Intelligent System of Badminton Serve Action Based on YOLOv5 and OpenPose

In badminton matches, players from both sides hit the ball very fast. In peacetime training, coaches and players often need to watch videos repeatedly to capture action information, which leads to the slow efficiency of correcting action. In view of the above problems, this paper proposes a method of using YOLOv5 technology for target detection of players and using OpenPose technology to establish a comparative scoring model of serving motion. First, the standard action images of badminton players were shot in professional venues, and the key nodes of all players were marked to initially establish the exclusive data set of badminton matches. Secondly, players with data similarity concentration or fuzzy data are intercepted and trained for many times to obtain a large number of accurate data sets and improve the recognition degree of the model to occluded or fuzzy targets. Finally, through the comparison of the figure service action and standard service action, score from the perspective of difference, judge whether the player's action is standard, so as to guide each player's action. The experimental results show that the badminton analysis model has a more accurate detection effect on the players, greatly shortens the training time, and improves the training efficiency of badminton players.