Cycle face aging generative adversarial networks

Abstract

The facial features of human changes with age. It is important to model the human face for cross-age verification and recognition. In this paper, we introduce a Cycle Face Aging Generative Adversarial Network (CFA-GANs) framework which preserves original face identity in the aged version of his/her face. Due to the shortage of paired data of human faces, we used CycleConsistent Generative Adversarial Network (CycleGANs) which transform an image from source domain X to target domain Y in absence of paired example. Our aim is to translate an input age group image into target age group image for face aging problems. Train on the various images, we demonstrate that our CFA-GAN learns and transfer the features of the face from the input group to target group. Results have been taken on UTKFace database to obtain aged and regenerated face images.