AI Smart Rep Counter using OpenCV

This project is an AI-powered exercise tracking system that detects and counts repetitions for various exercises like bicep curls, squats, push-ups, lateral raises, and shoulder presses using OpenCV and MediaPipe. It provides real-time feedback with rep counts, exercise stage, and smooth angle measurements.

---

📂 Dataset & Input

• Real-time webcam video feed is used as input.
• No static dataset required; the system processes your body pose live using MediaPipe Pose landmarks.

---

🔧 Tools & Libraries

• Python
• OpenCV
• MediaPipe
• NumPy
• Flask (for web interface)
• HTML/CSS/JavaScript (for frontend UI)

---

📊 Project Highlights

1. Exercise Detection & Rep Counting

• Detects body landmarks using MediaPipe Pose
• Calculates joint angles for relevant exercises
• Counts repetitions with hysteresis thresholds and smooth angle tracking
• Tracks the stage of exercise (up/down) for accurate rep counting

2. Supported Exercises

• Bicep Curl – Elbow angle detection
• Squat – Knee angle detection
• Push-Up – Average elbow angle
• Lateral Raise – Shoulder abduction angle
• Shoulder Press – Elbow/shoulder extension

3. Real-Time Feedback

• Displays live rep count, stage, and current angle on the video feed
• Alerts when landmarks are not visible

---

📈 Sample Interface

• Live video feed with overlays for:

  • Exercise name
  • Reps counted
  • Stage (up/down)
  • Joint angle

• Buttons to switch between exercises dynamically

---

📌 Future Improvements

• Store full workout sessions for analysis
• Generate visual insights like graphs of reps, sets, and exercise distribution
• Add more exercises and custom workout routines
• Integrate with a mobile-friendly interface

---

🤝 Acknowledgements

• MediaPipe for pose estimation
• OpenCV for video processing