TY - GEN
T1 - DBLFace
T2 - 2020 IEEE/IAPR International Joint Conference on Biometrics, IJCB 2020
AU - Le, Ha A.
AU - Kakadiaris, Ioannis A.
N1 - Publisher Copyright:
© 2020 IEEE.
PY - 2020/9/28
Y1 - 2020/9/28
N2 - Deep learning-based domain-invariant feature learning methods are advancing in near-infrared and visible (NIR-VIS) heterogeneous face recognition. However, these methods are prone to overfitting due to the large intra-class variation and the lack of NIR images for training. In this paper, we introduce Domain-Based Label Face (DBLFace), a learning approach based on the assumption that a subject is not represented by a single label but by a set of labels. Each label represents images of a specific domain. In particular, a set of two labels per subject, one for the NIR images and one for the VIS images, are used for training a NIR-VIS face recognition model. The classification of images into different domains reduces the intra-class variation and lessens the negative impact of data imbalance in training. To train a network with sets of labels, we introduce a domain-based angular margin loss and a maximum angular loss to maintain the inter-class discrepancy and to enforce the close relationship of labels in a set. Quantitative experiments confirm that DBLFace significantly improves the rank-1 identification rate by 6.7% on the EDGE20 dataset and achieves state-of-the-art performance on the CASIA NIR-VIS 2.0 dataset.
AB - Deep learning-based domain-invariant feature learning methods are advancing in near-infrared and visible (NIR-VIS) heterogeneous face recognition. However, these methods are prone to overfitting due to the large intra-class variation and the lack of NIR images for training. In this paper, we introduce Domain-Based Label Face (DBLFace), a learning approach based on the assumption that a subject is not represented by a single label but by a set of labels. Each label represents images of a specific domain. In particular, a set of two labels per subject, one for the NIR images and one for the VIS images, are used for training a NIR-VIS face recognition model. The classification of images into different domains reduces the intra-class variation and lessens the negative impact of data imbalance in training. To train a network with sets of labels, we introduce a domain-based angular margin loss and a maximum angular loss to maintain the inter-class discrepancy and to enforce the close relationship of labels in a set. Quantitative experiments confirm that DBLFace significantly improves the rank-1 identification rate by 6.7% on the EDGE20 dataset and achieves state-of-the-art performance on the CASIA NIR-VIS 2.0 dataset.
UR - http://www.scopus.com/inward/record.url?scp=85099684648&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85099684648&partnerID=8YFLogxK
U2 - 10.1109/IJCB48548.2020.9304884
DO - 10.1109/IJCB48548.2020.9304884
M3 - Conference contribution
AN - SCOPUS:85099684648
T3 - IJCB 2020 - IEEE/IAPR International Joint Conference on Biometrics
BT - IJCB 2020 - IEEE/IAPR International Joint Conference on Biometrics
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 28 September 2020 through 1 October 2020
ER -