Article: Biomechanical analysis of multi-sensor fusion shooting action based on spatiotemporal graph convolutional network Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.82 pp.1 - 16 Abstract: In the field of basketball biomechanical analysis, traditional methods often rely on single sensors or two-dimensional videos, making it difficult to comprehensively capture the complex spatiotemporal characteristics of shooting movements. This study proposes a multi-sensor fusion analysis method based on spatiotemporal graph convolutional networks (ST-GCN), aiming to achieve accurate and automated analysis of shooting movements. Specifically, a novel multi-stream fusion ST-GCN model architecture is first constructed, which can simultaneously process heterogeneous time series data from inertial measurement unit (IMU) and motion capture systems. Secondly, a cross-modal data fusion strategy based on attention mechanism is proposed, effectively enhancing the representation ability of key features such as joint angles and motion trajectories. Finally, end-to-end motion recognition and synchronous analysis of biomechanical parameters (such as joint torques and angular velocities) are achieved. Experiments collected multimodal data from over 2,000 shooting movements of 50 athletes. The results show that the method achieves an action classification accuracy of 96.7% on the complete dataset, and an accuracy of 92.5% in the IMU+sEMG sensor combination verification. When predicting key biomechanical parameters, the correlation coefficient with professional force platform data is as high as 0.93, significantly outperforming traditional analysis methods based on rules or single sensors. Inderscience Publishers - linking academia, business and industry through research

Title: Biomechanical analysis of multi-sensor fusion shooting action based on spatiotemporal graph convolutional network

Authors: Haolin Pang

Addresses: Faculty of Physical Education, Pingdingshan University, Pingdingshan, 467000, China

Abstract: In the field of basketball biomechanical analysis, traditional methods often rely on single sensors or two-dimensional videos, making it difficult to comprehensively capture the complex spatiotemporal characteristics of shooting movements. This study proposes a multi-sensor fusion analysis method based on spatiotemporal graph convolutional networks (ST-GCN), aiming to achieve accurate and automated analysis of shooting movements. Specifically, a novel multi-stream fusion ST-GCN model architecture is first constructed, which can simultaneously process heterogeneous time series data from inertial measurement unit (IMU) and motion capture systems. Secondly, a cross-modal data fusion strategy based on attention mechanism is proposed, effectively enhancing the representation ability of key features such as joint angles and motion trajectories. Finally, end-to-end motion recognition and synchronous analysis of biomechanical parameters (such as joint torques and angular velocities) are achieved. Experiments collected multimodal data from over 2,000 shooting movements of 50 athletes. The results show that the method achieves an action classification accuracy of 96.7% on the complete dataset, and an accuracy of 92.5% in the IMU+sEMG sensor combination verification. When predicting key biomechanical parameters, the correlation coefficient with professional force platform data is as high as 0.93, significantly outperforming traditional analysis methods based on rules or single sensors.

Keywords: spatiotemporal graph convolutional network; ST-GCN; multi-sensor fusion; shooting action analysis; biomechanics; attention mechanism.

DOI: 10.1504/IJICT.2026.155119

International Journal of Information and Communication Technology, 2026 Vol.27 No.82, pp.1 - 16

Received: 24 Oct 2025
Accepted: 15 Dec 2025
Published online: 27 Jul 2026 *

Title: Biomechanical analysis of multi-sensor fusion shooting action based on spatiotemporal graph convolutional network

Keep up-to-date