Design and Implementation of Track and Field Trajectory Analysis System Combining Object Detection Algorithm

With the growing demand for scientific training in the field of athletics, data analysis has become increasingly rigorous, and object detection algorithms have emerged as highly promising tools in trajectory analysis systems. This paper aims to solve the problem of track and field sports trajectory analysis, namely, how to efficiently and accurately capture and analyze the movement trajectory of athletes. This paper designs and implements a trajectory analysis system for track and field sports based on an object detection algorithm. The system architecture consists of five core modules: video preprocessing, object detection, trajectory reconstruction, trajectory analysis, and user interface. The system utilizes the YOLOv2 object detection algorithm, the Hungarian algorithm for association, and the Kalman filter for trajectory smoothing. Compared with traditional trajectory analysis methods, the average analysis time of the proposed system is only 347 milliseconds, with an average point loss rate of 4.91%, and significantly reduced endpoint displacement error. User feedback indicates strong performance in usability, functional completeness, and stability, although further optimization in interface design and feature scalability is still needed.