video1.0<iframe src="https://www.loom.com/embed/d1db300c4d984e2bbacc141165de3885" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/d1db300c4d984e2bbacc141165de3885-817216a19f1aaf05.gif323.935Hi team, this is my walkthrough for the Computer Vision Internship assignment. I used a YouTube video I downloaded because it forces the tracker to handle real world challenges like fast camera panning, similar jerseys, and heavy occlusions. My pipeline uses YOLOv8m for detection and BoT SORT with camera motion compensation, and I set the ID persistence to 120 frames, about 4 seconds, to reduce ID switching. I also generated heatmaps using KDE plus a player count over time, and I pushed the full code to GitHub. No specific viewer action was requested.