TeoFil795/CrossfitAI

CrossFit Movement Analyzer AI 🏋️‍♂️

An AI-powered application that analyzes CrossFit movements using computer vision and machine learning techniques. Currently focused on Squat Clean analysis.

Features

• —Real-time pose detection using MediaPipe
• —Barbell tracking with YOLOv8
• —Movement phase analysis
• —Form feedback and corrections
• —Performance metrics
• —Video optimization for better analysis

Technical Stack

• —Computer Vision: MediaPipe Pose Detection, YOLOv8
• —Backend: Python, OpenCV
• —Frontend: Gradio UI
• —ML Framework: PyTorch

Usage

1. 1.Upload a video of a CrossFit athlete performing a Squat Clean
2. 2.Wait for the AI to analyze the movement
3. 3.Review the annotated video with pose and barbell tracking
4. 4.Check the detailed movement analysis report

Performance Notes

• —Videos are automatically optimized for analysis
• —Processes every other frame for better performance
• —Supports various video formats (MP4, MOV)
• —Maximum frame dimension: 640px (maintains aspect ratio)

Future Development

• —Support for additional CrossFit movements
• —Enhanced form feedback
• —Custom YOLOv8 training for better barbell detection
• —Real-time analysis capabilities

License

MIT License

Authors

• —@kogy9