Research on badminton action feature recognition based on improved HMM model

The badminton movement speed is fast, and the movement is complicated. Therefore, it is difficult to effectively recognize the athlete's movement through the monitoring level in the competition and training, which makes it difficult for the athlete to effectively improve his skill. In order to effectively improve the training effect and the quality of the athletes, this study uses badminton as the research object, analyzes the sports characteristics research algorithm through literature review, and finds the shortcomings of traditional algorithms. At the same time, this paper combines the actual situation to improve the algorithm and combines GMM and HMM to builds the GMM-HMM model. In addition, this paper uses the Baum-Welch unsupervised learning algorithm for data processing, and based on the learning machine training, the recognition results are obtained. Finally, in order to verify the validity of the model, this study uses the mobile phone badminton action as the data foundation and performs training recognition in the model to summarize the recognition results. The research shows that the algorithm has good performance and can meet the actual needs and can be used as a reference for the subsequent related research corporal punishment theory.