Title: A track and field technical evaluation model based on DeepSORT algorithm and YOLOv5 algorithm

Authors: Wu Wen

Addresses: College of Physical Education, Yichun University, Yichun, 336000, China

Abstract: This study presents a run-up speed measurement model for long jump athletes, combining a lightweight target detector (YOLOv5) with a real-time tracking algorithm (DeepSORT). Traditional tracking methods often suffer from high computational load and slow detection speeds; our model addresses these limitations by improving efficiency and accuracy. Experimental results on the Sports-1M and UCF101 datasets show reduced memory usage - down to 64.88% and 68.32%, respectively - after replacing the original detector. On UCF101, the model achieves a precision of 0.957, recall of 0.913, F1 score of 0.927, and accuracy of 0.940, significantly outperforming the baseline. The average speed measurement deviation remains within 0.1m/s, confirming high accuracy. Overall, this model offers an effective tool for tracking and performance evaluation in track and field, providing valuable support for technical analysis.

Keywords: athletics; technical evaluation; target tracking; speed measurement; DeepSORT; YOLOv5.

DOI: 10.1504/IJISTA.2026.153197

International Journal of Intelligent Systems Technologies and Applications, 2026 Vol.24 No.2, pp.136 - 156

Received: 14 Apr 2025
Accepted: 28 Aug 2025
Published online: 29 Apr 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article