ICCV 2017 Papers

We present a deep-learning framework for real-time multiple spatio-temporal (S/T) action localisation and classification. Current state-of-the-art approaches work offline, and are too slow to be useful in real-world settings. To overcome their limitations we introduce two major developments. Firstly, we adopt real-time SSD (Single Shot MultiBox Detector) CNNs to regress and classify detection boxes in each video frame potentially containing an action of interest. Secondly, we design an original and efficient online algorithm to incrementally construct and label "action tubes" from the SSD frame level detections. As a result, our system is not only capable of performing S/T detection in real time, but can also perform early action prediction in an online fashion. We achieve new state-of-the-art results in both S/T action localisation and early action prediction on the challenging UCF101-24 and J-HMDB-21 benchmarks, even when compared to the top offline competitors. To the best of our knowledge, ours is the first real-time (up to 40fps) system able to perform online S/T action localisation on the untrimmed videos of UCF101-24.

In this paper, we propose a CNN-based framework for online MOT. This framework utilizes the merits of single object trackers in adapting appearance models and searching for target in the next frame. Simply applying single object tracker for MOT will encounter the problem in computational efficiency and drifted results caused by occlusion. Our framework achieves computational efficiency by sharing features and using ROI-Pooling to obtain individual features for each target. Some online learned target-specific CNN layers are used for adapting the appearance model for each target. In the framework, we introduce spatial-temporal attention mechanism (STAM) to handle the drift caused by occlusion and interaction among targets. The visibility map of the target is learned and used for inferring the spatial attention map. The spatial attention map is then applied to weight the features. Besides, the occlusion status can be estimated from the visibility map, which controls the online updating process via weighted loss on training samples with different occlusion statuses in different frames. It can be considered as temporal attention mechanism. The proposed algorithm achieves 34.3% and 46.0% in MOTA on challenging MOT15 and MOT16 benchmark dataset respectively.

While machine learning approaches to image restoration offer great promise, current methods risk training models fixated on performing well only for image corruption of a particular level of difficulty--such as a certain level of noise or blur. First, we examine the weakness of conventional "fixated" models and demonstrate that training general models to handle arbitrary levels of corruption is indeed non-trivial. Then, we propose an on-demand learning algorithm for training image restoration models with deep convolutional neural networks. The main idea is to exploit a feedback mechanism to self-generate training instances where they are needed most, thereby learning models that can generalize across difficulty levels. On four restoration tasks--image inpainting, pixel interpolation, image deblurring, and image denoising--and three diverse datasets, our approach consistently outperforms both the status quo training procedure and curriculum learning alternatives.

The handwritten signature is perhaps the most accustomed way for the acknowledgement of the consent of an individual or the authentication of the identity of a person in numerous transactions. In addition, the authenticity of a questioned offline or static handwritten signature still poses a case of interest, especially in forensic related applications. A common approach in offline signature verification system is to apply several predetermined image analysis models. Consequently, any offline signature sample which originates from either authentic persons or forgers, utilizes a fixed feature extraction base. In this proposed study, the feature space and the corresponding projection values depend on the training samples only; thus the proposed method can be found useful in forensic cases. In order to do so, we reenter a groundbreaking unsupervised learning method named archetypal analysis, which is connected to effective data analysis approaches such as sparse coding. Due to the fact that until recently there was no efficient implementation publicly available, archetypal analysis had only few cases of use. However, a fast optimization scheme using an active set strategy is now available. The main goal of this work is to introduce archetypal analysis for offline signature verification. The output of the archetypal analysis of few reference samples is a set of archetypes which are used to form the base of the feature space. Then, given a set of archetypes and a signature sample under examination archetypal analysis and average pooling provides the corresponding features. The promising performance of the proposed approach is demonstrated with the use of an evaluation method which employs the popular CEDAR and MCYT75 signature datasets.

We present a deep convolutional decoder architecture that can generate volumetric 3D outputs in a compute- and memory-efficient manner by using an octree representation. The network learns to predict both the structure of the octree, and the occupancy values of individual cells. This makes it a particularly valuable technique for generating 3D shapes. In contrast to standard decoders acting on regular voxel grids, the architecture does not have cubic complexity. This allows representing much higher resolution outputs with a limited memory budget. We demonstrate this in several application domains, including 3D convolutional autoencoders, generation of objects and whole scenes from high-level representations, and shape from a single image.

