Our latest work has been published in Neurocomputing (Volume 637, 7 July 2025).

Title: Dstsa-Gcn: Advancing Skeleton-Based Gesture Recognition with Semantic-Aware Spatio-Temporal Topology Modeling
Authors: Hu Cui, Renjing Huang, Ruoyu Zhang, Tessai Hayama

https://doi.org/10.1016/j.neucom.2025.130066

Graph Convolutional Networks (GCNs) have great potential for recognizing human gestures from skeleton data. However, existing methods struggle to capture dynamic and multiscale patterns. Our proposed method, DSTSA-GCN, addresses this with:

• GC-GC & GT-GC: New modules for modeling correlations across channels and time frames.

• MS-TCN: A multi-scale convolutional module to handle temporal diversity.

• Semantic-Aware Topology Modeling: Better understanding of gesture structure and motion.

Our method achieves state-of-the-art performance on key benchmarks like SHREC’17, DHG-14/28, and NTU-RGB+D.

Code Available: https://hucui2022.github.io/dstsa_gcn/