DS-CNN: A pre-trained Xception model based on depth-wise separable convolutional neural network for finger vein recognition

作者：

Highlights：

• A pre-trained model based on a depth-wise separable convolution layer is employed.

• The proposed model has low computation cost and better generalizability.

• A comparative analysis of the various learning models has been presented.

• The results are promising and show excellent authentication accuracy on a small dataset.

摘要

•A pre-trained model based on a depth-wise separable convolution layer is employed.•The proposed model has low computation cost and better generalizability.•A comparative analysis of the various learning models has been presented.•The results are promising and show excellent authentication accuracy on a small dataset.

论文关键词：Biometric,Convolutional neural network,Classification,Deep learning,Finger vein recognition,Transfer learning,Xception

论文评审过程：Received 3 April 2021, Revised 29 October 2021, Accepted 22 November 2021, Available online 2 December 2021, Version of Record 7 December 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.116288