Design and Research of a Martial Arts Routine Scoring System Based on Pose Recognition

With the continuous development of artificial intelligence, martial arts routine scoring systems based on pose recognition have gradually become an important auxiliary tool in martial arts training and competitions. This study aims to design a scoring system that utilizes deep learning and computer vision technologies to accurately recognize and evaluate martial arts movements through pose recognition algorithms. First, the fundamental concepts and developmental history of pose recognition technology are reviewed, and its applications in martial arts scoring are explored. Next, the design requirements for martial arts routine scoring are analyzed, and the system's functional, performance, and user requirements are proposed. Then, a complete system architecture is designed, which includes a data processing module, a pose recognition module, an action recognition and scoring module, and a user interaction module. To achieve high-precision action scoring, this paper details the pose recognition and action recognition methods, focusing on deep learning-based pose recognition techniques and scoring strategies. Finally, the system is developed, implemented, and tested, with its accuracy and scoring consistency evaluated. Experimental results demonstrate that the proposed scoring system can effectively improve the scoring precision of martial arts movements, providing robust technical support for martial arts teaching and competitions.