Detection and Tracking Moving Targets for Martial Arts Arena Video Using Reinforcement Learning

The existing detection and tracking algorithm of moving objects in martial arts challenge videos is not smooth in the implementation of video feedback mode, which leads to a large error in the algorithm. A detection and tracking algorithm of moving objects in martial arts challenge video based on reinforcement learning is designed. Based on the principles of movement and mechanics, the paper identifies the competitive rules of martial arts arena competition, strengthens learning, optimizes the video feedback mode, establishes the evaluation criteria of video classification, rating and remarks, describes the video background with multi-peak distribution model, detects the moving target, and tracks the moving target according to the detection content. The results show that the average error of the detection and tracking algorithm of moving objects in the martial arts arena competition video designed this time and the other two algorithms are respectively 3.314, 5.329, and 5.306, indicating that the performance of the detection and tracking algorithm of moving objects in the martial arts arena competition video designed after the combination of reinforcement learning is better.