Football trajectory prediction and real-time feedback mechanism based on Temporal Convolutional Network

Abstract

In modern football matches, athlete trajectory analysis and real-time feedback play a crucial role in tactical planning and game management. Traditional trajectory prediction methods struggle to provide accurate real-time feedback due to an insufficient capture of dynamic changes. This study constructs a model for football trajectory prediction and real-time feedback based on temporal convolutional networks (TCN) and spatial channel attention. This model can effectively capture the complex motion trajectories of players and balls in time and space representation, and generate real-time feedback information to assist coach decision-making through comparison with historical data. In addition, by introducing joint trajectory prediction, we aim to address the issue of poor handling of long-term dependencies in traditional methods, and improve the accuracy and real-time performance of trajectory prediction. In the experiment, we validated the method using actual game data, and our method successfully predicted the trajectory of football in actual games. The experimental results showed that the method has high accuracy in predicting the trajectory of players and balls, with PPA and TPA reaching 65.85% and 87.12% improved by 1.73% and 1.6%, respectively, which is better than existing baseline models and can provide accurate tactical feedback.

Keywords

• Trajectory tracking
• Real-time detection
• Deep learning applications
• Athletic sports
• Attention