Overview
Explore computer vision and deep learning techniques to create an automatic basketball game detection system. Learn to extract frames from YouTube videos, generate automatic annotations using Autodistill and GroundingDINO, train a custom YOLOv8 model for object detection in sports, and perform real-time inference on new game footage. Gain hands-on experience with essential tools for AI-based sports analytics while working with game footage from a Real Madrid vs. Maccabi Tel-Aviv match. Master the full pipeline from data labeling to live inference, and develop practical skills in deep learning for sports video analysis.
Syllabus
Introduction & Demo
Installation
Generate the dataset
Automatic dataset annotation
Train a model
Test the mode
Taught by
Eran Feit