Basketball AI with RF-DETR, SAM2, and SmolVLM2

Learn to build a comprehensive computer vision system for basketball analytics by combining cutting-edge AI models including RF-DETR for object detection, SAM2 for player tracking, and SmolVLM2 for jersey number recognition. Follow along as this 37-minute tutorial demonstrates frame-by-frame analysis of a complete basketball possession, covering player identification, team clustering using SigLIP and UMAP, shot detection and classification, and homography mapping to visualize player movements on a top-down court representation. Master the integration of multiple open-source models to create a robust sports analytics pipeline, with access to datasets for basketball player detection, court keypoint detection, and jersey number OCR, plus comprehensive Jupyter notebooks for hands-on implementation of each component in the system.