Speaker: Dr. Una Kelly 

Abstract: Face recognition plays an important role in modern society. We depend on it when we travel through airports that use automated border control systems (eGates), unlock our phones with it, and police use it to find criminals in their databases. However, face recognition systems are vulnerable to morphing attacks, which are combined images of two people that can trick systems into falsely matching both identities. If such images are used to apply for ID documents, this poses security risks. In this talk, I will present my research on analysing feature spaces of face recognition systems to understand their weaknesses.  So-called "worst-case morphs" can be used to measure theoretical vulnerabilities and help to explain why the most accurate face recognition systems are paradoxically the most vulnerable to morphing, which can be addressed by introducing smarter decision thresholds. Overall, the aim is to make face recognition systems more robust to both known and future morphing attacks by addressing fundamental flaws.

Short bio: Una Kelly is a postdoctoral researcher at the chair of Machine Learning and Data Engineering at the University of Münster. She completed her PhD degree at the chair of Data Management and Biometrics at the University of Twente. Her research focuses on developing computer vision models for biometrics and for earth observation.