Topic: Deep Face Age Progression: An Introduction
by Marcel Grimmer
online Big Blue Button Room: D19/2.03a, December 10, 2020 (Thursday), 12.00 noon
Keywords — Face Aging, Deep Learning, Generative Adversarial Networks
Face recognition is becoming increasingly popular today and is used, for example, in large-scale projects such as the European “Entry/Exit System” or the “Schengen Information System”. At the same time, it is important to develop face recognition models that are robust to intrapersonal face variations caused by various factors, such as facial expression, head poses, or aging effect. In this context, collecting images of the same person over time turns out to be particularly time and cost intensive. Therefore, face age progression (FAP) methods are used to automatically synthesize faces with aging effects in order to create cross-age training datasets. Specifically, deep generative networks have proven its capability to generate photorealistic aging effects, while preserving identity-related information. In this talk, the main concepts of deep generative FAP methods will be presented with an overview of common evaluation techniques and open challenges.