TrakAthlete4D: Multi-View On-Field Player Position Tracking in Sports


Nitish Agarwal (KinaTrax), Steven Cadavid (University of Miami)
The 35th British Machine Vision Conference
PDFPosterVideo (Right click to download)Supplementary

Abstract

The field of sports science has undergone substantial evolution, spurred by the development of sophisticated automated data collection techniques. The ability to track player locations and movements is central to this advancement. This paper presents a unique approach for 3D multi-object tracking (3D MOT) in sports settings, utilizing a multi-camera system that captures high-fidelity player data. We make four key contributions: (i) Our method demonstrates successful 3D MOT for sporting environments without requiring custom annotations and additional training; (ii) we prove superior detection performance with an alternate Non-Maximum Suppression (NMS) metric improving over the traditional Intersection Over Union (IOU), effective in occluding scenarios like nets; (iii) We introduce a novel hierarchical strategy to address the minimum k-Clique problem effectively for reliable cross-view bounding box linking; (iv) Lastly, we show efficacy of a third-degree Kalman filter that enhances smoothness and stability of detections over its more commonly used second-degree counterpart.

Citation

@inproceedings{Agarwal_2024_BMVC,
author    = {Nitish Agarwal and Steven Cadavid},
title     = {TrakAthlete4D: Multi-View On-Field Player Position Tracking in Sports},
booktitle = {35th British Machine Vision Conference 2024, {BMVC} 2024, Glasgow, UK, November 25-28, 2024},
publisher = {BMVA},
year      = {2024},
url       = {https://papers.bmvc2024.org/0929.pdf}
}


Copyright © 2024 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection