Advancing Martial Arts Training: Neural Network-Based Recognition and Assistance Systems in Biotechnological Applications

Abstract

This study harnesses the capabilities of neural network algorithms to enhance the recognition and assisted training of martial arts movements, a vital component of cultural heritage and physical education. Utilizing decomposed node data from martial arts routines, we developed a method to capture and describe subtle movement changes through descriptive data analytics. Directional vector data, derived from these analyses, enables the precise identification of complete martial arts sequences, facilitating the matching of highly similar martial arts routines based on action data vector selection. Further, we introduced an innovative technique to track joint angle changes over time, significantly enhancing the accuracy of movement recognition. Our results indicate a peak accuracy of 94.58% in action recognition, with the neural network's ability to support martial arts bending-assisted training reaching an effectiveness of 97%. These findings not only demonstrate the potential of neural network technologies to improve training outcomes but also highlight their role in promoting the preservation and development of martial arts as a traditional cultural practice. By integrating neural network algorithms with biotechnological approaches, this research contributes to the development of advanced biotechnological tools in sports and cultural training applications. It underscores the commercial and educational potentials of applying advanced computational techniques in traditional disciplines, paving the way for further innovations in the field of biotechnology and martial arts education.