AI Fitness Assistant Based on RTMPose

English | 中文

• 2025-06-07: Major update! Dropped YOLO models and all GPU support. Now uses only RTMPose for pose detection, and runs on CPU only. Simpler, more compatible, and easier to use.
• 2025-06-12：Optimize exercise_counters.py for counting accuracy, code structure optimization
• 2025-11-14: Reverted asynchronous pose detection due to accuracy issues, restored synchronous pose detection. Fixed crash when switching from statistics mode back to detection mode.
• 2025-11-15: New exercise database feature! All exercise configurations are now managed in data/exercises.json file. You can easily add, modify, or remove exercise types without modifying code.

• Multi-language interface
• Improve pose detection accuracy
• Add support for more exercise types
• Add custom exercise types template
• Recognizing Motion Accuracy
• Motion Error Correction Indication
• Add voice feedback
• Mobile Application Support

• Real-time Exercise Counting - Automatically counts your repetitions
• Multiple Exercise Support - Including squats, push-ups, sit-ups, bicep curls, and many more
• Advanced Pose Detection - Powered by RTMPose for accurate tracking
• CPU Only - No GPU required, works on most computers
• Visual Feedback - Live skeleton visualization with angle measurements
• Workout Statistics - Track your progress over time
• User-friendly Interface - Clean PyQt5 GUI with intuitive controls
• Works with any webcam - No special hardware required
• Runs locally - Complete privacy

• If you don't want to set up a Python environment, you can download our pre-packaged executable:

  Windows EXE package:

  Baidu Netdisk Link code: 8866

  Google Drive

• Use the interface buttons to select different exercises
• Real-time feedback shows your current form and repetition count
• Press the "Reset" button to reset the counter
• Use manual adjustment buttons to correct the count if needed
• Toggle skeleton visualization on/off
• View your workout statistics over time

All exercise types are now stored in the data/exercises.json file. You can easily add, modify, or remove exercise types without modifying code!

1. Keypoint Index Reference

   • The system uses COCO 17 keypoint format:
     • 0: nose, 1: left_eye, 2: right_eye, 3: left_ear, 4: right_ear
     • 5: left_shoulder, 6: right_shoulder, 7: left_elbow, 8: right_elbow
     • 9: left_wrist, 10: right_wrist, 11: left_hip, 12: right_hip
     • 13: left_knee, 14: right_knee, 15: left_ankle, 16: right_ankle

2. Configuration Parameters

   • down_angle: Angle threshold when lowering (degrees)
   • up_angle: Angle threshold when raising (degrees)
   • keypoints.left: Left side three keypoint indices [pt1, pt2, pt3] for angle calculation
   • keypoints.right: Right side three keypoint indices [pt1, pt2, pt3] for angle calculation
   • is_leg_exercise: Whether it's a leg exercise (true/false), affects counting logic
   • angle_point: Keypoint indices [pt1, pt2, pt3] for displaying angle lines on video

3. Example: Adding a New Exercise

   "my_custom_exercise": {
     "name_zh": "我的自定义运动",
     "name_en": "My Custom Exercise",
     "down_angle": 120,
     "up_angle": 170,
     "keypoints": {
       "left": [5, 7, 9],
       "right": [6, 8, 10]
     },
     "is_leg_exercise": false,
     "angle_point": [6, 8, 10]
   }

4. Restart the Application

• Python 3.9
• Webcam
• Windows/Mac/Linux: CPU only, no GPU required. Performance may vary by hardware.

1. Clone and install

   git clone https://github.com/yo-WASSUP/Good-GYM.git
   cd Good-GYM
   
   # Create virtual environment
   python -m venv venv
   # Activate (Windows)
   .\venv\Scripts\activate
   # or (Mac/Linux)
   source venv/bin/activate
   
   # Install dependencies
   pip install -r requirements.txt

2. Run the application

   python workout_qt_modular.py

Contributions are welcome! Please feel free to submit a Pull Request.

Thanks to RTMPose open source pose detection model: https://github.com/Tau-J/rtmlib

This project is licensed under the MIT License - see the LICENSE file for details.