Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

My Eyes Are Up Here: Promoting Focus on Uncovered Regions in Masked Face Recognition

 
: Neto, Pedro C.; Boutros, Fadi; Pinto, Joao Ribeiro; Saffari, Mohsen; Damer, Naser; Sequeira, Ana F.; Cardoso, Jaime S.

:

Brömme, Arslan (Editor); Busch, Christoph (Editor); Damer, Naser; Dantcheva, Antitza; Gomez-Barrero, M.; Raja, K.; Rathgeb, Christian; Sequeira, A.F.; Uhl, Andreas ; Gesellschaft für Informatik -GI-, Bonn:
BIOSIG 2021, 20th International Conference of the Biometrics Special Interest Group. Proceedings : 15.-17. September 2021, International Digital Conference
Bonn: GI, 2021 (GI-Edition - Lecture Notes in Informatics (LNI). Proceedings P-315)
ISBN: 978-3-88579-709-8
DOI: 10.1109/BIOSIG52210.2021
5 S.
Gesellschaft für Informatik, Special Interest Group on Biometrics and Electronic Signatures (BIOSIG International Conference) <11, 2012, Darmstadt>
Bundesministerium für Bildung und Forschung BMBF (Deutschland)
SFRH/BD/137720/2018; ATHENE
Englisch
Konferenzbeitrag
Fraunhofer IGD ()
Lead Topic: Digitized Work; Lead Topic: Smart City; Lead Topic: Visual Computing as a Service; Research Line: Computer vision (CV); Research Line: Machine Learning (ML); biometrics; face recognition; deep learning; CRISP; ATHENE

Abstract
The recent Covid-19 pandemic and the fact that wearing masks in public is now mandatory in several countries, created challenges in the use of face recognition systems (FRS). In this work, we address the challenge of masked face recognition (MFR) and focus on evaluating the verification performance in FRS when verifying masked vs unmasked faces compared to verifying only unmasked faces. We propose a methodology that combines the traditional triplet loss and the mean squared error (MSE) intending to improve the robustness of an MFR system in the masked-unmasked comparison mode. The results obtained by our proposed method show improvements in a detailed step-wise ablation study. The conducted study showed significant performance gains induced by our proposed training paradigm and modified triplet loss on two evaluation databases.

: http://publica.fraunhofer.de/dokumente/N-640709.html