CVPR 2016 论文列表

Hashing has become an increasingly popular technique for fast nearest neighbor search in large databases. Despite its successful progress in classic point-to-point search, there are few studies regarding point-to-hyperplane search, which has strong practical capabilities of scaling up in many applications like active learning with SVMs. Existing hyperplane hashing methods enable the fast search based on the randomly generated hash codes, but still suffer from a low collision probability and thus usually require long codes for a satisfying performance. To overcome this problem, this paper proposes a multilinear hyperplane hashing that generates a hash bit using multiple linear projections. Our theoretical analysis shows that as a product of an even number of random linear projections, the multilinear hash function possesses an increasing power of locality sensitivity to the hyperplane queries. To leverage its sensitivity to the angle distance, we further introduce an angular quantization based learning framework for compact multilinear hashing, which considerably boosts the search performance with less hash bits. Experiments with applications to large-scale (up to one million) active learning on two datasets demonstrate the overall superiority of the proposed approach.

Multicamera rigs are used in a large number of 3D Vision applications, such as 3D modeling, motion capture or telepresence and a robust calibration is of utmost importance in order to achieve a high accuracy results. In many practical configurations the cameras in a rig are arranged in such a way, that they can observe each other, in other words a number of epipoles correspond to the real image points. In this paper we propose a solution for the automatic recovery of the external calibration of a multicamera system by enforcing only simple geometrical constraints, arising from the epipole visibility, without using any calibration object, such as checkerboards, laser pointers or similar. Additionally, we introduce an extension of the method that handles the case of epipoles being visible in the reflection of a planar mirror, which makes the algorithm suitable for the calibration of any multicamera system, irrespective of the number of cameras and their actual mutual visibility, and furthermore we remark that it requires only one or a few images per camera and therefore features a high speed and usability. We produce an evidence of the algorithm effectiveness by presenting a wide set of tests performed on synthetic as well as real datasets and we compare the results with those obtained using a traditional LED-based algorithm. The real datasets have been captured using a multicamera Virtual Reality (VR) rig and a spherical dome configuration for 3D reconstruction.

This paper presents a hierarchical composition approach for multi-view object tracking. The key idea is to adaptively exploit multiple cues in both 2D and 3D, e.g., ground occupancy consistency, appearance similarity, motion coherence etc., which are mutually complementary while tracking the humans of interests over time. While feature online selection has been extensively studied in the past literature, it remains unclear how to effectively schedule these cues for the tracking purpose especially when encountering various challenges, e.g. occlusions, conjunctions, and appearance variations. To do so, we propose a hierarchical composition model and re-formulate multi-view multi-object tracking as a problem of compositional structure optimization. We setup a set of composition criteria, each of which corresponds to one particular cue. The hierarchical composition process is pursued by exploiting different criteria, which impose constraints between a graph node and its offsprings in the hierarchy. We learn the composition criteria using MLE on annotated data and efficiently construct the hierarchical graph by an iterative greedy pursuit algorithm. In the experiments, we demonstrate superior performance of our approach on three public datasets, one of which is newly created by us to test various challenges in multi-view multi-object tracking.

Cross-view recognition that intends to classify samples between different views is an important problem in computer vision. The large discrepancy between different even heterogenous views make this problem quite challenging. To eliminate the complex (maybe even highly nonlinear) view discrepancy for favorable cross-view recognition, we propose a multi-view deep network (MvDN), which seeks for a non-linear discriminant and view-invariant representation shared between multiple views. Specifically, our proposed MvDN network consists of two sub-networks, view-specific sub-network attempting to remove view-specific variations and the following common sub-network attempting to obtain common representation shared by all views. As the objective of MvDN network, the Fisher loss, i.e. the Rayleigh quotient objective, is calculated from the samples of all views so as to guide the learning of the whole network. As a result, the representation from the topmost layers of the MvDN network is robust to view discrepancy, and also discriminative. The experiments of face recognition across pose and face recognition across feature type on three datasets with 13 and 2 views respectively demonstrate the superiority of the proposed method, especially compared to the typical linear ones.

