Identify and manipulate faces in a web browser.
The facial recognition software uses a combination of OpenCV, DLib, and some custom C++ code all compiled into web assembly. The front end utilizes Javascript's new found love for primitive data types to pass pointers to and from native machine code to modify images.
Our sister project Celebraphone uses this for p2p video chatting in a browser. You can find that project here.
The project also uses a base image for compilation that you can find the source for here and pull from here.
To run locally, make sure you have docker up and type docker-compose up --build
.
This project was made possible by some of the people below.