Dynamic hypergraph network improves skeleton action recognition accuracy
Geometry-Constrained Dynamic Hypergraph Convolutional Network with Contrastive Score Refinement for Skeleton-Based Action Recognition
Computer Vision and Pattern Recognition
Summary
Recognizing human actions from body skeleton movements is important for many technologies but is challenging due to the complex way body parts move together. The authors created a new method called DyGCN that builds flexible connections between body joints that can change with each action, rather than fixed connections. They also introduced a way to make these connections more precise by refining scores during training using contrastive learning. This approach helped their method perform better than previous ones on popular datasets for action recognition.
What this means in practice
- •For computer vision developers: Enhance software that interprets human actions using body skeleton data by integrating dynamic and geometry-aware hypergraph models for better accuracy.
- •For robotics engineers: Improve robot understanding of human gestures and movements by employing dynamic hypergraph convolutional networks that capture complex joint interactions.
Authors
Minh Ta Tran Nhat, Yuning Ye, Yuseok Ban
Abstract
Skeleton-based action recognition is a pivotal task in computer vision; however, conventional Graph Convolutional Networks (GCNs) are often constrained by pairwise joint modeling, which fails to capture the intricate high-order interactions inherent in human movement. While recent hypergraph-based methods attempt to address this limitation, they typically rely on static structures derived from anatomical priors, lacking the flexibility to capture dynamic, action-specific joint correlations. To this end, we propose DyGCN, a Geometry-Constrained Dynamic Hypergraph Convolutional Network for skeleton-based action recognition. At the core of DyGCN is the Dynamic Hypergraph Construction Network with Contrastive Score Refinement (DyGCT) block, which comprises two novel components: the Geometry-Constrained Dynamic Hypergraph Construction (GeoHC) module and the Contrastive Score Refinement (CSR) module. Specifically, GeoHC dynamically constructs action-specific hyperedges by leveraging geometry-constrained second-order compatibility, transcending the limitations of static hyperedge structures. Furthermore, CSR explicitly refines the pairwise compatibility scores during hypergraph construction through prototype-based contrastive learning, fostering more discriminative action-specific hyperedges, with zero inference-time overhead as CSR operates only during training. Extensive experiments on NTU RGB+D, NTU RGB+D 120, and Northwestern-UCLA validate the effectiveness of DyGCN, achieving state-of-the-art performance on widely used benchmarks. The source code is available at https://github.com/marvineta0405/DyGCN.