A Computer Vision Framework for Multi-Class Detection and Tracking in Soccer Broadcast Footage

A new research paper presents a computer vision framework that could democratize professional sports analytics by extracting detailed tracking data from standard broadcast footage. The system, developed by researcher Daniel Tshiani, combines a YOLO (You Only Look Once) object detector with the ByteTrack multi-object tracking algorithm to create an end-to-end pipeline for soccer analysis. It identifies and tracks four key classes throughout a match: players, referees, goalkeepers, and the ball.

Experimental results show the pipeline achieves high performance in detecting and tracking players and officials, with strong precision, recall, and mAP50 (mean Average Precision at 50% Intersection over Union) scores. However, the research identifies ball detection as the primary remaining challenge due to its small size and rapid movement. Despite this limitation, the framework successfully demonstrates that meaningful player-level spatial information can be extracted from a single broadcast camera feed.

The approach addresses a significant inequality in sports analytics where elite clubs with access to expensive multi-camera setups or GPS tracking systems gain competitive advantages through detailed performance data. By reducing reliance on specialized hardware, this computer vision solution enables colleges, academies, and amateur clubs to adopt scalable, data-driven analysis methods previously accessible only to professional teams. The research highlights the growing potential for affordable AI-powered sports analytics that could level the playing field across different tiers of competition.