Battle strategy auxiliary training method based on historical tennis game video analysis

The invention provides a battle strategy aided training method based on historical tennis game video analysis, and the method comprises the steps: carrying out the analysis according to a tennis motion track, obtaining a batting track of a fighter and related refined data, and giving an auxiliary suggestion. A two-dimensional tennis track is provided with three-dimensional information by defining single-shot ball return quality, collecting and marking battle data information of a tennis match and rendering colors, a perception information code of the track is obtained through a perception network, and scene memory and semantic memory are generated through a memory network; and corresponding quantized data are added to the generated prediction track of the response network, the prediction track and the actual player return track are confronted, a supervision error is minimized, a prediction response module is continuously optimized, a tennis battle return auxiliary model is formed, historical tennis game data