Multiview 4D facial keypoints tracking

The accurate localization of facial keypoints in static images, and their tracking in video has many potential applications. In this project we will address the problem of fine-grained facial keypoint tracking using multiple images and cameras, possibly including depth cameras. We have identified three potential partners, each of whom focus on applications that require the accurate localization and/or tracking of facial keypoints. In one potential application we are interested in detecting certain congenital syndromes using detailed facial analysis. In another scenario such features could aide in the detection of driver drowsiness, a major cause of road accidents. In yet another example, detailed tracking of facial keypoints can be used for high quality computer animation.

Md. Kamrul Hasan
Faculty Supervisor: 
Christopher Pal
Project Year: