Skeleton-based Action Recognition Research Articles

Frame-Level Label Refinement for Skeleton-Based Weakly-Supervised Action Recognition

In recent years, skeleton-based action recognition has achieved remarkable performance in understanding human motion from sequences of skeleton data, which is an important medium for synthesizing realistic human movement in various applications. However, existing methods assume that each action clip is manually trimmed to contain one specific action, which requires a significant amount of effort for annotation. To solve this problem, we consider a novel problem of skeleton-based weakly-supervised temporal action localization (S-WTAL), where we need to recognize and localize human action segments in untrimmed skeleton videos given only the video-level labels. Although this task is challenging due to the sparsity of skeleton data and the lack of contextual clues from interaction with other objects and the environment, we present a frame-level label refinement framework based on a spatio-temporal graph convolutional network (ST-GCN) to overcome these difficulties. We use multiple instance learning (MIL) with video-level labels to generate the frame-level predictions. Inspired by advances in handling the noisy label problem, we introduce a label cleaning strategy of the frame-level pseudo labels to guide the learning process. The network parameters and the frame-level predictions are alternately updated to obtain the final results. We extensively evaluate the effectiveness of our learning approach on skeleton-based action recognition benchmarks. The state-of-the-art experimental results demonstrate that the proposed method can recognize and localize action segments of the skeleton data.

Hierarchical Consistent Contrastive Learning for Skeleton-Based Action Recognition with Growing Augmentations

Contrastive learning has been proven beneficial for self-supervised skeleton-based action recognition. Most contrastive learning methods utilize carefully designed augmentations to generate different movement patterns of skeletons for the same semantics. However, it is still a pending issue to apply strong augmentations, which distort the images/skeletons’ structures and cause semantic loss, due to their resulting unstable training. In this paper, we investigate the potential of adopting strong augmentations and propose a general hierarchical consistent contrastive learning framework (HiCLR) for skeleton-based action recognition. Specifically, we first design a gradual growing augmentation policy to generate multiple ordered positive pairs, which guide to achieve the consistency of the learned representation from different views. Then, an asymmetric loss is proposed to enforce the hierarchical consistency via a directional clustering operation in the feature space, pulling the representations from strongly augmented views closer to those from weakly augmented views for better generalizability. Meanwhile, we propose and evaluate three kinds of strong augmentations for 3D skeletons to demonstrate the effectiveness of our method. Extensive experiments show that HiCLR outperforms the state-of-the-art methods notably on three large-scale datasets, i.e., NTU60, NTU120, and PKUMMD. Our project is publicly available at: https://jhang2020.github.io/Projects/HiCLR/HiCLR.html.

Anonymization for Skeleton Action Recognition

Skeleton-based action recognition attracts practitioners and researchers due to the lightweight, compact nature of datasets. Compared with RGB-video-based action recognition, skeleton-based action recognition is a safer way to protect the privacy of subjects while having competitive recognition performance. However, due to improvements in skeleton recognition algorithms as well as motion and depth sensors, more details of motion characteristics can be preserved in the skeleton dataset, leading to potential privacy leakage. We first train classifiers to categorize private information from skeleton trajectories to investigate the potential privacy leakage from skeleton datasets. Our preliminary experiments show that the gender classifier achieves 87% accuracy on average, and the re-identification classifier achieves 80% accuracy on average with three baseline models: Shift-GCN, MS-G3D, and 2s-AGCN. We propose an anonymization framework based on adversarial learning to protect potential privacy leakage from the skeleton dataset. Experimental results show that an anonymized dataset can reduce the risk of privacy leakage while having marginal effects on action recognition performance even with simple anonymizer architectures. The code used in our experiments is available at https://github.com/ml-postech/Skeleton-anonymization/

Self-Supervised Action Representation Learning from Partial Spatio-Temporal Skeleton Sequences

Self-supervised learning has demonstrated remarkable capability in representation learning for skeleton-based action recognition. Existing methods mainly focus on applying global data augmentation to generate different views of the skeleton sequence for contrastive learning. However, due to the rich action clues in the skeleton sequences, existing methods may only take a global perspective to learn to discriminate different skeletons without thoroughly leveraging the local relationship between different skeleton joints and video frames, which is essential for real-world applications. In this work, we propose a Partial Spatio-Temporal Learning (PSTL) framework to exploit the local relationship from a partial skeleton sequences built by a unique spatio-temporal masking strategy. Specifically, we construct a negative-sample-free triplet steam structure that is composed of an anchor stream without any masking, a spatial masking stream with Central Spatial Masking (CSM), and a temporal masking stream with Motion Attention Temporal Masking (MATM). The feature cross-correlation matrix is measured between the anchor stream and the other two masking streams, respectively. (1) Central Spatial Masking discards selected joints from the feature calculation process, where the joints with a higher degree of centrality have a higher possibility of being selected. (2) Motion Attention Temporal Masking leverages the motion of action and remove frames that move faster with a higher possibility. Our method achieves state-of-the-art performance on NTURGB+D 60, NTURGB+D 120 and PKU-MMD under various downstream tasks. Furthermore, to simulate the real-world scenarios, a practical evaluation is performed where some skeleton joints are lost in downstream tasks.In contrast to previous methods that suffer from large performance drops, our PSTL can still achieve remarkable results under this challenging setting, validating the robustness of our method.

Hierarchical Contrast for Unsupervised Skeleton-Based Action Representation Learning

This paper targets unsupervised skeleton-based action representation learning and proposes a new Hierarchical Contrast (HiCo) framework. Different from the existing contrastive-based solutions that typically represent an input skeleton sequence into instance-level features and perform contrast holistically, our proposed HiCo represents the input into multiple-level features and performs contrast in a hierarchical manner. Specifically, given a human skeleton sequence, we represent it into multiple feature vectors of different granularities from both temporal and spatial domains via sequence-to-sequence (S2S) encoders and unified downsampling modules. Besides, the hierarchical contrast is conducted in terms of four levels: instance level, domain level, clip level, and part level. Moreover, HiCo is orthogonal to the S2S encoder, which allows us to flexibly embrace state-of-the-art S2S encoders. Extensive experiments on four datasets, i.e., NTU-60, NTU-120, PKU-I and PKU-II, show that HiCo achieves a new state-of-the-art for unsupervised skeleton-based action representation learning in two downstream tasks including action recognition and retrieval, and its learned action representation is of good transferability. Besides, we also show that our framework is effective for semi-supervised skeleton-based action recognition. Our code is available at https://github.com/HuiGuanLab/HiCo.

Spatial–Temporal gated graph attention network for skeleton-based action recognition

Spatial–Temporal gated graph attention network for skeleton-based action recognition

Multi-stream Global–Local Motion Fusion Network for skeleton-based action recognition

Skeleton-based action recognition is widely used in varied areas such as human–machine interaction and virtual reality. Benefit from the powerful expression ability to depict structural data, graph convolutional networks (GCNs) have been developed to address this task by modeling the human body skeletons as spatial–temporal graphs. However, most existing GCN-based methods usually ignore the diversity of the motion information between channels of the input feature. And how to enhance the ability to capture the long-term global correlations in spatial and temporal dimensions is also a fundamental challenge. In this work, we propose a novel multi-stream framework Global–Local Motion Fusion Network (GLMFN), which integrates the global and local motion information of spatial–temporal dimensions. Specifically, we design a grouping graph convolution module to enforce the ability to aggregate local spatial motion information. Besides, to learn richer semantic features, we propose two modules based on the self-attention operator: a spatial self-attention module and a temporal self-attention module. The former is responsible for extracting spatial long-term motion relationships, while the latter aims to capture temporal long-term motion relationships. Moreover, we present a multi-stream fusion strategy with a series of treatments for body joints to achieve a better recognition effect. To validate the efficacy and efficiency of the proposed model, we perform exhaustive experiments on the NTU-RGBD dataset and NTU-RGBD-120 dataset, and our method achieves the state-of-the-art performance on both datasets.

TFC-GCN: Lightweight Temporal Feature Cross-Extraction Graph Convolutional Network for Skeleton-Based Action Recognition

For skeleton-based action recognition, graph convolutional networks (GCN) have absolute advantages. Existing state-of-the-art (SOTA) methods tended to focus on extracting and identifying features from all bones and joints. However, they ignored many new input features which could be discovered. Moreover, many GCN-based action recognition models did not pay sufficient attention to the extraction of temporal features. In addition, most models had swollen structures due to too many parameters. In order to solve the problems mentioned above, a temporal feature cross-extraction graph convolutional network (TFC-GCN) is proposed, which has a small number of parameters. Firstly, we propose the feature extraction strategy of the relative displacements of joints, which is fitted for the relative displacement between its previous and subsequent frames. Then, TFC-GCN uses a temporal feature cross-extraction block with gated information filtering to excavate high-level representations for human actions. Finally, we propose a stitching spatial–temporal attention (SST-Att) block for different joints to be given different weights so as to obtain favorable results for classification. FLOPs and the number of parameters of TFC-GCN reach 1.90 G and 0.18 M, respectively. The superiority has been verified on three large-scale public datasets, namely NTU RGB + D60, NTU RGB + D120 and UAV-Human.

