A real-time workout tracking application using computer vision to analyze shoulder presses. This tool provides immediate feedback on form, tracks repetitions, and records workout data for further analysis. Built with MediaPipe, OpenCV, and Python, this project is a powerful tool for fitness enthusiasts and trainers.
The Computer Vision Workout Tracker leverages pose estimation and video processing to track key body movements during shoulder presses. This project provides:
- Real-time visual feedback with color-coded exercise states.
- Accurate repetition counting.
- A CSV file containing detailed workout data for performance analysis.
- Real-Time Tracking: Detects key body landmarks such as shoulders, elbows, and wrists.
- State Classification:
- RESTING: Wrist is below the midpoint between the shoulder and hip.
- FLEXED: Arm is bent during the press motion.
- EXTENDED: Arm is straight, typically at the top of the press.
- Color-Coded Feedback:
- Green: RESTING
- Yellow: FLEXED
- Red: EXTENDED
- Live Camera Feed Integration: Tracks movements in real-time using a webcam.
- Video Analysis: Analyze pre-recorded workout videos.
- Data Logging: Saves workout data (keypoints, angles, reps, states) to a CSV file.
- Step 1: Ensure that you install app.py, the static folder, and the templats folder into one folder.
- Step 2: Run app.py in any IDE (VSCode, Pycharm, etc.)
- Step 3: Once you run it, you will see a URL pop open. Ensure you can click it by using CTRL + Click (PC) or Command + Click (Mac)
- The data generated consists of a CSV of 32 landmarks that corresponds to different points on the body
- Camera Feed: It should output the number of repetitions (reps) for a workout, angle measurements, and display feedback for every frame
- Post-Workout Feedback: A summary on the number of reps, % of form accuracy for the entire workout, and most common feedback