Search

Facial expression recognition (FER) is of great interest to the current studies of human-computer interaction. In this paper, we propose a novel geometry-guided facial expression recognition framework, based on graph convolutional networks and transformers, to perform effective emotion recognition from videos. Specifically, we detect and utilize facial landmarks to construct a spatial-temporal graph, based on both the landmark coordinates and local appearance, for representing a facial expression sequence. The graph convolutional blocks and transformer modules are employed to produce high-semantic emotion-related representations from the structured facial graphs, which facilitate the framework to establish both the local and non-local dependency between the vertices. Moreover, spatial and temporal attention mechanisms are introduced into graph-based learning to promote FER reasoning, via the emphasis on the most informative facial components and frames. Extensive experiments demonstrate that the proposed framework achieves promising performance for geometry-based FER and shows great generalization and robustness in real-world applications. [ABSTRACT FROM AUTHOR]

Learning discriminative representations with good robustness from facial observations serves as a fundamental step towards intelligent facial expression recognition (FER). In this article, we propose a novel geometry-aware FER framework to boost the FER performance based on both the geometric and appearance knowledge. Specifically, we propose an encoding strategy for facial landmarks, and adopt a graph convolutional network (GCN) to fully explore the structural information of the facial components behind different expressions. A convolutional neural network (CNN) is further applied to the whole facial observation to learn the global characteristics of different expressions. The features from these two networks are fused into a comprehensive high-semantic representation, which promotes the FER reasoning from both visual and structural perspectives. Moreover, to facilitate the networks to concentrate on the most informative facial regions and components, we introduce multi-level attention mechanisms into the proposed framework, which enhance the reliability of the learned representations for effective FER. Experiments on two challenging FER benchmarks demonstrate that the attentive graph-based learning on the facial geometry boosts the FER accuracy. Furthermore, the insensitivity of the geometric information to the appearance variations also improves the generalization of the proposed framework. [ABSTRACT FROM AUTHOR]

The stable continuation of trajectories among different targets has always been the key to the tracking performance of multi-object tracking (MOT) tasks. If features of the target are aggregated and classified simply, the discriminant features of the target will be ignored. This will affect the robustness of the trajectory generated by the model. Meanwhile, many popular models are keen to execute detection and feature extraction tasks in parallel. But these two tasks will conflict with each other when optimized respectively. Therefore, we propose our tracker MOTFR to solve the above problems. In this paper, we propose a Locally Shared Information Decoupling Module (LSIDM) to reduce task optimization conflicts while ensuring the necessary information sharing. Meanwhile, a feature recoding module for deep extraction of identity discriminative features is proposed, which is called the Feature Purification Module (FPM). By combining LSIDM and FPM modules, the model utilizes the discriminative appearance features to guide the optimization of detection and further improves the performance of our model. To solve the problem of targets disappearing due to various abnormal occlusion, a Short-term Trajectory Online Complement Strategy (STOCS) is proposed to realize the trajectories continuation of these targets in the tracking stage. Through sufficient experiments, we demonstrate the superior performance of our MOTFR, which guarantees high-quality detection while achieving the stability of the target trajectory. [ABSTRACT FROM AUTHOR]

Unsupervised domain adaptation (UDA) person re-identification (ReID) faces enormous challenges due to the severe shift between the source and target domains, as well as the dramatic variations within the target domain. In this paper, to address these issues, we propose a multi-loss gap minimization learning (MGML) approach for UDA person ReID. Firstly, we introduce the part model to learn discriminative patch features and design a Patch-based Part Ignoring (PPI) loss to select reliable instances for the efficient learning of the part model. Then, given the gap that typically occurs because of the inter-domain shift and intra-domain variations, a Gap-based Minimum Camera Discrepancy (G-MCD) loss is proposed. Specifically, in terms of the inter-domain, we propose to leverage the tracklet and camera information to label each distance vector, and accordingly align pair-wise distance distributions to bridge the inter-domain gap. As for the intra-domain, to alleviate the biased search, we propose to perform the mined neighborhood for intra-camera and inter-camera separately to optimize matching results by exploring neighborhood relations more deeply. Finally, experimental results on three challenging datasets demonstrate that applying our method to unlabeled target domain outperforms current UDA methods for person ReID. [ABSTRACT FROM AUTHOR]