Edge and surface are two fundamental visual elements of an object. The majority of existing object proposal approaches utilize edge or edge-like cues to rank candidates, while we consider that the surface cue containing the 3D characteristic of objects should be captured effectively for proposals, which has been rarely discussed before. In this paper, an object-level proposal model is presented, which constructs an occlusion-based objectness taking the surface cue into account. Specifically, the better detection of occlusion edges is focused on to enrich the surface cue into proposals, namely, the occlusion-dominated fusion and normalization criterion are designed to obtain the approximately overall contour information, to enhance the occlusion edge map at utmost and thus boost proposals. Experimental results on the PASCAL VOC 2007 and MS COCO 2014 dataset demonstrate the effectiveness of our approach, which achieves around 6% improvement on the average recall than Edge Boxes at 1000 proposals and also leads to a modest gain on the performance of object detection.

The recently developed variational autoencoders (VAEs) have proved to be an effective confluence of the rich representational power of neural networks with Bayesian methods. However, most work on VAEs use a rather simple prior over the latent variables such as standard normal distribution, thereby restricting its applications to relatively simple phenomena. In this work, we propose hierarchical nonparametric variational autoencoders, which combines tree-structured Bayesian nonparametric priors with VAEs, to enable infinite flexibility of the latent representation space. Both the neural parameters and Bayesian priors are learned jointly using tailored variational inference. The resulting model induces a hierarchical structure of latent semantic concepts underlying the data corpus, and infers accurate representations of data instances. We apply our model in video representation learning. Our method is able to discover highly interpretable activity hierarchies, and obtain improved clustering accuracy and generalization capacity based on the learned rich representations.

We present an approach for blind image deblurring, which handles non-uniform blurs. Our algorithm has two main components: (i) A new method for recovering the unknown blur-field directly from the blurry image, and (ii) A method for deblurring the image given the recovered nonuniform blur-field. Our blur-field estimation is based on analyzing the spectral content of blurry image patches by Re-blurring them. Being unrestricted by any training data, it can handle a large variety of blur sizes, yielding superior blur-field estimation results compared to training based deep-learning methods. Our non-uniform deblurring algorithm is based on the internal image-specific patch recurrence prior. It attempts to recover a sharp image which, on one hand - results in the blurry image under our estimated blur-field, and on the other hand - maximizes the internal recurrence of patches within and across scales of the recovered sharp image. The combination of these two components gives rise to a blind-deblurring algorithm, which exceeds the performance of state-of-the-art CNN-based blind-deblurring by a significant margin, without the need for any training data.

Part-based trackers are effective in exploiting local details of the target object for robust tracking. In contrast to most existing part-based methods that divide all kinds of target objects into a number of fixed rectangular patches, in this paper, we propose a novel framework in which a set of deformable patches dynamically collaborate on tracking of non-rigid objects. In particular, we proposed a shape-preserved kernelized correlation filter (SP-KCF) which can accommodate target shape information for robust tracking. The SP-KCF is introduced into the level set framework for dynamic tracking of individual patches. In this manner, our proposed deformable patches are target-dependent, have the capability to assume complex topology, and are deformable to adapt to target variations. As these deformable patches properly capture individual target subregions, we exploit their photometric discrimination and shape variation to reveal the trackability of individual target subregions, which enables the proposed tracker to dynamically take advantage of those subregions with good trackability for target likelihood estimation. Finally the shape information of these deformable patches enables accurate object contours to be computed as the tracking output. Experimental results on the latest public sets of challenging sequences demonstrate the effectiveness of the proposed method.

Many state-of-the-art approaches to multi-object tracking rely on detecting them in each frame independently, grouping detections into short but reliable trajectory segments, and then further grouping them into full trajectories. This grouping typically relies on imposing local smoothness constraints but almost never on enforcing more global ones on the trajectories. In this paper, we propose a non-Markovian approach to imposing global consistency by using behavioral patterns to guide the tracking algorithm. When used in conjunction with state-of-the-art tracking algorithms, this further increases their already good performance on multiple challenging datasets. We show significant improvements both in supervised settings where ground truth is available and behavioral patterns can be learned from it, and in completely unsupervised settings.

