CVPR 2016 论文列表

While current approaches to action recognition on pre-segmented video clips already achieve high accuracies, temporal action detection is still far from comparably good results. Automatically locating and classifying the relevant action segments in videos of varying lengths proves to be a challenging task. We propose a novel method for temporal action detection including statistical length and language modeling to represent temporal and contextual structure. Our approach aims at globally optimizing the joint probability of three components, a length and language model and a discriminative action model, without making intermediate decisions. The problem of finding the most likely action sequence and the corresponding segment boundaries in an exponentially large search space is addressed by dynamic programming. We provide an extensive evaluation of each model component on Thumos 14, a large action detection dataset, and report state-of-the-art results on three datasets.

With the recent popularity of animated GIFs on social media, there is need for ways to index them with rich metadata. To advance research on animated GIF understanding, we collected a new dataset, Tumblr GIF (TGIF), with 100K animated GIFs from Tumblr and 120K natural language descriptions obtained via crowdsourcing. The motivation for this work is to develop a testbed for image sequence description systems, where the task is to generate natural language descriptions for animated GIFs or video clips. To ensure a high quality dataset, we developed a series of novel quality controls to validate free-form text input from crowdworkers. We show that there is unambiguous association between visual content and natural language descriptions in our dataset, making it an ideal benchmark for the visual content captioning task. We perform extensive statistical analyses to compare our dataset to existing image and video description datasets. Next, we provide baseline results on the animated GIF description task, using three representative techniques: nearest neighbor, statistical machine translation, and recurrent neural networks. Finally, we show that models fine-tuned from our animated GIF description dataset can be helpful for automatic movie description.

In this paper we introduce a new method for text detection in natural images. The method comprises two contributions: First, a fast and scalable engine to generate synthetic images of text in clutter. This engine overlays synthetic text to existing background images in a natural way, accounting for the local 3D scene geometry. Second, we use the synthetic images to train a Fully-Convolutional Regression Network (FCRN) which efficiently performs text detection and bounding-box regression at all locations and multiple scales in an image. We discuss the relation of FCRN to the recently-introduced YOLO detector, as well as other end-to-end object detection systems based on deep learning. The resulting detection network significantly out performs current methods for text detection in natural images, achieving an F-measure of 84.2% on the standard ICDAR 2013 benchmark. Furthermore, it can process 15 images per second on a GPU.

In this paper, we propose a structured feature learning framework to reason the correlation among body joints at the feature level in human pose estimation. Different from existing approaches of modeling structures on score maps or predicted labels, feature maps preserve substantially richer descriptions of body joints. The relationships between feature maps of joints are captured with the introduced geometrical transform kernels, which can be easily implemented with a convolution layer. Features and their relationships are jointly learned in an end-to-end learning system. A bi-directional tree structured model is proposed, so that the feature channels at a body joint can well receive information from other joints. The proposed framework improves feature learning substantially. With very simple post processing, it reaches the best mean PCP on the LSP and FLIC datasets. Compared with the baseline of learning features at each joint separately with ConvNet, the mean PCP has been improved by 18% on FLIC. The code is released to the public.

This paper addresses the challenge of 3D full-body human pose estimation from a monocular image sequence. Here, two cases are considered: (i) the image locations of the human joints are provided and (ii) the image locations of joints are unknown. In the former case, a novel approach is introduced that integrates a sparsity-driven 3D geometric prior and temporal smoothness. In the latter case, the former case is extended by treating the image locations of the joints as latent variables in order to take into account considerable uncertainties in 2D joint locations. A deep fully convolutional network is trained to predict the uncertainty maps of the 2D joint locations. The 3D pose estimates are realized via an Expectation-Maximization algorithm over the entire sequence, where it is shown that the 2D joint location uncertainties can be conveniently marginalized out during inference. Empirical evaluation on the Human3.6M dataset shows that the proposed approaches achieve greater 3D pose estimation accuracy over state-of-the-art baselines. Further, the proposed approach outperforms a publicly available 2D pose estimation baseline on the challenging PennAction dataset.