The emergence of wearable devices has opened up new potentials for egocentric activity recognition. Although some methods integrate attention mechanisms into deep neural networks to capture fine-grained human-object interactions in a weak-supervision manner, they either ignore exploiting the temporal consistency or generate attention based on considering appearance cues only. To address these limitations, in this paper, we propose an enhanced attention-tracking method, combined with multi-branch network (EAT-MBNet), for egocentric activity recognition. Specifically, we propose class-aware attention maps (CAAMs) by employing a self-attention-based module to refine the class activation maps (CAMs). Our proposed method can enhance the semantic dependency between the activity categories and the feature maps. To highlight the discriminative features from the regions of interest across frames, we propose a flow-guided attention-tracking (F-AT) module, by simultaneously leveraging historical attention and motion patterns. Furthermore, we propose a cross-modality modeling branch based on an interactive GRU module, which captures the time-synchronized long-term relationships between the appearance and motion branches. Experimental results on four egocentric activity benchmarks demonstrate that the proposed method achieves state-of-the-art performance. [ABSTRACT FROM AUTHOR]

Visible-infrared person re-identification (VI-ReID) is a challenging cross-modality pedestrian retrieval task which aims to match person images between the visible and infrared modality of the same identity. Existing methods usually adopt two-stream network to solve cross-modality gap, but they ignore the pixel-level discrepancy between the visible and infrared images. Some methods introduce auxiliary modalities in the network, but they lack powerful constraints on the feature distribution of multiple modalities. In this letter, we propose a Dynamic Center Aggregation (DCA) loss with mixed modality for VI-ReID. Concretely, we employ a mixed modality as a bridge between the visible and infrared modality, reducing the difference of the two modalities at the pixel-level. The mixed modality is generated by a Dual-modality Feature Mixer (DFM), which combines the features of visible and infrared images. Moreover, we dynamically adjust the relative distance across multi-modality through DCA loss, which is conducive to explore the modality-invariant feature. We evaluate the proposed method on two public available VI-ReID datasets (SYSU-MM01 and RegDB). Experimental results demonstrate that our method achieves competitive performance. [ABSTRACT FROM AUTHOR]

Most existing multi-object tracking models only focus on the spatial information generated by image-level input while ignoring the necessary temporal information. The temporal motion information between consecutive frames can effectively reflect the target's motion status, which is essential to improving the performance in dealing with occlusion and motion blur of the model. We propose MISTracker to realize the motion information supplement to the original tracking model. Specifically, we divide multi-scale feature maps into two categories from the perspective of space and channel information. Meanwhile, the spatial-level frame differences processing (SFDP) module and the channel-level frame differences processing (CFDP) module are proposed to deal with these differences between continuous frames, respectively. The SFDP processes the differences from the perspective of spatial information and supplements the motion information through the perception of pixel-level information changes in the feature maps. The CFDP processes the differences from the perspective of channels and enhances the information of motion-sensitive channels through the overall pixel differences of different channels. Eventually, temporal and motion information are complementary to each other after upsampling fusion. The whole process is realized by simple convolution, which reduces the computational force as much as possible and enhances the tracking performance of the model. [ABSTRACT FROM AUTHOR]

