Title: Research on GCN-based aerobics movement recognition under the background of big data

Authors: Li Shang

Addresses: Physical Culture Institute, Xianyang Normal University, Xianyang, 712000, China

Abstract: The effective recognition of aerobics is a powerful guarantee to predict athletes' physical injury in time, and also a powerful tool to improve the standard level of aerobics. To solve the problem of low nonlinear ability of spatiotemporal graph Convolutional network (STGCN) model, a dense connected network structure based on STGCN is proposed in this paper. Finally, an aerobics recognition model is constructed by combining the dense spatiotemporal graph Convolutional network (DSTGCN) algorithm. The integrated model first analyses and pre-processes the images in aerobic exercise videos to construct a directed spatiotemporal map of human bones. The resulting skeleton topology is then embedded in the DSTGCN network, where it is learned and updated along with the model. The experimental results show that the final recognition accuracy of DSTGCN model is stable at about 86.95%, which is better than other existing algorithms.

Keywords: big data; STGCN; DSTCGN; directed space-time graph; aerobics exercise; motion recognition.

DOI: 10.1504/IJCSYSE.2025.149218

International Journal of Computational Systems Engineering, 2025 Vol.9 No.2/3/4, pp.187 - 195

Received: 10 Apr 2023
Accepted: 19 Jul 2023
Published online: 20 Oct 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article