SAGCN: A spatiotemporal attention-weighted graph convolutional network with IoT integration for adolescent tennis motion analysis

Adolescent athletes, especially tennis players, face challenges in motion and health management during skill development, with traditional methods often lacking accuracy and real-time efficiency. While deep learning-based action recognition has advanced, issues like computational complexity and real-time performance in dynamic environments remain. To address these issues, we propose the SAGCN model, which integrates real-time feedback from IoT-enabled wearables to enhance athlete motion and health analysis. The model combines dynamic attention-weighted Graph Convolutional Networks (GCNs) with lightweight channel attention mechanisms to optimize joint relationships and efficiently capture spatiotemporal features. This approach significantly reduces computational complexity while maintaining high accuracy. The SAGCN model combines spatiotemporal feature extraction with dynamic GCNs and attention mechanisms to provide an efficient and accurate solution for both motion analysis and health intervention. Experimental results show that SAGCN outperforms state-of-the-art methods, achieving 96.4%, 96.7%, and 95.3% accuracy on the NTU RGB+D 60, NTU RGB+D 120, and UCF101 datasets, respectively. This model provides a robust solution for real-time monitoring and intervention in adolescent tennis players, with broad potential applications in smart sports and health management.