Finger-Vein Recognition Using a NASNet with a Cutout

Abstract

Traditional information security systems use passwords or identification cards that might be deciphered or stolen. Many methods have been developed to improve the security of personal information, such as the finger-vein recognition to replace traditional recognition. This study proposes a cutout for data augmentation (DA) and a neural architecture search network (NASNet). Experiments show that the proposed method is 98.89% accurate for the FV-USM public dataset.