HandRecognition

Hand recognition (soon to connect with an audioplayer) using mediapipe, opencv, and tensorflow

Setup

• Download zip file from FinalTouchUp
• Download python 3.10.x
  • Anything more or less will break the code
  • verify with python --version
• Open terminal and create virtual environment
  • python -m venv desired_name_of_virtual_env
• activate venv with
  • source venvName/bin/activate for linux/mac
  • venvName\Scripts\activate for Windows
• Download modules for mediapipe, opencv, and tensorflow
  • pip install mediapipe opencv-python numpy tensorflow pyqt5 keyboard pycaw

How to use

• run app.py
• Edit Commands will allow you to edit commands to your liking, as long as it is a valid command
• Reset will reset all commands to its default dataset labels (like, palm, fist, etc)
• Start recognition will open the camera and start hand recognition
  • Open Spotify first before pressing this button

Limitations

• Made without Spotify API, thus commands are severely limited
• Mapping of gesture to commands are through global key commands and therefore, if other media players are opened, it may not directly communicate with Spotify
• Allows cross-compatibility but may experience performance overhead in Start Recognition (slow startup ~20 seconds), especially in Windows