In this paper, we propose an unconventional approach for text detection in natural images. Both global and local cues are taken into account for localizing text lines in a coarse-to-fine procedure. First, a Fully Convolutional Network (FCN) model is trained for predicting a salient map of text regions in a holistic manner. Then, a set of hypotheses text lines are estimated by combining the salient map and MSER components. Finally, another FCN classifier is used for predicting the centroid of each character, in order to remove the false hypotheses. The framework is general for handling texts in multiple orientations, languages and fonts. The proposed method consistently achieves the state-of-the-art performance on three text detection benchmarks: MSRA-TD500, ICDAR2015, and ICDAR2013.

In this paper, we specifically examine the training of a multi-label classifier from data with incompletely assigned labels. This problem is fundamentally important in many multi-label applications because it is almost impossible for human annotators to assign a complete set of labels, although their judgments are reliable. In other words, a multi-label dataset usually has properties by which (1) assigned labels are definitely positive and (2) some labels are absent but are still considered positive. Such a setting has been studied as a positive and unlabeled (PU) classification problem in a binary setting. We treat incomplete label assignment problems as a multi-label PU ranking, which is an extension of classical binary PU problems to the well-studied rank-based multi-label classification. We derive the conditions that should be satisfied to cancel the negative effects of label incompleteness. Our experimentally obtained results demonstrate the effectiveness of these conditions.

Lineage tracing, the tracking of living cells as they move and divide, is a central problem in biological image analysis. Solutions, called lineage forests, are key to understanding how the structure of multicellular organisms emerges. We propose an integer linear program (ILP) whose feasible solutions define, for every image in a sequence, a decomposition into cells (segmentation) and, across images, a lineage forest of cells (tracing). In this ILP, path-cut inequalities enforce the morality of lineages, i.e., the constraint that cells do not merge. To find feasible solutions of this NP-hard problem, with certified bounds to the global optimum, we define efficient separation procedures and apply these as part of a branch-and-cut algorithm. To show the effectiveness of this approach, we analyze feasible solutions for real microscopy data in terms of bounds and run-time, and by their weighted edit distance to lineage forests traced by humans.

The goal of this paper is to perform 3D object detection in single monocular images in the domain of autonomous driving. Our method first aims to generate a set of candidate class-specific object proposals, which are then run through a standard CNN pipeline to obtain high-quality object detections. The focus of this paper is on proposal generation. In particular, we propose a probabilistic model that places object candidates in 3D using a prior on ground-plane. We then score each candidate box projected to the image plane via several intuitive potentials such as semantic segmentation, contextual information, size and location priors and typical object shape. The weights in our model are trained with S-SVM. Experiments show that our object proposal generation approach significantly outperforms all monocular baselines, and achieves the best detection performance on the challenging KITTI benchmark, among the published monocular competitors.

Cascaded regression has recently become the method of choice for solving non-linear least squares problems such as deformable image alignment. Given a sizeable training set, cascaded regression learns a set of generic rules that are sequentially applied to minimise the least squares problem. Despite the success of cascaded regression for problems such as face alignment and head pose estimation, there are several shortcomings arising in the strategies proposed thus far. Specifically, (a) the regressors are learnt independently, (b) the descent directions may cancel one another out and (c) handcrafted features (e.g., HoGs, SIFT etc.) are mainly used to drive the cascade, which may be sub-optimal for the task at hand. In this paper, we propose a combined and jointly trained convolutional recurrent neural network architecture that allows the training of an end-to-end to system that attempts to alleviate the aforementioned drawbacks. The recurrent module facilitates the joint optimisation of the regressors by assuming the cascades form a nonlinear dynamical system, in effect fully utilising the information between all cascade levels by introducing a memory unit that shares information across all levels. The convolutional module allows the network to extract features that are specialised for the task at hand and are experimentally shown to outperform hand-crafted features. We show that the application of the proposed architecture for the problem of face alignment results in a strong improvement over the current state-of-the-art.

