KinshipGAN : generating kin images

Abstract

Automatic kinship verifi cation using face images involves analyzing features and computing similarities between two input images to establish kin-relationship. It has gained signi cant interest from the research community and several approaches including deep learning architectures are proposed. One of the law enforcement applications of kinship analysis involves predicting the kin image given an input image. In other words, the question posed here is: \given an input image, is it possible to generate a kin-image?" This paper, for the rst time in the literature attempts to generate kin-images using Generative Adversarial Learning. The proposed Kinship GAN model incorporates three information, kin-gender, kinship loss, reconstruction loss, in a GAN model and generates kin images. On the WVU Kinship Video database, the proposed model shows very promising results in generating kin images. Kinship veri cation accuracy is used as an evaluation metric and the results show 70% accuracy.