Multi-stream ternary enhanced graph convolutional network for skeleton-based action recognition

Multi-stream ternary enhanced graph convolutional network for skeleton-based action recognition

Multi-Modality Adaptive Feature Fusion Graph Convolutional Network for Skeleton-Based Action Recognition

Graph convolutional networks are widely used in skeleton-based action recognition because of their good fitting ability to non-Euclidean data. While conventional multi-scale temporal convolution uses several fixed-size convolution kernels or dilation rates at each layer of the network, we argue that different layers and datasets require different receptive fields. We use multi-scale adaptive convolution kernels and dilation rates to optimize traditional multi-scale temporal convolution with a simple and effective self attention mechanism, allowing different network layers to adaptively select convolution kernels of different sizes and dilation rates instead of being fixed and unchanged. Besides, the effective receptive field of the simple residual connection is not large, and there is a great deal of redundancy in the deep residual network, which will lead to the loss of context when aggregating spatio-temporal information. This article introduces a feature fusion mechanism that replaces the residual connection between initial features and temporal module outputs, effectively solving the problems of context aggregation and initial feature fusion. We propose a multi-modality adaptive feature fusion framework (MMAFF) to simultaneously increase the receptive field in both spatial and temporal dimensions. Concretely, we input the features extracted by the spatial module into the adaptive temporal fusion module to simultaneously extract multi-scale skeleton features in both spatial and temporal parts. In addition, based on the current multi-stream approach, we use the limb stream to uniformly process correlated data from multiple modalities. Extensive experiments show that our model obtains competitive results with state-of-the-art methods on the NTU-RGB+D 60 and NTU-RGB+D 120 datasets.

Multi-Granularity Anchor-Contrastive Representation Learning for Semi-Supervised Skeleton-Based Action Recognition.

In the semi-supervised skeleton-based action recognition task, obtaining more discriminative information from both labeled and unlabeled data is a challenging problem. As the current mainstream approach, contrastive learning can learn more representations of augmented data, which can be considered as the pretext task of action recognition. However, such a method still confronts three main limitations: 1) It usually learns global-granularity features that cannot well reflect the local motion information. 2) The positive/negative pairs are usually pre-defined, some of which are ambiguous. 3) It generally measures the distance between positive/negative pairs only within the same granularity, which neglects the contrasting between the cross-granularity positive and negative pairs. Toward these limitations, we propose a novel Multi-granularity Anchor-Contrastive representation Learning (dubbed as MAC-Learning) to learn multi-granularity representations by conducting inter- and intra-granularity contrastive pretext tasks on the learnable and structural-link skeletons among three types of granularities covering local, context, and global views. To avoid the disturbance of ambiguous pairs from noisy and outlier samples, we design a more reliable Multi-granularity Anchor-Contrastive Loss (dubbed as MAC-Loss) that measures the agreement/disagreement between high-confidence soft-positive/negative pairs based on the anchor graph instead of the hard-positive/negative pairs in the conventional contrastive loss. Extensive experiments on both NTU RGB+D and Northwestern-UCLA datasets show that the proposed MAC-Learning outperforms existing competitive methods in semi-supervised skeleton-based action recognition tasks.

Spatial–Temporal Self-Attention Enhanced Graph Convolutional Networks for Fitness Yoga Action Recognition

Fitness yoga is now a popular form of national fitness and sportive physical therapy. At present, Microsoft Kinect, a depth sensor, and other applications are widely used to monitor and guide yoga performance, but they are inconvenient to use and still a little expensive. To solve these problems, we propose spatial–temporal self-attention enhanced graph convolutional networks (STSAE-GCNs) that can analyze RGB yoga video data captured by cameras or smartphones. In the STSAE-GCN, we build a spatial–temporal self-attention module (STSAM), which can effectively enhance the spatial–temporal expression ability of the model and improve the performance of the proposed model. The STSAM has the characteristics of plug-and-play so that it can be applied in other skeleton-based action recognition methods and improve their performance. To prove the effectiveness of the proposed model in recognizing fitness yoga actions, we collected 960 fitness yoga action video clips in 10 action classes and built the dataset Yoga10. The recognition accuracy of the model on Yoga10 achieves 93.83%, outperforming the state-of-the-art methods, which proves that this model can better recognize fitness yoga actions and help students learn fitness yoga independently.

