Presented at the CVPR 2024 Workshop on Computer Vision in Sports (CVSports), held alongside the main Computer Vision and Pattern Recognition conference in Seattle, WA.
This work examined the generalization capabilities of action spotting models — originally developed for football/soccer — when applied to other sports domains including handball and basketball. The study provided the first systematic cross-domain transfer analysis using the SoccerNet and Play-by-Play datasets.
Presenting at CVPR brought this work to an international audience of leading computer vision researchers and highlighted the breadth of the sports AI research being conducted at UDC in collaboration with teams from KU Leuven, KAUST, and other institutions.