This notebook delivers a full end-to-end pipeline using Autodistill + Grounded SAM to generate segmentation-based annotations from video frames, convert them to YOLO bounding boxes, and fine-tune a YOLOv8 model all inside a single environment.