Research on Traditional Martial Arts Action Recognition Algorithm Based on OpenPose

This study puts forth a traditional martial arts action recognition algorithm using OpenPose with the intention of enhancing the accuracy and analysis of traditional martial arts movements. Through the analysis of the principles of OpenPose, we developed a systematic recognition framework, in which we applied data preprocessing algorithms, extracted features, and modeled martial arts postures. In combination with the optimization algorithms, this approach increased the precision and fastness of recognition. The experiment results showed that this proposed algorithm was very efficient and could recognize certain movements in martial arts more accurately. The research concludes that the suggested method can be utilized to support martial arts education, movement monitoring, and training, and it offers a novel method for the digital transmission of traditional culture.