AI Gym Bro — Computer Vision Fitness Tracker
A Python-based application using computer vision to track exercise reps via webcam with real-time form feedback. Detects body pose landmarks to count reps and alert on incorrect form.
View on GitHubHow it works
AI Gym Bro uses your webcam feed and MediaPipe's pose estimation model to detect 33 body landmark points in real time. By analyzing the angles between key joints (shoulder, elbow, wrist), the app determines whether you are in the "up" or "down" position of an exercise rep.
Each time a full rep is completed — going from fully extended to contracted and back — the counter increments. If your form deviates from expected joint angles, an on-screen alert prompts you to correct your posture, helping prevent injury during training.
Tech stack
The app is built entirely in Python, leveraging OpenCV for webcam capture and frame rendering, and MediaPipe for real-time pose landmark detection. NumPy is used for joint angle calculations. The system runs at 30+ FPS on a standard laptop without a dedicated GPU.
Project outcomes
AI Gym Bro successfully tracks bicep curls, squats, and push-ups with high accuracy. The form-feedback system alerts users in real time, reducing the risk of injury during self-guided workouts. The project demonstrates practical applications of computer vision and showcases Python skills in building interactive, real-time systems.
View on GitHub