Holistic Pose Detection

Description

This project uses MediaPipe's Holistic model to detect and visualize human pose, face mesh, and hand landmarks in real-time from a webcam feed.

Features

• Real-time pose detection: Tracks body pose landmarks
• Face mesh detection: Detects facial contours and landmarks
• Hand tracking: Tracks hand landmarks (when visible)
• Live visualization: Displays detected landmarks overlaid on the video feed

Requirements

• Python 3.7+
• OpenCV (cv2)
• MediaPipe (mediapipe)

Installation

Install the required dependencies:

pip install opencv-python mediapipe

Usage

Run the main script:

python main.py

• The application will open a window showing the webcam feed with detected pose and face landmarks
• Press q to quit the application

Configuration

The holistic model is configured with:

• min_detection_confidence=0.5: Minimum confidence for initial detection
• min_tracking_confidence=0.5: Minimum confidence for tracking landmarks

You can adjust these values in main.py for different sensitivity levels.

Notes

• The script uses camera index 1 by default. If your camera doesn't work, try changing cv2.VideoCapture(1) to cv2.VideoCapture(0) in main.py
• The display window shows a flipped (mirrored) view of the camera feed for a more natural experience

License

This project is open source and available for educational purposes.