NOTICE: this project has been replaced by Tracking4All which is actively supported and features much better quality, is cross-platform with embedded Unity support and more!
Testing hand tracking inside of Unity using Google MediaPipe Hands Python framework. Webcam readings, piping, and MediaPipe Hands all run on a different thread.
- Requires Python, Unity Hub, a WebCam, and decently fast CPU.
- pip install mediapipe
- Clone this repository.
- Run the Unity project FIRST which acts as a server.
- Next run main.py which actually runs MediaPipe Hands.
- Go back to the running Unity project to see your hands inside of the game view.
- You can set the DEBUG flag True in hands.py to visualize what is being seen and how your hands are being interpreted.
- Improve the accuracy of the model by setting MODEL_COMPLEXITY to 1 inside hands.py