Improving kinship verification using face age transformation

Abstract

Kinship verification, the task of determining family relationships based on facial images, has gained significant attention in recent years due to its applications in areas such as social networks, forensics, and genealogy. However, accurately verifying kinship remains a challenging problem, particularly when accounting for variations in age between family members. This thesis explores the potential of using Face Age Transformation techniques, specifically through Generative Adversarial Networks (GANs), to improve the accuracy of kinship verification models. This work involves developing a GAN-based age transformation model that can simulate the aging process in facial images. By augmenting kinship datasets with these age-transformed images, we aim to enhance the robustness and reliability of kinship verification systems. Experimental results indicate that incorporating age-transformed facial images into the kinship verification process leads to a more accurate representation of familial relationships, particularly in cases where age differences are pronounced. This work contributes to the growing field of kinship verification by elaborating a novel approach that leverages the power of GANs for age progression, offering a promising direction for future research and practical applications.