Attention-aware ensemble learning for face-periocular cross-modality matching

Open Webpage

Tiong-Sik Ng, Andrew Beng Jin Teoh

Published: 01 Jan 2025, Last Modified: 11 Nov 2025Appl. Soft Comput. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•AELNet: An attention-aware ensemble for face-periocular cross-modality matching.•AELNet uses attention-aware encoder and multiple heads to enhance joint embeddings.•AELNet enhances FPCM by diversifying classifier heads via embeddings and sampling.•AELNet proves robust through experiments on five unconstrained face datasets.