The probabilistic principal component analysis (PPCA) is built upon a global linear mapping, with which it is insufficient to model complex data variation. This paper proposes a mixture of bilateral-projection probabilistic principal component analysis model (mixB2DPPCA) on 2D data. With multi-components in the mixture, this model can be seen as a `soft' cluster algorithm and has capability of modeling data with complex structures. A Bayesian inference scheme has been proposed based on the variational EM (Expectation-Maximization) approach for learning model parameters. Experiments on some publicly available databases show that the performance of mixB2DPPCA has been largely improved, resulting in more accurate reconstruction errors and recognition rates than the existing PCA-based algorithms.

This paper addresses the problem of mirror surface reconstruction, and a solution based on observing the reflections of a moving reference plane on the mirror surface is proposed. Unlike previous approaches which require tedious work to calibrate the camera, our method can recover both the camera intrinsics and extrinsics together with the mirror surface from reflections of the reference plane under at least three unknown distinct poses. Our previous work has demonstrated that 3D poses of the reference plane can be registered in a common coordinate system using reflection correspondences established across images. This leads to a bunch of registered 3D lines formed from the reflection correspondences. Given these lines, we first derive an analytical solution to recover the camera projection matrix through estimating the line projection matrix. We then optimize the camera projection matrix by minimizing reprojection errors computed based on a cross-ratio formulation. The mirror surface is finally reconstructed based on the optimized cross-ratio constraint. Experimental results on both synthetic and real data are presented, which demonstrate the feasibility and accuracy of our method.

Fine-grained classification involves distinguishing between similar sub-categories based on subtle differences in highly localized regions; therefore, accurate localization of discriminative regions remains a major challenge. We describe a patch-based framework to address this problem. We introduce triplets of patches with geometric constraints to improve the accuracy of patch localization, and automatically mine discriminative geometrically-constrained triplets for classification. The resulting approach only requires object bounding boxes. Its effectiveness is demonstrated using four publicly available fine-grained datasets, on which it outperforms or obtains comparable results to the state-of-the-art in classification.

Recognizing an action from a sequence of 3D skeletal poses is a challenging task. First, different actors may perform the same action in various performing styles. Second, the estimated poses are sometimes inaccurate due to sensory noises. These challenges can cause large variations between instances of the same class. Third, the datasets are usually small, with only a few actors performing few repetitions of each action. Hence training complex classifiers risks over-fitting the data. We address this task by mining a set of key-pose-motifs for each action class. A key-pose-motif contains a set of ordered poses or action units(a short sequence of poses), which are required to be close but not necessarily adjacent in the action sequences. The representation is robust to style variations and outlier poses. The key-pose-motifs are represented in terms of a dictionary using soft-quantization (probabilities) to deal with inaccuracies caused by quantization. We propose an efficient algorithm to mine key-pose-motifs taking into account these probabilities. We classify a sequence by matching it to the motifs of each class and select the class that maximizes the matching score. This simple classifier obtains state-of-the-art performance on two benchmark datasets and outperforms a deep network approach.

In computer vision, many problems can be formulated as finding a low rank approximation of a given measurement matrix. Ideally, if all elements of the measurement matrix are available, this is easily solved in the L2-norm using factorization. However, in practice this is rarely the case. Lately, this problem has been addressed using different approaches, one is to replace the rank term by the convex nuclear norm, another is to derive the convex envelope of the rank term plus a data term. In the latter case, matrices are divided into sub-matrices and the envelope is computed for each sub-block individually. In this paper a new convex envelope is derived which takes all sub-matrices into account simultaneously. This leads to a simpler formulation, using only one parameter, for applications where one seeks low rank approximations of multiple matrices with the same rank. We show in this paper how our general framework can be used for manifold denoising of several images at once, as well as just denoising one image. We get comparable results to other well-known methods and our framework can also be used for other applications such as linear shape models.

