MediaPipe Hand, Pose, and Face Detection

This project demonstrates the usage of MediaPipe for hand, pose, and face detection in videos.

Language Stats:

• Python: 39660 bytes
• HTML: 2905 bytes
• Jupyter Notebook: 691 bytes

Total Size Of Project : 13433 kilobytes

Total Lines Of Code : 1302

End of Language Stats

Features

• Detects hand, pose, and face landmarks using MediaPipe.
• Displays the detected landmarks overlaid on the video frames.
• Supports resizing frames to fit the screen while maintaining aspect ratio.
• Displays the name of the video file as a text overlay on the frames.

Requirements

• Python 3.x
• OpenCV
• MediaPipe

Installation

1. Clone this repository:

git clone https://github.com/cagatay-softgineer/MediaPipe.git

2. Install the required libraries:

pip install opencv-python mediapipe websockets

Usage

1. Place your video files in the videos directory.
2. Update the Video Path in main.py.

useMediaPipe("videos/Test.mp4")

3. Run the main.py script:

python main.py

4. Press 'q' to exit the program.

WebSocket Usage

1. Start the WebSocket Server:

• Navigate to the WebSocketServer folder.

FOR WINDOWS

• Run the executable websoc.exe file to start the WebSocket server.

FOR MACOS OR LINUX

• Create a script for run websocket.py

python3 websoc.py # Use python if not working with python3

• Save the file with a .sh extension, for example run_websoc.sh
• Make the Script

• Open Terminal.
• Navigate to the directory where you saved run_websoc.sh.
• Make the script executable by running the following command

chmod +x run_websoc.sh

• Run the Script

./run_websoc.sh

2. Add Send2WSS=True to:

useMediaPipe("videos/Test.mp4",Send2WSS=True)

Customization

• You can customize the screen width and height in the useMediaPipe function to adjust the size of the displayed frames.
• Additional customization can be done by modifying the code in modelUsageTests.py according to your requirements.

License

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

Acknowledgments

This project utilizes the following third-party applications and libraries

- MediaPipe library developed by Google.

- Blender