Scientific Training Model for Table Tennis Players Based on Artificial Neural Network Algorithm

Table tennis has five core elements: speed, power, spin, arc, and landing point, of which spin is the most crucial. This paper focuses on table tennis players and proposes a detection and tracking algorithm based on artificial neural networks. It designs and implements a table tennis tactical indicator training model, which includes the detection of table tennis movement trajectories. The calculation of table tennis trajectories can also be applied in formal table tennis matches to display real-time spin speed, providing feedback to athletes for scientific training. After in-depth research, we developed an LSTM-based trajectory prediction algorithm, which effectively addresses the complexity of table tennis chasing and spin measurement during training, confirmed through extensive testing.