Research summary
This project seeks to improve the experience of trying products out virtually using augmented reality.
Virtual try-on for faces is popular in gaming and cosmetics, but can also be applied in medical practices such as facial reconstruction surgery and medicinal skin products.
We are developing a new technique that can fill in missing parts of the facial image through inpainting in real-time to create a more realistic appearance. Our solution uses machine intelligence to integrate facial wrinkling information that can be used on a mobile platform with a quicker processing time than existing applications.
Working alongside face-tracking and AR specialists Image Metrics, we are testing our method in real-world applications.
This research will investigate deep learning architectures for various inpainting tasks.