Over the past ten years, metric learning allowed the improvement of the numerous machine learning approaches that manipulate distances or similarities. In this field, local metric learning has been shown to be very efficient, especially to take into account non linearities in the data and better capture the peculiarities of the application of interest. However, it is well known that local metric learning (i) can entail overfitting and (ii) face difficulties to compare two instances that are assigned to two different local models. In this paper, we address these two issues by introducing a novel metric learning algorithm that linearly combines local models (C2LM). Starting from a partition of the space in regions and a model (a score function) for each region, C2LM defines a metric between points as a weighted combination of the models. A weight vector is learned for each pair of regions, and a spatial regularization ensures that the weight vectors evolve smoothly and that nearby models are favored in the combination. The proposed approach has the particularity of working in a regression setting, of working implicitly at different scales, and of being generic enough so that it is applicable to similarities and distances. We prove theoretical guarantees of the approach using the framework of algorithmic robustness. We carry out experiments with datasets using both distances (perceptual color distances, using Mahalanobis-like distances) and similarities (semantic word similarities, using bilinear forms), showing that C2LM consistently improves regression accuracy even in the case where the amount of training data is small.

Multi-label submodular Markov Random Fields (MRFs) have been shown to be solvable using max-flow based on an encoding of the labels proposed by Ishikawa, in which each variable X_i is represented by l nodes (where l is the number of labels) arranged in a column. However, this method in general requires 2l^2 edges for each pair of neighbouring variables. This makes it inapplicable to realistic problems with many variables and labels, due to excessive memory requirement. In this paper, we introduce a variant of the max-flow algorithm that requires much less storage. Consequently, our algorithm makes it possible to optimally solve multi-label submodular problems involving large numbers of variables and labels on a standard computer.

We propose a material classification method using raw time-of-flight (ToF) measurements. ToF cameras capture the correlation between a reference signal and the temporal response of material to incident illumination. Such measurements encode unique signatures of the material, i.e. the degree of subsurface scattering inside a volume. Subsequently, it offers an orthogonal domain of feature representation compared to conventional spatial and angular reflectance-based approaches. We demonstrate the effectiveness, robustness, and efficiency of our method through experiments and comparisons of real-world materials.

Superpixels are perceptually meaningful atomic regions that can effectively capture image features. Among various methods for computing uniform superpixels, simple linear iterative clustering (SLIC) is popular due to its simplicity and high performance. In this paper, we extend SLIC to compute content-sensitive superpixels, i.e., small superpixels in content-dense regions (e.g., with high intensity or color variation) and large superpixels in content-sparse regions. Rather than the conventional SLIC method that clusters pixels in R5, we map the image I to a 2-dimensional manifold M in R5, whose area elements are a good measure of the content density in I. We propose an efficient method to compute restricted centroidal Voronoi tessellation (RCVT) --- a uniform tessellation --- on M, which induces the content-sensitive superpixels in I. Unlike other algorithms that characterize content-sensitivity by geodesic distances, manifold SLIC tackles the problem by measuring areas of Voronoi cells on M, which can be computed at a very low cost. As a result, it runs 10 times faster than the state-of-the-art content-sensitive superpixels algorithm. We evaluate manifold SLIC and seven representative methods on the BSDS500 benchmark and observe that our method outperforms the existing methods.

While there has been increasing interest in the task of describing video with natural language, current computer vision algorithms are still severely limited in terms of the variability and complexity of the videos and their associated language that they can recognize. This is in part due to the simplicity of current benchmarks, which mostly focus on specific fine-grained domains with limited videos and simple descriptions. While researchers have provided several benchmark datasets for image captioning, we are not aware of any large-scale video description dataset with comprehensive categories yet diverse video content. In this paper we present MSR-VTT (standing for "ABC-Video to Text") which is a new large-scale video benchmark for video understanding, especially the emerging task of translating video to text. This is achieved by collecting 257 popular queries from a commercial video search engine, with 118 videos for each query. In its current version, MSR-VTT provides 10K web video clips with 38.7 hours and 200K clip-sentence pairs in total, covering the most comprehensive categories and diverse visual content, and representing the largest dataset in terms of sentence and vocabulary. Each clip is annotated with about 20 natural sentences by 1,327 AMT workers. We present a detailed analysis of MSR-VTT in comparison to a complete set of existing datasets, together with a summarization of different state-of-the-art video-to-text approaches. We also provide an extensive evaluation of these approaches on this dataset, showing that the hybrid Recurrent Neural Network-based approach, which combines single-frame and motion representations with soft-attention pooling strategy, yields the best generalization capability on MSR-VTT.