ChildFace: Gender Aware Child Face Aging

Abstract

Child face aging and rejuvenation has amassed considerable active research interest due to its immense impact on monitoring applications especially for finding lost/abducted children with childhood photos and hence protect children. Prior studies are primarily motivated to enhance the generation quality and aging of face images, rather than quantifying face recognition performance. To address this challenge we propose ChildFace model. Our model does child face aging and rejuvenation while using gender as condition. Our model uses Conditional Generative Adversarial Nets (cGANs), VGG19 based perceptual loss and LightCNN29 age classifier and produces impressive results. Intense quantitative study based on verification, identification and age estimation proves that our model is competent to existing state-of-art models and can make a significant contribution in identifying missing children.