Hooplab

The Project

Every basketball player knows the frustration of practicing alone — shooting hundreds of shots with no feedback, no coach, and no way to know what's actually going wrong. HoopLab is an interactive AI-powered mobile experience, already available on the AppStore, that transforms solo practice into an engaging, game-like training loop where players shoot, see their results instantly, and level up their technique in real time. The interaction is simple and satisfying: a player records or uploads a video clip of their shot, trims it using HoopLab's built-in video editor, and selects an analysis mode — the app then overlays the optimal parabolic arc directly onto their actual shot, showing frame-by-frame exactly where their release went off and what needs to change. Powered by a custom-trained YOLOv11 computer vision model built on a dataset of over 8,200 images with 94.7% detection accuracy, HoopLab identifies the ball, rim, and player pose on-device in seconds — no internet required. Built with Flutter for cross-platform support on both iOS and Android, the app uses FFmpeg for high-quality frame extraction and parabolic interpolation to mathematically compute the ideal shot trajectory for any position on the court. What makes HoopLab truly unique is how it turns objective AI analysis into something visual, immediate, and genuinely fun — players compete against their own previous scores, chase a perfect arc, and actually see themselves improving. At a time when quality coaching is inaccessible to millions of young players, HoopLab democratizes elite-level feedback and puts it in everyone's pocket.

View this project on YouTube