Finger-Vein Recognition Using ResNet with Elastic Margin Loss

As technology evolves, people are starting to value the safety of their personal information more than ever. However, traditional keys and passwords as security features are subject to vulnerabilities, prompting a flux of new methods of biometric recognition with deep learning (DL). We proposed the RestNet model and elastic margin loss to perform a finger-vein recognition system. Experimental results indicated that the proposed DL architecture showed that precision, recall, F1-score, and accuracy metrics were higher than 99% in the FV-USM dataset.