A Study on Table Tennis Trajectory Prediction Methods Based on the Informer Model

The charm of table tennis lies in its fast pace and complex technical requirements, with accurate trajectory prediction being one of the key factors for improving competitive performance. With the rapid development of artificial intelligence and deep learning technologies, data-driven trajectory prediction methods have gradually become a research hotspot. In the study, by comparing various prediction algorithms, including SPM, STM, and Seq2Seq models, it was found that they have certain limitations in handling table tennis trajectories and cannot meet the high-precision real-time prediction demands. Based on this, the Informer model was introduced, leveraging its efficient time series processing capabilities to significantly enhance prediction accuracy and efficiency. Furthermore, improvements to the Informer model have further promoted its application in table tennis trajectory prediction, achieving even better results. By constructing a binocular vision acquisition system and combining it with the physical motion model of table tennis, the research successfully captured and analyzed the trajectories of table tennis movements, further validating the effectiveness of the prediction algorithms. This research not only provides theoretical support for the intelligent development of table tennis but also lays the foundation for optimizing training and competition strategies in practical applications, showcasing the infinite possibilities of combining technology and sports.