Super-symmetric tensors - a higher-order extension of scatter matrices - are becoming increasingly popular in machine learning and computer vision for modeling data statistics, co-occurrences, or even as visual descriptors. They were shown recently to outperform second-order approaches, however, the size of these tensors are exponential in the data dimensionality, which is a significant concern. In this paper, we study third-order super-symmetric tensor descriptors in the context of dictionary learning and sparse coding. For this purpose, we propose a novel non-linear third-order texture descriptor. Our goal is to approximate these tensors as sparse conic combinations of atoms from a learned dictionary. Apart from the significant benefits to tensor compression that this framework offers, our experiments demonstrate that the sparse coefficients produced by this scheme lead to better aggregation of high-dimensional data and showcase superior performance on two common computer vision tasks compared to the state of the art.

For people first impressions of someone are of determining importance. They are hard to alter through further information. This begs the question if a computer can reach the same judgement. Earlier research has already pointed out that age, gender, and average attractiveness can be estimated with reasonable precision. We improve the state-of-the-art, but also predict - based on someone's known preferences - how much that particular person is attracted to a novel face. Our computational pipeline comprises a face detector, convolutional neural networks for the extraction of deep features, standard support vector regression for gender, age and facial beauty, and - as the main novelties - visual regularized collaborative filtering to infer inter-person preferences as well as a novel regression technique for handling visual queries without rating history. We validate the method using a very large dataset from a dating site as well as images from celebrities. Our experiments yield convincing results, i.e. we predict 76% of the ratings correctly solely based on an image, and reveal some sociologically relevant conclusions. We also validate our collaborative filtering solution on the standard MovieLens rating dataset, augmented with movie posters, to predict an individual's movie rating. We demonstrate our algorithms on howhot.io which went viral around the Internet with more than 50 million pictures evaluated in the first month.

Humans navigate complex crowded environments based on social conventions: they respect personal space, yielding right-of-way and avoid collisions. In our work, we propose a data-driven approach to learn these human-human interactions for predicting their future trajectories. This is in contrast to traditional approaches which use hand-crafted functions such as Social forces. We present a new Long Short-Term Memory (LSTM) model which jointly reasons across multiple individuals in a scene. Different from the conventional LSTM, we share the information between multiple LSTMs through a new pooling layer. This layer pools the hidden representation from LSTMs corresponding to neighboring trajectories to capture interactions within this neighborhood. We demonstrate the performance of our method on several public datasets. Our model outperforms previous forecasting methods by more than 42% . We also analyze the trajectories predicted by our model to demonstrate social behaviours such as collision avoidance and group movement, learned by our model.

How can unlabeled video augment visual learning? Existing methods perform "slow" feature analysis, encouraging temporal coherence, where the image representations of temporally close frames to exhibit only small differences. While this standard approach captures the fact that high-level visual signals change slowly over time, it fails to capture *how* the visual content changes. We propose to generalize slow feature analysis to "steady" feature analysis. The key idea is to impose a prior that higher order derivatives in the learned feature space must be small. To this end, we train a convolutional neural network with a regularizer that minimizes a contrastive loss on tuples of sequential frames from unlabeled video. Focusing on the case of triplets of frames, the proposed method encourages that feature changes over time should be smooth, i.e., similar to the most recent changes. Using five diverse image and video datasets, including unlabeled YouTube and KITTI videos, we demonstrate our method's impact on object recognition, scene classification, and action recognition tasks. We further show that our features learned from unlabeled video can even surpass a standard heavily supervised pretraining approach.

Learning and capturing both appearance and dynamic representations are pivotal for crowd video understanding. Convolutional Neural Networks (CNNs) have shown its remarkable potential in learning appearance representations from images. However, the learning of dynamic representation, and how it can be effectively combined with appearance features for video analysis, remains an open problem. In this study, we propose a novel spatio-temporal CNN, named Slicing CNN (S-CNN), based on the decomposition of 3D feature maps into 2D spatio- and 2D temporal-slices representations. The decomposition brings unique advantages: (1) the model is capable of capturing dynamics of different semantic units such as groups and objects, (2) it learns separated appearance and dynamic representations while keeping proper interactions between them, and (3) it exploits the selectiveness of spatial filters to discard irrelevant background clutter for crowd understanding. We demonstrate the effectiveness of the proposed S-CNN model on the WWW crowd video dataset for attribute recognition and observe significant performance improvements to the state-of-the-art methods (62.55% from 51.84% [21]).