During the last years, Convolutional Neural Networks (CNNs) have achieved state-of-the-art performance in image classification. Their architectures have largely drawn inspiration by models of the primate visual system. However, while recent research results of neuroscience prove the existence of non-linear operations in the response of complex visual cells, little effort has been devoted to extend the convolution technique to non-linear forms. Typical convolutional layers are linear systems, hence their expressiveness is limited. To overcome this, various non-linearities have been used as activation functions inside CNNs, while also many pooling strategies have been applied. We address the issue of developing a convolution method in the context of a computational model of the visual cortex, exploring quadratic forms through the Volterra kernels. Such forms, constituting a more rich function space, are used as approximations of the response profile of visual cells. Our proposed second-order convolution is tested on CIFAR-10 and CIFAR-100. We show that a network which combines linear and non-linear filters in its convolutional layers, can outperform networks that use standard linear filters with the same architecture, yielding results competitive with the state-of-the-art on these datasets.

This paper considers the problem of recovering either a low rank matrix or a sparse vector from observations of linear combinations of the vector or matrix elements. Recent methods replace the non-convex regularization with l1 or nuclear norm relaxations. It is well known that this approach recovers near optimal solutions if a so called restricted isometry property (RIP) holds. On the other hand it also has a shrinking bias which can degrade the solution. In this paper we study an alternative non-convex regularization term that does not suffer from this bias. Our main theoretical results show that if a RIP holds then the stationary points are often well separated, in the sense that their differences must be of high cardinality/rank. Thus, with a suitable initial solution the approach is unlikely to fall into a bad local minimum. Our numerical tests show that the approach is likely to converge to a better solution than standard l1/nuclear-norm relaxation even when starting from trivial initializations. In many cases our results can also be used to verify global optimality of our method.

Despite the recent success of deep-learning based semantic segmentation, deploying a pre-trained road scene segmenter to a city whose images are not presented in the training set would not achieve satisfactory performance due to dataset biases. Instead of collecting a large number of annotated images of each city of interest to train or refine the segmenter, we propose an unsupervised learning approach to adapt road scene segmenters across different cities. By utilizing Google Street View and its time-machine feature, we can collect unannotated images for each road scene at different times, so that the associated static-object priors can be extracted accordingly. By advancing a joint global and class-specific domain adversarial learning framework, adaptation of pre-trained segmenters to that city can be achieved without the need of any user annotation or interaction. We show that our method improves the performance of semantic segmentation in multiple cities across continents, while it performs favorably against state-of-the-art approaches requiring annotated training data.

We address the problem of distance metric learning (DML), defined as learning a distance consistent with a notion of semantic similarity. Traditionally, for this problem supervision is expressed in the form of sets of points that follow an ordinal relationship -- an anchor point x is similar to a set of positive points Y, and dissimilar to a set of negative points Z, and a loss defined over these distances is minimized. While the specifics of the optimization differ, in this work we collectively call this type of supervision Triplets and all methods that follow this pattern Triplet-Based methods. These methods are challenging to optimize. A main issue is the need for finding informative triplets, which is usually achieved by a variety of tricks such as increasing the batch size, hard or semi-hard triplet mining, etc. Even with these tricks, the convergence rate of such methods is slow. In this paper we propose to optimize the triplet loss on a different space of triplets, consisting of an anchor data point and similar and dissimilar proxy points which are learned as well. These proxies approximate the original data points, so that a triplet loss over the proxies is a tight upper bound of the original loss. This proxy-based loss is empirically better behaved. As a result, the proxy-loss improves on state-of-art results for three standard zero-shot learning datasets, by up to 15% points, while converging three times as fast as other triplet-based losses.

