A prototype-based framework for open-set heterogeneous federated face recognition

T.H. Kim

doi:10.1016/j.eswa.2025.130020

A prototype-based framework for open-set heterogeneous federated face recognition

2025-10-15

0

PRE

AI

Andrew Beng Jin Teoh *

DOI：10.1016/j.eswa.2025.130020

Original

Original request for help

Save

Abstract

En 中文

• Prototype-based federated face recognition for unseen new identities. • Shares identity prototypes without averaging to preserve class meaning. • Contrastive prototype-anchor loss boosts separation of face identities. • Works across diverse client model architectures and device conditions. • Improves strict low false accept verification while protecting privacy.

A prototype-based framework for open-set heterogeneous federated face recognition

Abstract

Journal

Organization