Skip to main content

Automatic Facial Landmark Annotation of Facial Images

Researcher: Marios Savvides

Research Area: Security of Cyber-Physical Systems

Abstract

Scope: Sponsored by the FBI, the aim of this project is to develop automatic algorithms that will robustly identify about 79 fiducial points on the face for forensic analysis. Current systems are not reliable and require significant manual markup by a human operator; this project aims at building robust models that will generalize to unseen real world data and fit robustly to these points. This project will also provide the ability to do other facial forensic comparison in the future (such as shape analysis for face matching or detecting other uniquely identifiable features).

Outcomes: The algorithms developed here will be integrated into the FBI's Universal Face Workstation, which is the facial forensic analysis system. It will also be integrated into a Government Owned Facial Recognition system.