Generative Face Models for Video Chat and Broadcast

In today’s world, the video chat and conferencing has been a necessary part of overall routine life (both personal and work). Currently, the video chat and broadcast technologies require lots of bandwidth to transmit high quality data. In overall project, we aim to make video transmission data intelligent by using the state-of-the-art AI technology. This will not only reduce the size of video data but will also include the ‘instructions’ to give users creative control of their representations and serve as a segue to the metaverse – the intelligent part.
In this part of the project, we are planning to utilize the state-of-the-art face generating StyleGan2 neural network that convert the face video data into the compressed, intelligent, and ready-for-transmission data format. At the receiving end, the transmitted instructions will be used to regenerate the transmitted video data. The additional instructions to manipulate the video will provide the useful mechanism to generate the avatars with realistic faces – a must have feature for upcoming metaverse.

Faculty Supervisor:

Abdul Bais

Student:

Partner:

Cya Live

Discipline:

Engineering

Sector:

Artificial Intelligence; Information and Communications Technology; Entertainment and Media

University:

University of Regina

Program: