Hand Tracker

This is a python script using OpenCV and Google's MediaPipe to track hand signs.

Installation

Use the package manager pip to install OpenCV and MediaPipe.

pip install -r requirements.txt

Usage

Using mediapipe to label landmarks with an id can be used to compare others, thus allowing you to find patterns within a hand.

There is also a function called "distance" that compares 2 landmark ids and finds a distance value between them.

distance(5,8) #compares distance from landmark 5 to landmark 8

Using the distance between 2 landmarks you can see for example, if 2 fingers are touching. This can be expanded upon by using it to check for hand signs.

#see if pointer finger and thumb are touching, thus making an OK sign
if distance(8, 4) < 40 + scalar and distance(8, 12) > 50 + scalar and distance(12, 16) < 60 + scalar: 

#check if all fingers are touching palm, thus making a fist
if distance(8,5)<30+scalar and distance(12,9)<30+scalar and distance(16,13)<30+scalar and distance(20,17)<30+scalar:

You can use these checks to write over the image with OpenCV's cv2.putText function. This can be used to display text when a hand sign is detected.

Extra

You may notice a variable named "scalar" being used. This is used in addition with the other file, named baseline- which creates a baseline image of your hand to more accurately scale at distance.

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

License

The Unlicense