PathCRF: Ball-Free Soccer Event Detection via Possession Path Inference from Player Trajectories
Researchers have developed PathCRF, a new framework for detecting soccer events using only player tracking data, eliminating the need for ball tracking. This method models player trajectories as a dynamic graph and uses a Conditional Random Field to infer possession states. Experiments demonstrate PathCRF's accuracy in identifying events like passes and controls, which can significantly reduce manual annotation efforts and costs. AI
IMPACT Enables more accessible and cost-effective data collection for sports analytics, potentially democratizing data-driven insights in soccer.