Using Computer Vision to Analyze Basketball Player Movement

This presentation explores advanced computer vision techniques in sports, focusing on player detection challenges and an innovative analysis pipeline. It provides an in-depth overview of a multi-model system utilising cutting-edge tools such as RF-DETR, SAM2, SmolVLM2, and SigLIP for tasks like tracking, jersey classification, and team clustering. The workflow highlights detection and processing methods while addressing challenges like motion blur and occlusion. Results emphasise optimisation...