ICCV 2015 Papers

While feedforward deep convolutional neural networks (CNNs) have been a great success in computer vision, it is important to remember that the human visual contex contains generally more feedback connections than foward connections. In this paper, we will briefly introduce the background of feedbacks in the human visual cortex, which motivates us to develop a computational feedback mechanism in the deep neural networks. The proposed networks perform inference from image features in a bottom-up manner as traditional convolutional networks; while during feedback loops it sets up high-level semantic labels as the agoala to infer the activation status of hidden layer neurons. The feedback networks help us better visualize and understand on how deep neural networks work as well as capture visual attention on expected objects, even in the images with cluttered background and multiple objects.

Image-based localization is an essential complement to GPS localization. Current image-based localization methods are based on either 2D-to-3D or 3D-to-2D to find the correspondences, which ignore the real scene geometric attributes. The main contribution of our paper is that we use a 3D model reconstructed by a short video as the query to realize 3D-to-3D localization under a multi-task point retrieval framework. Firstly, the use of a 3D model as the query enables us to efficiently select location candidates. Furthermore, the reconstruction of 3D model exploits the correlations among different images, based on the fact that images captured from different views for SfM share information through matching features. By exploring shared information (matching features) across multiple related tasks (images of the same scene captured from different views), the visual feature's view-invariance property can be improved in order to get to a higher point retrieval accuracy. More specifically, we use multi-task point retrieval framework to explore the relationship between descriptors and the 3D points, which extracts the discriminant points for more accurate 3D-to-3D correspondences retrieval. We further apply multi-task learning (MTL) retrieval approach on thermal images to prove that our MTL retrieval framework also provides superior performance for the thermal domain. This application is exceptionally helpful to cope with the localization problem in an environment with limited light sources.

Subspace models have been widely used for appearance based object tracking. Most existing subspace based trackers employ a linear subspace to represent object appearances, which are not accurate enough to model large variations of objects. To address this, this paper presents a local subspace collaborative tracking method for robust visual tracking, where multiple linear and nonlinear subspaces are learned to better model the nonlinear relationship of object appearances. First, we retain a set of key samples and compute a set of local subspaces for each key sample. Then, we construct a hyper sphere to represent the local nonlinear subspace for each key sample. The hyper sphere of one key sample passes the local key samples and also is tangent to the local linear subspace of the specific key sample. In this way, we are able to represent the nonlinear distribution of the key samples and also approximate the local linear subspace near the specific key sample, so that local distributions of the samples can be represented more accurately. Experimental results on challenging video sequences demonstrate the effectiveness of our method.

Patch-level descriptors underlie several important computer vision tasks, such as stereo-matching or content-based image retrieval. We introduce a deep convolutional architecture that yields patch-level descriptors, as an alternative to the popular SIFT descriptor for image retrieval. The proposed family of descriptors, called Patch-CKN, adapt the recently introduced Convolutional Kernel Network (CKN), an unsupervised framework to learn convolutional architectures. We present a comparison framework to benchmark current deep convolutional approaches along with Patch-CKN for both patch and image retrieval, including our novel ``RomePatches'' dataset. Patch-CKN descriptors yield competitive results compared to supervised CNN alternatives on patch and image retrieval.

The task of counting the number of repetitions of approximately the same action in an input video sequence is addressed. The proposed method runs online and not on the complete pre-captured video. It analyzes sequentially blocks of 20 non-consecutive frames. The cycle length within each block is evaluated using a convolutional neural network and the information is then integrated over time. The entropy of the network's predictions is used in order to automatically start and stop the repetition counter and to select the appropriate time scale. Coupled with a region of interest detection mechanism, the method is robust enough to handle real world videos, even when the camera is moving. A unique property of our method is that it is shown to successfully train on entirely unrealistic data created by synthesizing moving random patches.

This paper presents a novel feature learning method for visual speech recognition using Deep Boltzmann Machines (DBM). Unlike all existing visual feature extraction techniques which solely extracts features from video sequences, our method is able to explore both acoustic information and visual information to learn a better visual feature representation in the training stage. During the test stage, instead of using both audio and visual signals, only the videos are used for generating the missing audio feature, and both the given visual and given audio features are used to obtain a joint representation. We carried out our experiments on a large scale audio-visual data corpus, and experimental results show that our proposed techniques outperforms the performance of the hadncrafted features and features learned by other commonly used deep learning techniques.

Due to unavoidable appearance variations caused by occlusion, deformation, and other factors, classifiers for visual tracking are nonlinear as a necessity. Building on the theory of globally linear approximations to nonlinear functions, we introduce an elegant method that jointly learns a nonlinear classifier and a visual dictionary for tracking objects in a semi-supervised sparse coding fashion. This establishes an obvious distinction from conventional sparse coding based discriminative tracking algorithms that usually maintain two-stage learning strategies, i.e., learning a dictionary in an unsupervised way then followed by training a classifier. However, the treating dictionary learning and classifier training as separate stages may not produce both descriptive and discriminative models for objects. By contrast, our method is capable of constructing a dictionary that not only fully reflects the intrinsic manifold structure of the data, but also possesses discriminative power. This paper presents an optimization method to obtain such an optimal dictionary, associated sparse coding, and a classifier in an iterative process. Our experiments on a benchmark show our tracker attains outstanding performance compared with the state-of-the-art algorithms.

