I co-developed this project with a Hackathon team during a VTHacks 12. This project won the Peraton sponsor's award for "Best Implementation of Generative AI". Rep-Right is a browser-based exercise form analysis application that uses a device’s camera and real-time pose detection to evaluate exercise technique. The app supports squats, push-ups, lunges, and planks, tracking repetitions or hold time while analyzing joint angles and body positioning. It provides live visual feedback, highlights potentially problematic joints, identifies issues such as shallow depth, knee valgus, excessive torso lean, hip sagging, or hip piking, and maintains a history of completed repetitions and form quality. Powered by TensorFlow.js and MoveNet, all video processing occurs locally in the browser, meaning camera data is not uploaded to an external server.
- Won Peraton's "Best Implementation of Generative AI".
Video is processed locally in your browser while using the application. No data is uploaded to any sources. This app is intended to provide general exercise form cues and help collegiate athletes and bodybuilders to correct their training form. This application should never be used as a substitute for coaching or physical therapy. Avoid using the application to perform activities that may cause damage or physical injury to yourself or others. Exercise at your own risk.