Article: Sports movement analysis method considering early fusion network structure and key points of human body Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.43 pp.36 - 60 Abstract: With the rapid development of sports and computer technology, accurate sports movement analysis has become crucial for enhancing athlete performance and rehabilitation. Traditional methods face challenges such as difficulty in recognising multi-scene actions and inconsistent sequence lengths. To address this, a novel approach combining an early fusion network with human key point data is proposed. By integrating skeleton node information and using Neville interpolation, the method enhances feature extraction and temporal localisation. Experimental results show significant improvements: compared to traditional models such as LSTM and ST-GCN, the EF-GCN model proposed in this study achieves an increase in classification accuracy of up to 18.5% across various neural networks, and performance metrics such as accuracy, precision, recall, and F1-score improve by around 10%. This approach offers substantial advancements in motion analysis and holds great potential for future sports training and rehabilitation applications. Inderscience Publishers - linking academia, business and industry through research

Title: Sports movement analysis method considering early fusion network structure and key points of human body

Authors: Jianjun Yin; Jie Chen

Addresses: Department of Physical Education, Guangdong University of Finance and Economics, Guangzhou, 510320, China ' School of Humanities and Communication, Guangdong University of Finance and Economics, Guangzhou, 510320, China

Abstract: With the rapid development of sports and computer technology, accurate sports movement analysis has become crucial for enhancing athlete performance and rehabilitation. Traditional methods face challenges such as difficulty in recognising multi-scene actions and inconsistent sequence lengths. To address this, a novel approach combining an early fusion network with human key point data is proposed. By integrating skeleton node information and using Neville interpolation, the method enhances feature extraction and temporal localisation. Experimental results show significant improvements: compared to traditional models such as LSTM and ST-GCN, the EF-GCN model proposed in this study achieves an increase in classification accuracy of up to 18.5% across various neural networks, and performance metrics such as accuracy, precision, recall, and F1-score improve by around 10%. This approach offers substantial advancements in motion analysis and holds great potential for future sports training and rehabilitation applications.

Keywords: early fusion network structure; key points of the human body; sports movement analysis; Neville interpolation method; temporal positioning.

DOI: 10.1504/IJICT.2025.150400

International Journal of Information and Communication Technology, 2025 Vol.26 No.43, pp.36 - 60

Received: 08 Aug 2025
Accepted: 24 Oct 2025
Published online: 12 Dec 2025 *

Title: Sports movement analysis method considering early fusion network structure and key points of human body

Keep up-to-date