.

This paper presents the first snapshot hyperspectral light field imager in practice. Specifically, we design a novel hybrid camera system to obtain two complementary measurements that sample the angular and spectral dimensions respectively. To recover the full 5D hyperspectral light field from the severely undersampled measurements, we then propose an efficient computational reconstruction algorithm by exploiting the large correlations across the angular and spectral dimensions through self-learned dictionaries. Simulation on an elaborate hyperspectral light field dataset validates the effectiveness of the proposed approach. Hardware experimental results demonstrate that, for the first time to our knowledge, a 5D hyperspectral light field containing 9x9 angular views and 27 spectral bands can be acquired in a single shot.

Recently, there has been a lot of interest in automatically generating descriptions for an image. Most existing language-model based approaches for this task learn to generate an image description word by word in its original word order. However, for humans, it is more natural to locate the objects and their relationships first, and then elaborate on each object, describing notable attributes. We present a coarse-to-fine method that decomposes the original image description into a skeleton sentence and its attributes, and generates the skeleton sentence and attribute phrases separately. By this decomposition, our method can generate more accurate and novel descriptions than the previous state-of-the-art. Experimental results on the MS-COCO and a larger scale Stock3M datasets show that our algorithm yields consistent improvements across different evaluation metrics, especially on the SPICE metric, which has much higher correlation with human ratings than the conventional metrics. Furthermore, our algorithm can generate descriptions with varied length, benefiting from the separate control of the skeleton and attributes. This enables image description generation that better accommodates user preferences.

Reflections are a common artifact in images taken through glass windows. Automatically removing the reflection artifacts after the picture is taken is an ill-posed problem. Attempts to solve this problem using optimization schemes therefore rely on various prior assumptions from the physical world. Instead of removing reflections from a single image, which has met with limited success so far, we propose a novel approach to suppress reflections. It is based on a Laplacian data fidelity term and an l-zero gradient sparsity term imposed on the output. With experiments on artificial and real-world images we show that our reflection suppression method performs better than the state-of-the-art reflection removal techniques.

Tensor completion has attracted attention because of its promising ability and generality. However, there are few studies on noisy scenarios which directly solve an optimization problem consisting of a "noise inequality constraint". In this paper, we propose a new tensor completion and denoising model including tensor total variation and tensor nuclear norm minimization with a range of values and noise inequalities. Furthermore, we developed its solution algorithm based on a primal-dual splitting method, which is computationally efficient as compared to tensor decomposition based non-convex optimization. Lastly, extensive experiments demonstrated the advantages of the proposed method for visual data retrieval such as for color images, movies, and 3D-volumetric data.