Design and Implementation of Horse Riding Action Monitoring Platform Based on Deep Learning

In order to meet the increasing demand of video analysis and its related applications, the method of temporal motion detection based on deep learning has attracted more and more attention. Through the motion detection system, users can use the labeled data for supervised training through the training module of the system to generate a deep learning model for extracting the target semantics, and can also call the model through the prediction module of the system to calculate and analyze the silhouette of the moving person in the video. Accurate analysis of sports training posture by technical means is one of the important means to improve people's competitive level in modern sports. The method adopted in this paper is deep learning. CNN training model is applied to spatial stream (video image) and temporal stream (dense optical stream) respectively to judge the scores of the categories of actions, and different methods are explored to fuse the scores of the two networks to get the final classification results. According to the research in this paper, this method is effective and can be widely used.