We investigate the problem of person search in the wild in this work. Instead of comparing the query against all candidate regions generated in a query-blind manner, we propose to recursively shrink the search area from the whole image till achieving precise localization of the target person, by fully exploiting information from the query and contextual cues in every recursive search step. We develop the Neural Person Search Machines (NPSM) to implement such recursive localization for person search. Benefiting from its neural search mechanism, NPSM is able to selectively shrink its focus from a loose region to a tighter one containing the target automatically. In this process, NPSM employs an internal primitive memory component to memorize the query representation which modulates the attention and augments its robustness to other distracting regions. Evaluations on two benchmark datasets, CUHK-SYSU Person Search dataset and PRW dataset, have demonstrated that our method can outperform current state-of-the-arts in both mAP and top-1 evaluation protocols.

In this paper, we approach the problem of segmentation-free query-by-string word spotting for handwritten documents. In other words, we use methods inspired from computer vision and machine learning to search for words in large collections of digitized manuscripts. In particular, we are interested in historical handwritten texts, which are often far more challenging than modern printed documents. This task is important, as it provides people with a way to quickly find what they are looking for in large collections that are tedious and difficult to read manually. To this end, we introduce an end-to-end trainable model based on deep neural networks that we call Ctrl-F-Net. Given a full manuscript page, the model simultaneously generates region proposals, and embeds these into a distributed word embedding space, where searches are performed. We evaluate the model on common benchmarks for handwritten word spotting, outperforming the previous state-of-the-art segmentation-free approaches by a large margin, and in some cases even segmentation-based approaches. One interesting real-life application of our approach is to help historians to find and count specific words in court records that are related to women's sustenance activities and division of labor. We provide promising preliminary experiments that validate our method on this task.

In this paper, we propose the first higher frame rate video dataset (called Need for Speed - NfS) and benchmark for visual object tracking. The dataset consists of 100 videos (380K frames) captured with now commonly available higher frame rate (240 FPS) cameras from real world scenarios. All frames are annotated with axis aligned bounding boxes and all sequences are manually labelled with nine visual attributes - such as occlusion, fast motion, background clutter, etc. Our benchmark provides an extensive evaluation of many recent and state-of-the-art trackers on higher frame rate sequences. We ranked each of these trackers according to their tracking accuracy and real-time performance. One of our surprising conclusions is that at higher frame rates, simple trackers such as correlation filters outperform complex methods based on deep networks. This suggests that for practical applications (such as in robotics or embedded vision), one needs to carefully tradeoff bandwidth constraints associated with higher frame rate acquisition, computational costs of real-time analysis, and the required application accuracy. Our dataset and benchmark allows for the first time (to our knowledge) systematic exploration of such issues, and will be made available to allow for further research in this space.

Detecting logo frequency and duration in sports videos provides sponsors an effective way to evaluate their advertising efforts. However, general-purposed object detection methods cannot address all the challenges in sports videos. In this paper, we propose a mutual-enhanced approach that can improve the detection of a logo through the information obtained from other simultaneously occurred logos. In a Fast-RCNN-based framework, we first introduce a homogeneity-enhanced re-ranking method by analyzing the characteristics of homogeneous logos in each frame, including type repetition, color consistency, and mutual exclusion. Different from conventional enhance mechanism that improves the weak proposals with the dominant proposals, our mutual method can also enhance the relatively significant proposals with weak proposals. Mutual enhancement is also included in our frame propagation mechanism that improves logo detection by utilizing the continuity of logos across frames. We use a tennis video dataset and an associated logo collection for detection evaluation. Experiments show that the proposed method outperforms existing methods with a higher accuracy.

In this work, we address multimodal learning problem with Gaussian process latent variable models (GPLVMs) and their application to cross-modal retrieval. Existing GPLVM based studies generally impose individual priors over the model parameters and ignore the intrinsic relations among these parameters. Considering the strong complementarity between modalities, we propose a novel joint prior over the parameters for multimodal GPLVMs to propagate multimodal information in both kernel hyperparameter spaces and latent space. The joint prior is formulated as a harmonization constraint on the model parameters, which enforces the agreement among the modality-specific GP kernels and the similarity in the latent space. We incorporate the harmonization mechanism into the learning process of multimodal GPLVMs. The proposed methods are evaluated on three widely used multimodal datasets for cross-modal retrieval. Experimental results show that the harmonization mechanism is beneficial to the GPLVM algorithms for learning non-linear correlation among heterogeneous modalities.