DiverAge: Reliable Pluralistic Face Aging with Cross-Age Identity Relation Guidance
Researchers have developed DiverAge, a new framework for generating realistic and diverse face aging sequences. This method uses diffusion autoencoding to create varied appearances within a single age group while ensuring identity consistency across an aging sequence. DiverAge incorporates a Cross-age Identity Relation Regulator (CARR) to maintain identity preservation and ordinal reliability, addressing limitations of previous deterministic and pluralistic aging techniques. AI
IMPACT Enhances realism and diversity in AI-driven face aging applications for biometrics and forensics.