Tracking multiple objects in sports broadcasts is challenging because athletes move quickly and change direction suddenly, while the camera simultaneously pans and zooms. These combined motions make standard tracking algorithms unreliable.
ARC-SORT (Adaptive, Robust, Compensated SORT) is a new method designed for this setting. Its name indicates that it adapts to changing motion, remains robust under difficult conditions, and compensates for camera movement. The approach is presented in a preprint on arXiv, though the abstract does not yet include experimental results.
The work highlights the need for trackers that handle both target motion and camera motion together, rather than treating them separately. As sports analytics and automated broadcasting grow, such methods could improve player and ball tracking in real time.