Near-Infrared (NIR) images of most materials exhibit less texture or albedo variations making them beneficial for vision tasks such as intrinsic image decomposition and structured light depth estimation. Understanding the reflectance properties (BRDF) of materials in the NIR wavelength range can be further useful for many photometric methods including shape from shading and inverse rendering. However, even with less albedo variation, many materials e.g. fabrics, leaves, etc. exhibit complex fine-scale surface detail making it hard to accurately estimate BRDF. In this paper, we present an approach to simultaneously estimate NIR BRDF and fine-scale surface details by imaging materials under different IR lighting and viewing directions. This is achieved by an iterative scheme that alternately estimates surface detail and NIR BRDF of materials. Our setup does not require complicated gantries or calibration and we present the first NIR dataset of 100 materials including a variety of fabrics (knits, weaves, cotton, satin, leather), and organic (skin, leaves, jute, trunk, fur) and inorganic materials (plastic, concrete, carpet). The NIR BRDFs measured from material samples are used with a shape-from-shading algorithm to demonstrate fine-scale reconstruction of objects from a single NIR image.

Estimating the number of clusters remains a difficult model selection problem. We consider this problem in the domain where the affinity relations involve groups of more than two nodes. Building on the previous formulation for the pairwise affinity case, we exploit the mathematical structures in the higher order case. We express the original minimal-rank and positive semi-definite (PSD) constraints in a form amenable for numerical implementation, as the original constraints are either intractable or even undefined in general in the higher order case. To scale to large problem sizes, we also propose an alternative formulation, so that it can be efficiently solved via stochastic optimization in an online fashion. We evaluate our algorithm with different applications to demonstrate its superiority, and show it can adapt to varying levels of unbalancedness of clusters.

Semantic object parsing is a fundamental task for understanding objects in detail in computer vision community, where incorporating multi-level contextual information is critical for achieving such fine-grained pixel-level recognition. Prior methods often leverage the contextual information through post-processing predicted confidence maps. In this work, we propose a novel deep Local-Global Long Short-Term Memory (LG-LSTM) architecture to seamlessly incorporate short-distance and long-distance spatial dependencies into the feature learning over all pixel positions. In each LG-LSTM layer, local guidance from neighboring positions and global guidance from the whole image are imposed on each position to better exploit complex local and global contextual information. Individual LSTMs for distinct spatial dimensions are also utilized to intrinsically capture various spatial layouts of semantic parts in the images, yielding distinct hidden and memory cells of each position for each dimension. In our parsing approach, several LG-LSTM layers are stacked and appended to the intermediate convolutional layers to directly enhance visual features, allowing network parameters to be learned in an end-to-end way. The long chains of sequential computation by stacked LG-LSTM layers also enable each pixel to sense a much larger region for inference benefiting from the memorization of previous dependencies in all positions along all dimensions. Comprehensive evaluations on three public datasets well demonstrate the significant superiority of our LG-LSTM over other state-of-the-art methods for object parsing.

Pedestrian detection and semantic segmentation are high potential tasks for many real-time applications. However most of the top performing approaches provide state of art results at high computational costs. In this work we propose a fast solution for achieving state of art results for both pedestrian detection and semantic segmentation. As baseline for pedestrian detection we use sliding windows over cost efficient multiresolution filtered LUV+HOG channels. We use the same channels for classifying pixels into eight semantic classes. Using short range and long range multiresolution channel features we achieve more robust segmentation results compared to traditional codebook based approaches at much lower computational costs. The resulting segmentations are used as additional semantic channels in order to achieve a more powerful pedestrian detector. To also achieve fast pedestrian detection we employ a multiscale detection scheme based on a single flexible pedestrian model and a single image scale. The proposed solution provides competitive results on both pedestrian detection and semantic segmentation benchmarks at 8 FPS on CPU and at 15 FPS on GPU, being the fastest top performing approach.