Facial landmark detection, as a vital topic in computer vision, has been studied for many decades and lots of datasets have been collected for evaluation. These datasets usually have different annotations, e.g., 68-landmark markup for LFPW dataset, while 74-landmark markup for GTAV dataset. Intuitively, it is meaningful to fuse all the datasets to predict a union of all types of landmarks from multiple datasets (i.e., transfer the annotations of each dataset to all other datasets), but this problem is nontrivial due to the distribution discrepancy between datasets and incomplete annotations of all types for each dataset. In this work, we propose a deep regression network coupled with sparse shape regression (DRN-SSR) to predict the union of all types of landmarks by leveraging datasets with varying annotations, each dataset with one type of annotation. Specifically, the deep regression network intends to predict the union of all landmarks, and the sparse shape regression attempts to approximate those undefined landmarks on each dataset so as to guide the learning of the deep regression network for face alignment. Extensive experiments on two challenging datasets, IBUG and GLF, demonstrate that our method can effectively leverage the multiple datasets with different annotations to predict the union of all types of landmarks.

Hands appear very often in egocentric video, and their appearance and pose give important cues about what people are doing and what they are paying attention to. But existing work in hand detection has made strong assumptions that work well in only simple scenarios, such as with limited interaction with other people or in lab settings. We develop methods to locate and distinguish between hands in egocentric video using strong appearance models with Convolutional Neural Networks, and introduce a simple candidate region generation approach that outperforms existing techniques at a fraction of the computational cost. We show how these high-quality bounding boxes can be used to create accurate pixelwise hand regions, and as an application, we investigate the extent to which hand segmentation alone can distinguish between different activities. We evaluate these techniques on a new dataset of 48 first-person videos (along with pixel-level ground truth for over 15,000 hand instances) of people interacting in realistic environments.

In this article we explore the problem of constructing person-specific models for the detection of facial Action Units (AUs), addressing the problem from the point of view of Transfer Learning and Multi-Task Learning. Our starting point is the fact that some expressions, such as smiles, are very easily elicited, annotated, and automatically detected, while others are much harder to elicit and to annotate. We thus consider a novel problem: all AU models for the target subject are to be learnt using person-specific annotated data for a reference AU (AU12 in our case), and no data or little data regarding the target AU. In order to design such a model, we propose a novel Multi-Task Learning and the associated Transfer Learning framework, in which we consider both relations across subjects and AUs. That is to say, we consider a tensor structure among the tasks. Our approach hinges on learning the latent relations among tasks using one single reference AU, and then transferring these latent relations to other AUs. We show that we are able to effectively make use of the annotated data for AU12 when learning other person-specific AU models, even in the absence of data for the target task. Finally, we show the excellent performance of our method when small amounts of annotated data for the target tasks are made available.

Online Multi-Object Tracking (MOT) has wide applications in time-critical video analysis scenarios, such as robot navigation and autonomous driving. In tracking-by-detection, a major challenge of online MOT is how to robustly associate noisy object detections on a new video frame with previously tracked objects. In this work, we formulate the online MOT problem as decision making in Markov Decision Processes (MDPs), where the lifetime of an object is modeled with a MDP. Learning a similarity function for data association is equivalent to learning a policy for the MDP, and the policy learning is approached in a reinforcement learning fashion which benefits from both advantages of offline-learning and online-learning for data association. Moreover, our framework can naturally handle the birth/death and appearance/disappearance of targets by treating them as state transitions in the MDP while leveraging existing online single object tracking methods. We conduct experiments on the MOT Benchmark to verify the effectiveness of our method.

We propose an effective approach for spatio-temporal action localization in realistic videos. The approach first detects proposals at the frame-level and scores them with a combination of static and motion CNN features. It then tracks high-scoring proposals throughout the video using a tracking-by-detection approach. Our tracker relies simultaneously on instance-level and class-level detectors. The tracks are scored using a spatio-temporal motion histogram, a descriptor at the track level, in combination with the CNN features. Finally, we perform temporal localization of the action using a sliding-window approach at the track level. We present experimental results for spatio-temporal localization on the UCF-Sports, J-HMDB and UCF-101 action localization datasets, where our approach outperforms the state of the art with a margin of 15%, 7% and 12% respectively in mAP.

The current dominant paradigm for feature learning in computer vision relies on training neural networks for the task of object recognition using millions of hand labelled images. Is it also possible to learn features for a diverse set of visual tasks using any other form of supervision? In biology, living organisms developed the ability of visual perception for the purpose of moving and acting in the world. Drawing inspiration from this observation, in this work we investigated if the awareness of egomotion(i.e. self motion) can be used as a supervisory signal for feature learning. As opposed to the knowledge of class labels, information about egomotion is freely available to mobile agents. We found that using the same number of training images, features learnt using egomotion as supervision compare favourably to features learnt using class-label as supervision on the tasks of scene recognition, object recognition, visual odometry and keypoint matching.

