This is an experimental project testing hand gesture recognition in the browser, using a simple webcam. It uses two Tensorflow JS projects (Hand Pose Detection and MoveNet) which detect and estimate poses of hands and bodies, to then track the hand positions and recognize static hand gestures.
WARNING! This project is a WORK IN PROGRESS and experimental, everything is subject to change! Use at your own risk!
This project is based on Angular and uses the Angular CLI and the Yarn package Manager.
To install them, you need a recent version of Node.js.
Run the following commands to setup and run the project:
(On Linux, you might need to use sudo to install things globally)
# If node or npm versions are not up to date:
npm install --global n
n lts
# To install angular cli (ng) and yarn:
npm install --global @angular/cli yarn
# To download all project dependencies:
cd mirror-hgr # navigate to your cloned directory
yarn
# To run a dev server:
ng serve --open