We propose an approach for dense semantic 3D reconstruction which uses a data term that is defined as potentials over viewing rays, combined with continuous surface area penalization. Our formulation is a convex relaxation which we augment with a crucial non-convex constraint that ensures exact handling of visibility. To tackle the non-convex minimization problem, we propose a majorize-minimize type strategy which converges to a critical point. We demonstrate the benefits of using the non-convex constraint experimentally. For the geometry-only case, we set a new state of the art on two datasets of the commonly used Middlebury multi-view stereo benchmark. Moreover, our general-purpose formulation directly reconstructs thin objects, which are usually treated with specialized algorithms. A qualitative evaluation on the dense semantic 3D reconstruction task shows that we improve significantly over previous methods.

Feature tracking is a fundamental problem in computer vision with applications in various tasks including 3D reconstruction and visual SLAM. While many methods have been devoted to making these tasks robust to noise and outliers, less attention has been attracted to improving the feature tracking itself. This paper introduces a novel multi-body feature tracker that takes advantage of the multi-body rigidity assumption to improve tracking robustness. A conventional approach to addressing this problem would consist of alternating between solving two subtasks: motion segmentation and feature tracking under rigidity constraints for each segment. This approach, however, requires knowing the number of motions, as well as assigning points to motion groups, which is typically sensitive to the motion estimates. By contrast, here, we introduce a segmentation-free solution to multi-body feature tracking that bypasses the motion assignment step and reduces to solving a series of subproblems with closed-form solutions. Our experiments demonstrate the benefits of our approach in terms of tracking accuracy and robustness to noise.

We present an approach for detecting and matching building facades between aerial view and street-view images. We exploit the regularity of urban scene facades as captured by their lattice structures and deduced from median-tiles' shape context, color, texture and spatial similarities. Our experimental results demonstrate effective matching of oblique and partially-occluded facades between aerial and ground views. Quantitative comparisons for automated urban scene facade matching from three cities show superior performance of our method over baseline SIFT, Root-SIFT and the more sophisticated Scale-Selective Self-Similarity and Binary Coherent Edge descriptors. We also illustrate regularity-based applications of occlusion removal from street views and higher-resolution texture-replacement in aerial views.

Deep neural networks with alternating convolutional, max-pooling and decimation layers are widely used in state of the art architectures for computer vision. Max-pooling purposefully discards precise spatial information in order to create features that are more robust, and typically organized as lower resolution spatial feature maps. On some tasks, such as whole-image classification, max-pooling derived features are well suited; however, for tasks requiring precise localization, such as pixel level prediction and segmentation, max-pooling destroys exactly the information required to perform well. Precise localization may be preserved by shallow convnets without pooling but at the expense of robustness. Can we have our max-pooled multi-layered cake and eat it too? Several papers have proposed summation and concatenation based methods for combining upsampled coarse, abstract features with finer features to produce robust pixel level predictions. Here we introduce another model --- dubbed Recombinator Networks --- where coarse features inform finer features early in their formation such that finer features can make use of several layers of computation in deciding how to use coarse features. The model is trained once, end-to-end and performs better than summation-based architectures, reducing the error from the previous state of the art on two facial keypoint datasets, AFW and AFLW, by 30% and beating the current state-of-the-art on 300W without using extra data. We improve performance even further by adding a denoising prediction model based on a novel convnet formulation.

We present a novel dataset and a novel algorithm for recognizing activities of daily living (ADL) from a first-person wearable camera. Handled objects are crucially important for egocentric ADL recognition. For specific examination of objects related to users' actions separately from other objects in an environment, many previous works have addressed the detection of handled objects in images captured from head-mounted and chest-mounted cameras. Nevertheless, detecting handled objects is not always easy because they tend to appear small in images. They can be occluded by a user's body. As described herein, we mount a camera on a user's wrist. A wrist-mounted camera can capture handled objects at a large scale, and thus it enables us to skip the object detection process. To compare a wrist-mounted camera and a head-mounted camera, we also developed a novel and publicly available dataset that includes videos and annotations of daily activities captured simultaneously by both cameras. Additionally, we propose a discriminative video representation that retains spatial and temporal information after encoding the frame descriptors extracted by convolutional neural networks.