A Sparsified Densely Connected Network with Separable Convolution for Finger-Vein Recognition
At present, ResNet and DenseNet have achieved significant performance gains in the field of finger-vein biometric recognition, which is partially attributed to the dominant design of cross-layer skip connection. In this manner, features from multiple layers can be effectively aggregated to provide s...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-12-01
|
Series: | Symmetry |
Subjects: | |
Online Access: | https://www.mdpi.com/2073-8994/14/12/2686 |
_version_ | 1797455148115558400 |
---|---|
author | Qiong Yao Xiang Xu Wensheng Li |
author_facet | Qiong Yao Xiang Xu Wensheng Li |
author_sort | Qiong Yao |
collection | DOAJ |
description | At present, ResNet and DenseNet have achieved significant performance gains in the field of finger-vein biometric recognition, which is partially attributed to the dominant design of cross-layer skip connection. In this manner, features from multiple layers can be effectively aggregated to provide sufficient discriminant representation. Nevertheless, an over-dense connection pattern may induce channel expansion of feature maps and excessive memory consumption. To address these issues, we proposed a low memory overhead and fairly lightweight network architecture for finger-vein recognition. The core components of the proposed network are a sequence of sparsified densely connected blocks with symmetric structure. In each block, a novel connection cropping strategy is adopted to balance the channel ratio of input/output feature maps. Beyond this, to facilitate smaller model volume and faster convergence, we substitute the standard convolutional kernels with separable convolutional kernels and introduce a robust loss metric that is defined on the geodesic distance of angular space. Our proposed sparsified densely connected network with separable convolution (hereinafter dubbed ‘SC-SDCN’) has been tested on two benchmark finger-vein datasets, including the Multimedia Lab of Chonbuk National University (MMCBNU)and Finger Vein of Universiti Sains Malaysia (FV-USM), and the advantages of our SC-SDCN can be evident from the experimental results. Specifically, an equal error rate (EER) of 0.01% and an accuracy of 99.98% are obtained on the MMCBNU dataset, and an EER of 0.45% and an accuracy of 99.74% are obtained on the FV-USM dataset. |
first_indexed | 2024-03-09T15:47:22Z |
format | Article |
id | doaj.art-1102e619603c49c2805867e097ff6f6f |
institution | Directory Open Access Journal |
issn | 2073-8994 |
language | English |
last_indexed | 2024-03-09T15:47:22Z |
publishDate | 2022-12-01 |
publisher | MDPI AG |
record_format | Article |
series | Symmetry |
spelling | doaj.art-1102e619603c49c2805867e097ff6f6f2023-11-24T18:21:25ZengMDPI AGSymmetry2073-89942022-12-011412268610.3390/sym14122686A Sparsified Densely Connected Network with Separable Convolution for Finger-Vein RecognitionQiong Yao0Xiang Xu1Wensheng Li2Artificial Intelligence and Computer Vision Laboratory, University of Electronic Science and Technology of China Zhongshan Institute, Zhongshan 528400, ChinaArtificial Intelligence and Computer Vision Laboratory, University of Electronic Science and Technology of China Zhongshan Institute, Zhongshan 528400, ChinaArtificial Intelligence and Computer Vision Laboratory, University of Electronic Science and Technology of China Zhongshan Institute, Zhongshan 528400, ChinaAt present, ResNet and DenseNet have achieved significant performance gains in the field of finger-vein biometric recognition, which is partially attributed to the dominant design of cross-layer skip connection. In this manner, features from multiple layers can be effectively aggregated to provide sufficient discriminant representation. Nevertheless, an over-dense connection pattern may induce channel expansion of feature maps and excessive memory consumption. To address these issues, we proposed a low memory overhead and fairly lightweight network architecture for finger-vein recognition. The core components of the proposed network are a sequence of sparsified densely connected blocks with symmetric structure. In each block, a novel connection cropping strategy is adopted to balance the channel ratio of input/output feature maps. Beyond this, to facilitate smaller model volume and faster convergence, we substitute the standard convolutional kernels with separable convolutional kernels and introduce a robust loss metric that is defined on the geodesic distance of angular space. Our proposed sparsified densely connected network with separable convolution (hereinafter dubbed ‘SC-SDCN’) has been tested on two benchmark finger-vein datasets, including the Multimedia Lab of Chonbuk National University (MMCBNU)and Finger Vein of Universiti Sains Malaysia (FV-USM), and the advantages of our SC-SDCN can be evident from the experimental results. Specifically, an equal error rate (EER) of 0.01% and an accuracy of 99.98% are obtained on the MMCBNU dataset, and an EER of 0.45% and an accuracy of 99.74% are obtained on the FV-USM dataset.https://www.mdpi.com/2073-8994/14/12/2686finger-vein recognitiondensely connectedsparsifiedseparable convolution |
spellingShingle | Qiong Yao Xiang Xu Wensheng Li A Sparsified Densely Connected Network with Separable Convolution for Finger-Vein Recognition Symmetry finger-vein recognition densely connected sparsified separable convolution |
title | A Sparsified Densely Connected Network with Separable Convolution for Finger-Vein Recognition |
title_full | A Sparsified Densely Connected Network with Separable Convolution for Finger-Vein Recognition |
title_fullStr | A Sparsified Densely Connected Network with Separable Convolution for Finger-Vein Recognition |
title_full_unstemmed | A Sparsified Densely Connected Network with Separable Convolution for Finger-Vein Recognition |
title_short | A Sparsified Densely Connected Network with Separable Convolution for Finger-Vein Recognition |
title_sort | sparsified densely connected network with separable convolution for finger vein recognition |
topic | finger-vein recognition densely connected sparsified separable convolution |
url | https://www.mdpi.com/2073-8994/14/12/2686 |
work_keys_str_mv | AT qiongyao asparsifieddenselyconnectednetworkwithseparableconvolutionforfingerveinrecognition AT xiangxu asparsifieddenselyconnectednetworkwithseparableconvolutionforfingerveinrecognition AT wenshengli asparsifieddenselyconnectednetworkwithseparableconvolutionforfingerveinrecognition AT qiongyao sparsifieddenselyconnectednetworkwithseparableconvolutionforfingerveinrecognition AT xiangxu sparsifieddenselyconnectednetworkwithseparableconvolutionforfingerveinrecognition AT wenshengli sparsifieddenselyconnectednetworkwithseparableconvolutionforfingerveinrecognition |