A computer vision-based exercise detection system that uses pose estimation to track and count different jump exercises.
- Real-time exercise detection and counting
- Supports multiple exercise types:
- Normal jumps
- High knee jumps
- Jumping jacks
- Criss cross jumps
- Visual feedback with exercise count and pose tracking
pip install -r requirements.txtpython main.pyThis will open a file dialog to select your exercise video. Once selected, the application will:
- Track your body pose using MediaPipe
- Count different exercise types automatically
- Display the counts in real-time
- Show visual tracking of key body points
Press 'q' to quit the application.
VisionJump uses MediaPipe's pose estimation to track body landmarks and applies specialized algorithms to detect different exercise patterns:
- Normal Jumps: Detected by tracking the vertical movement of hips
- High Knee Jumps: Detected by measuring knee angles
- Jumping Jacks: Detected by measuring pelvic angles between legs
- Criss Cross Jumps: Detected by tracking wrist crossover patterns