Every day, video conferencing apps connect individuals all over the world. Eye contact is a key element for establishing and maintaining a genuine connection between individuals, but this is largely lost when using a webcam. To solve this problem, we developed an open-source and extremely lightweight gaze correction model that uses machine learning to make the user's eyes appear to be directed at their web camera, creating the illusion of eye-to-eye contact. It can run without a dedicated GPU. To showcase the model, we created a web conferencing app centered around a user's eye gaze.

Team Members:

Rick Hanish

Lisong Sun

Rayan Tejani

Jonathan Vasilyev

Daniel Yamakov