Multi-scale spatial–temporal convolutional neural network for skeleton-based action recognition

Multi-scale spatial–temporal convolutional neural network for skeleton-based action recognition

A Lightweight Attentional Shift Graph Convolutional Network for Skeleton-Based Action Recognition

In the field of skeleton-based human behavior recognition, graph convolutional neural networks have made remarkable achievements. However, high precision networks are often accompanied by numerous parameters and computational cost, and their application in mobile devices has considerable limitations. Aiming at the problem of excessive spatiotemporal complexity of high-accuracy methods, this paper further analyzes the lightweight human action recognition model and proposes a lightweight architecture attentional shift graph convolutional network. There are three main improvements in this model. Firstly, shift convolution is a lightweight convolution method that can be combined with graph convolution to effectively reduce its complexity. At the same time, a shallow architecture for multi-stream early fusion is designed to reduce the network scale by merging multi-stream networks and reducing the number of network layers. In addition, the efficient channel attention module is introduced into the model to capture the underlying characteristic information in the channel domain. Experiments are conducted on the three existing skeleton datasets, NTU RGB+D, NTU-120 RGB+D, and Northwestern-UCLA. Results demonstrate that the proposed model is not only competitive in accuracy, but also outperforms current mainstream methods in parameter count and computational cost, and supports running in some devices with limited computing and storage resources.

Human action recognition model incorporating multiscale temporal convolutional network and spatiotemporal excitation network

Human action recognition is a research hotspot in the field of computer vision. Focusing on the problem of similar action recognition, we propose an improved two-stream adaptive graph convolutional network for skeleton-based action recognition, which incorporating a multiscale temporal convolutional network and a spatiotemporal excitation network. Using the multiscale temporal convolutional network, the temporal information can be effectively extracted by dilated convolution at different scales so as to broaden the width of the temporal network and extract more temporal features with slight difference between categories at the same time. By utilizing the spatiotemporal excitation network, the input features can be obtained through channel pooling to form single-channel features for two-dimensional convolution, by which important spatiotemporal information can be excited and the role of local nodes in similar actions can be effectively enhanced. Extensive tests and ablation studies on the three large-scale datasets, NTU-RGB+D60, NTU-RGB+D120, and Kinetics-Skeleton, were conducted. Our model outperforms the baseline by 7.2% and the state-of-the-art model by 4% in the similar action recognition on NTU-RGB+D60 dataset on average, which demonstrates the superiority of our model.

Skeleton-based Human Action Recognition via Large-kernel Attention Graph Convolutional Network.

The skeleton-based human action recognition has broad application prospects in the field of virtual reality, as skeleton data is more resistant to data noise such as background interference and camera angle changes. Notably, recent works treat the human skeleton as a non-grid representation, e.g., skeleton graph, then learns the spatio-temporal pattern via graph convolution operators. Still, the stacked graph convolution plays a marginal role in modeling long-range dependences that may contain crucial action semantic cues. In this work, we introduce a skeleton large kernel attention operator (SLKA), which can enlarge the receptive field and improve channel adaptability without increasing too much computational burden. Then a spatiotemporal SLKA module (ST-SLKA) is integrated, which can aggregate long-range spatial features and learn long-distance temporal correlations. Further, we have designed a novel skeleton-based action recognition network architecture called the spatiotemporal large-kernel attention graph convolution network (LKA-GCN). In addition, large-movement frames may carry significant action information. This work proposes a joint movement modeling strategy (JMM) to focus on valuable temporal interactions. Ultimately, on the NTU-RGBD 60, NTU-RGBD 120 and Kinetics-Skeleton 400 action datasets, the performance of our LKA-GCN has achieved a state-of-the-art level.