The emergence of low-cost depth cameras creates potential for RGB-D based human action recognition. However, most of the existing RGB-D based approaches simply concatenate original heterogeneous features without discovering the latent relations among different modalities. We propose a discriminative common structure learning (DCSL) model for human action recognition from RGB-D sequences. Specifically, we extract deep learning-based features and hand-crafted features from multimodal data (skeleton, depth, and RGB). In particular, we propose a deep architecture based on 3-D convolutional neural network to automatically extract deep spatiotemporal features from raw sequences. The proposed DCSL model utilizes a generalized version of collective matrix factorization to learn shared features among different modalities. To perform supervised learning and preserve intermodal similarity, we formulate a graph regularization term by considering both label information and similar geometric structure of multimodal data, which intends to improve the discriminative power of shared features. Moreover, we solve the objective function using an iterative optimization algorithm. Then, an improved collaborative representation classifier is employed to perform computationally efficient action recognition. Experimental results on four action datasets demonstrate the superior performance of the proposed method. [ABSTRACT FROM AUTHOR]

The launching of low-cost depth sensors opens up new potentials for RGB-D-based human action recognition. However, most of current RGB-D-based methods simply fuse multimodal features in a holistic manner and ignore the latent connections among different modalities. In this paper, we propose a multimodal correlative representation learning (MCRL) model for human action recognition from RGB-D videos. Specifically, we propose a spatio-temporal pyramid Fourier HOG feature (STPF-HOG) to capture local dynamic patterns around each human joint, which integrates both spatial arrangement and temporal structures. The proposed MCRL model utilizes multimodal data (skeleton, depth, and RGB), and learns shared structures among different modalities. The original low-level features are effectively compressed and projected to a latent subspace. Then, the discriminative shared features are learned by a supervised fashion. We formulate both subspace learning and shared features mining in a modified multi-task learning framework and solve the formulation using an iterative optimization algorithm. To perform computationally efficient action recognition, a robust collaborative representation classifier is presented by introducing a weight regularization matrix. Experimental results on three action datasets demonstrate that the proposed method leads to a more favorable performance compared with the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

End-to-end multiple object tracking (MOT) has been very popular recently in the field of computer vision. However, when performing the inter-frame data association, the majority of algorithms ignore the diversity among trajectories and treat them uniformly. In the joint-detection-and-tracking (JDT) paradigm, there exists an optimization conflict between detection and re-identification due to their different optimization directions. In this paper, we propose a dynamic trajectory quantification (DTQ) strategy for MOT with feature rearrangement. Initially, the DTQ strategy quantifies each trajectory to show its quality, and the score is renewed every frame. To adapt to different situations, we dynamically adopt different updating mechanisms. Additionally, we provide a channel-enhanced feature rearrangement module to alleviate the optimization conflict between subtasks in the JDT paradigm, which obtains more convincing detection results for the DTQ strategy. We evaluate our proposed tracker on several benchmarks, i.e., MOT15, MOT16, MOT17, and MOT20. Extensive experimental results demonstrate that our method achieves competitive results while still maintaining a sizable real-time tracking speed. [ABSTRACT FROM AUTHOR]

The skeleton-based action recognition task has recently drawn much attention in computer vision, and graph convolutional networks (GCNs) have shown great advantages in this task. However, in shallow layers of GCNs, the sharing of convolution kernels in regular graph convolution (RGC) may result in the underutilization of some convolution kernels due to the particularity of the skeleton data. To capture richer locally discriminative action features, we propose the exclusive kernel graph convolution (EKGC) for the shallow layers of GCNs, which assigns the vertexes exclusive convolution kernels for graph convolution. At the same time, following the mixed graph convolution strategy, RGC is retained in the deep layers of the network, and the mixed graph convolutional network (MGCN) is proposed to utilize EKGC and RGC reasonably in all layers of the network. In addition, as a new form of skeleton information in GCN-based methods, angle information is explored in our work. Compared with joint data and bone data, angle data have better invariance against rotation and scaling, and they expand the relationships between skeletons and enrich the data representation of action. Finally, our final model, the multi-stream MGCN, is validated on two challenging datasets, achieving the state-of-the-art performance on the NTU-RGBD dataset and making competitive performance on a kinetics-skeleton dataset. [ABSTRACT FROM AUTHOR]

