Disentangling Racial Phenotypes: Fine-Grained Control of Race-related Facial Phenotype Characteristics

arXiv:2403.19897v1 Announce Type: cross
Abstract: Achieving an effective fine-grained appearance variation over 2D facial images, whilst preserving facial identity, is a challenging task due to the high complexity and entanglement of common 2D facial feature encoding spaces. Despite these challenges, such fine-grained control, by way of disentanglement is a crucial enabler for data-driven racial bias mitigation strategies across multiple automated facial analysis tasks, as it allows to analyse, characterise and synthesise human facial diversity. In this paper, we propose a …

abstract arxiv challenges complexity control cs.cv cs.lg encoding entanglement feature fine-grained identity images race racial spaces type variation