We present a supervised learning to rank algorithm that effectively orders images by exploiting the structure in image sequences. Most often in the supervised learning to rank literature, ranking is approached either by analysing pairs of images or by optimizing a list-wise surrogate loss function on full sequences. In this work we propose MidRank, which learns from moderately sized sub-sequences instead. These sub-sequences contain useful structural ranking information that leads to better learnability during training and better generalization during testing. By exploiting sub-sequences, the proposed MidRank improves ranking accuracy considerably on an extensive array of image ranking applications and datasets.

This paper proposes a novel method, which learns to detect saliency of face images. To be more specific, we obtain a database of eye tracking over extensive face images, via conducting an eye tracking experiment. With analysis on eye tracking database, we verify that the fixations tend to cluster around facial features, when viewing images with large faces. For modeling attention on faces and facial features, the proposed method learns the Gaussian mixture model (GMM) distribution from the fixations of eye tracking data as the top-down features for saliency detection of face images. Then, in our method, the top-down features (i.e., face and facial features) upon the the learnt GMM are linearly combined with the conventional bottom-up features (i.e., color, intensity, and orientation), for saliency detection. In the linear combination, we argue that the weights corresponding to top-down feature channels depend on the face size in images, and the relationship between the weights and face size is thus investigated via learning from the training eye tracking data. Finally, experimental results show that our learning-based method is able to advance state-of-the-art saliency prediction for face images. The corresponding database and code are available online: www.ee.buaa.edu.cn/xumfiles/saliency_detection.html.

Online Multiple Target Tracking (MTT) is often addressed within the tracking-by-detection paradigm. Detections are previously extracted independently in each frame and then objects trajectories are built by maximizing specifically designed coherence functions. Nevertheless, ambiguities arise in presence of occlusions or detection errors. In this paper we claim that the ambiguities in tracking could be solved by a selective use of the features, by working with more reliable features if possible and exploiting a deeper representation of the target only if necessary. To this end, we propose an online divide and conquer tracker for static camera scenes, which partitions the assignment problem in local subproblems and solves them by selectively choosing and combining the best features. The complete framework is cast as a structural learning task that unifies these phases and learns tracker parameters from examples. Experiments on two different datasets highlights a significant improvement of tracking performances (MOTA +10%) over the state of the art.

In recent years, region proposals have replaced sliding windows in support of object recognition, offering more discriminating shape and appearance information through improved localization. One powerful approach for generating region proposals is based on minimizing parametric energy functions with parametric maxflow. In this paper, we introduce Parametric Min-Loss (PML), a novel structured learning framework for parametric energy functions. While PML is generally applicable to different domains, we use it in the context of region proposals to learn to combine a set of mid-level grouping cues to yield a small set of object region proposals with high recall. Our learning framework accounts for multiple diverse outputs, and is complemented by diversification seeds based on image location and color. This approach casts perceptual grouping and cue combination in a novel structured learning framework which yields baseline improvements on VOC 2012 and COCO 2014.

The challenging problem of filamentary structure segmentation has a broad range of applications in biological and medical fields. A critical yet challenging issue remains on how to detect and restore the small filamentary fragments from backgrounds: The small fragments are of diverse shapes and appearances, meanwhile the backgrounds could be cluttered and ambiguous. Focusing on this issue, this paper proposes an iterative two-step learning-based approach to boost the performance based on a base segmenter arbitrarily chosen from a number of existing segmenters: We start with an initial partial segmentation where the filamentary structure obtained is of high confidence based on this existing segmenter. We also define a scanning horizon as epsilon balls centred around the partial segmentation result. Step one of our approach centers on a data-driven latent classification tree model to detect the filamentary fragments. This model is learned via a training process, where a large number of distinct local figure/background separation scenarios are established and geometrically organized into a tree structure. Step two spatially restores the isolated fragments back to the current partial segmentation, which is accomplished by means of completion fields and matting. Both steps are then alternated with the growth of partial segmentation result, until the input image space is entirely explored. Our approach is rather generic and can be easily augmented to a wide range of existing supervised/unsupervised segmenters to produce an improved result. This has been empirically verified on specific filamentary structure segmentation tasks: retinal blood vessel segmentation as well as neuronal segmentations, where noticeable improvement has been shown over the original state-of-the-arts.

What makes an image appear realistic? In this work, we are answering this question from a data-driven perspective by learning the perception of visual realism directly from large amounts of data. In particular, we train a Convolutional Neural Network (CNN) model that distinguishes natural photographs from automatically generated composite images. The model learns to predict visual realism of a scene in terms of color, lighting and texture compatibility, without any human annotations pertaining to it. Our model outperforms previous works that rely on hand-crafted heuristics, for the task of classifying realistic vs. unrealistic photos. Furthermore, we apply our learned model to compute optimal parameters of a compositing method, to maximize the visual realism score predicted by our CNN model. We demonstrate its advantage against existing methods via a human perception study.