The popularity of wearable devices has increased the demands for the research on first-person activity recognition. However, most of the current first-person activity datasets are built based on the assumption that only the human-object interaction (HOI) activities, performed by the camera-wearer, are captured in the field of view. Since humans live in complicated scenarios, in addition to the first-person activities, it is likely that third-person activities performed by other people also appear. Analyzing and recognizing these two types of activities simultaneously occurring in a scene is important for the camera-wearer to understand the surrounding environments. To facilitate the research on concurrent first- and third-person activity recognition (CFT-AR), we first created a new activity dataset, namely PolyU concurrent first- and third-person (CFT) Daily, which exhibits distinct properties and challenges, compared with previous activity datasets. Since temporal asynchronism and appearance gap usually exist between the first- and third-person activities, it is crucial to learn robust representations from all the activity-related spatio-temporal positions. Thus, we explore both holistic scene-level and local instance-level (person-level) features to provide comprehensive and discriminative patterns for recognizing both first- and third-person activities. On the one hand, the holistic scene-level features are extracted by a 3-D convolutional neural network, which is trained to mine shared and sample-unique semantics between video pairs, via two well-designed attention-based modules and a self-knowledge distillation (SKD) strategy. On the other hand, we further leverage the extracted holistic features to guide the learning of instance-level features in a disentangled fashion, which aims to discover both spatially conspicuous patterns and temporally varied, yet critical, cues. Experimental results on the PolyU CFT Daily dataset validate that our method achieves the state-of-the-art performance.

Weakly supervised video anomaly detection (WS-VAD) aims to identify the snippets involving anomalous events in long untrimmed videos, with solely video-level binary labels. A typical paradigm among the existing WS-VAD methods is to employ multiple modalities as inputs, e.g., RGB, optical flow, and audio, as they can provide sufficient discriminative clues that are robust to the diverse, complicated real-world scenes. However, such a pipeline has high reliance on the availability of multiple modalities and is computationally expensive and storage demanding in processing long sequences, which limits its use in some applications. To address this dilemma, we propose a privileged knowledge distillation (KD) framework dedicated to the WS-VAD task, which can maintain the benefits of exploiting additional modalities, while avoiding the need for using multimodal data in the inference phase. We argue that the performance of the privileged KD framework mainly depends on two factors: 1) the effectiveness of the multimodal teacher network and 2) the completeness of the useful information transfer. To obtain a reliable teacher network, we propose a cross-modal interactive learning strategy and an anomaly normal discrimination loss, which target learning task-specific cross-modal features and encourage the separability of anomalous and normal representations, respectively. Furthermore, we design both representation-and logits-level distillation loss functions, which force the unimodal student network to distill abundant privileged knowledge from the well-trained multimodal teacher network, in a snippet-to-video fashion. Extensive experimental results on three public benchmarks demonstrate that the proposed privileged KD framework can train a lightweight yet effective detector, for localizing anomaly events under the supervision of video-level annotations.

Invertible image decolorization is a useful color compression technique to reduce the cost in multimedia systems. Invertible decolorization aims to synthesize faithful grayscales from color images, which can be fully restored to the original color version. In this paper, we propose a novel color compression method to produce invertible grayscale images using invertible neural networks (INNs). Our key idea is to separate the color information from color images, and encode the color information into a set of Gaussian distributed latent variables via INNs. By this means, we force the color information lost in grayscale generation to be independent of the input color image. Therefore, the original color version can be efficiently recovered by randomly re-sampling a new set of Gaussian distributed variables, together with the synthetic grayscale, through the reverse mapping of INNs. To effectively learn the invertible grayscale, we introduce the wavelet transformation into a UNet-like INN architecture, and further present a quantization embedding to prevent the information omission in format conversion, which improves the generalizability of the framework in real-world scenarios. Extensive experiments on three widely used benchmarks demonstrate that the proposed method achieves a state-of-the-art performance in terms of both qualitative and quantitative results, which shows its superiority in multimedia communication and storage systems.