Cross-stream contrastive learning for self-supervised skeleton-based action recognition

Self-supervised skeleton-based action recognition enjoys a rapid growth along with the development of contrastive learning. The existing methods rely on imposing invariance to augmentations of 3D skeleton within a single data stream, which merely leverages the easy positive pairs and limits the ability to explore the complicated movement patterns. In this paper, we advocate that the defect of single-stream contrast and the lack of necessary feature transformation are responsible for easy positives, and therefore propose a Cross-Stream Contrastive Learning framework for skeleton-based action Representation learning (CSCLR). Specifically, the proposed CSCLR not only utilizes intra-stream contrast pairs, but introduces inter-stream contrast pairs as hard samples to formulate a better representation learning. Besides, to further exploit the potential of positive pairs and increase the robustness of self-supervised representation learning, we propose a Positive Feature Transformation (PFT) strategy which adopts feature-level manipulation to increase the variance of positive pairs. To validate the effectiveness of our method, we conduct extensive experiments on three benchmark datasets NTU-RGB + D 60, NTU-RGB + D 120 and PKU-MMD. Experimental results show that our proposed CSCLR exceeds the state-of-the-art methods on a diverse range of evaluation protocols.

Employing FGP-3D, a Fully Gated and Anchored Methodology, to Identify Skeleton-Based Action Recognition

Recent years have seen an explosion in interest in and development of action recognition based on skeletal data. Contemporary methods using fully gated units can successfully extract characteristics from human skeletons by relying on the human topology that has been predefined. Despite advancements, fully gated unit-based techniques have trouble generalizing to other domains, particularly when dealing with various human topological structures. In this context, we introduce FGP-3D, a novel skeleton-based action recognition technique that can generalize across datasets while being effective at learning spatiotemporal features from human skeleton sequences. This is accomplished via a multi-head attention technique to learn an ideal dependence feature matrix from the uniform distribution. We next re-evaluate state-of-the-art techniques as well as the suggested novel descriptor FGP-3D in order to examine the cross-domain generalizability of skeleton-based action recognition in real-world video skeleton statistics. After being applied to commonly used action categorization datasets, experimental results demonstrate that the proposed FGP-3D, with pre-training, generalizes well and outperforms the state-of-the-art.

ICE-GCN: An interactional channel excitation-enhanced graph convolutional network for skeleton-based action recognition

Thanks to the development of depth sensors and pose estimation algorithms, skeleton-based action recognition has become prevalent in the computer vision community. Most of the existing works are based on spatio-temporal graph convolutional network frameworks, which learn and treat all spatial or temporal features equally, ignoring the interaction with channel dimension to explore different contributions of different spatio-temporal patterns along the channel direction and thus losing the ability to distinguish confusing actions with subtle differences. In this paper, an interactional channel excitation (ICE) module is proposed to explore discriminative spatio-temporal features of actions by adaptively recalibrating channel-wise pattern maps. More specifically, a channel-wise spatial excitation (CSE) is incorporated to capture the crucial body global structure patterns to excite the spatial-sensitive channels. A channel-wise temporal excitation (CTE) is designed to learn temporal inter-frame dynamics information to excite the temporal-sensitive channels. ICE enhances different backbones as a plug-and-play module. Furthermore, we systematically investigate the strategies of graph topology and argue that complementary information is necessary for sophisticated action description. Finally, together equipped with ICE, an interactional channel excited graph convolutional network with complementary topology (ICE-GCN) is proposed and evaluated on three large-scale datasets, NTU RGB+D 60, NTU RGB+D 120, and Kinetics-Skeleton. Extensive experimental results and ablation studies demonstrate that our method outperforms other SOTAs and proves the effectiveness of individual sub-modules. The code will be published at https://github.com/shuxiwang/ICE-GCN.

2s-GATCN: Two-Stream Graph Attentional Convolutional Networks for Skeleton-Based Action Recognition

As human actions can be characterized by the trajectories of skeleton joints, skeleton-based action recognition techniques have gained increasing attention in the field of intelligent recognition and behavior analysis. With the emergence of large datasets, graph convolutional network (GCN) approaches have been widely applied for skeleton-based action recognition and have achieved remarkable performances. In this paper, a novel GCN-based approach is proposed by introducing a convolutional block attention module (CBAM)-based graph attention block to compute the semantic correlations between any two vertices. By considering semantic correlations, our model can effectively identify the most discriminative vertex connections associated with specific actions, even when the two vertices are physically unconnected. Experimental results demonstrate that the proposed model is effective and outperforms existing methods.