Scalable Angular Discriminative Deep Metric Learning for Face Recognition
With the development of deep learning, Deep Metric Learning (DML) has achieved great improvements in face recognition. Specifically, the widely used softmax loss in the training process often bring large intra-class variations, and feature normalization is only exploited in the testing process to co...
Main Authors: | , , |
---|---|
Format: | Report |
Language: | unknown |
Published: |
arXiv
2018
|
Subjects: | |
Online Access: | https://dx.doi.org/10.48550/arxiv.1804.10899 https://arxiv.org/abs/1804.10899 |
id |
ftdatacite:10.48550/arxiv.1804.10899 |
---|---|
record_format |
openpolar |
spelling |
ftdatacite:10.48550/arxiv.1804.10899 2023-05-15T16:01:40+02:00 Scalable Angular Discriminative Deep Metric Learning for Face Recognition Wu, Bowen Wu, Huaming Zhang, Monica M. Y. 2018 https://dx.doi.org/10.48550/arxiv.1804.10899 https://arxiv.org/abs/1804.10899 unknown arXiv arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ Computer Vision and Pattern Recognition cs.CV Artificial Intelligence cs.AI FOS Computer and information sciences Preprint Article article CreativeWork 2018 ftdatacite https://doi.org/10.48550/arxiv.1804.10899 2022-04-01T09:51:42Z With the development of deep learning, Deep Metric Learning (DML) has achieved great improvements in face recognition. Specifically, the widely used softmax loss in the training process often bring large intra-class variations, and feature normalization is only exploited in the testing process to compute the pair similarities. To bridge the gap, we impose the intra-class cosine similarity between the features and weight vectors in softmax loss larger than a margin in the training step, and extend it from four aspects. First, we explore the effect of a hard sample mining strategy. To alleviate the human labor of adjusting the margin hyper-parameter, a self-adaptive margin updating strategy is proposed. Then, a normalized version is given to take full advantage of the cosine similarity constraint. Furthermore, we enhance the former constraint to force the intra-class cosine similarity larger than the mean inter-class cosine similarity with a margin in the exponential feature projection space. Extensive experiments on Labeled Face in the Wild (LFW), Youtube Faces (YTF) and IARPA Janus Benchmark A (IJB-A) datasets demonstrate that the proposed methods outperform the mainstream DML methods and approach the state-of-the-art performance. Report DML DataCite Metadata Store (German National Library of Science and Technology) Janus ENVELOPE(163.100,163.100,-71.067,-71.067) |
institution |
Open Polar |
collection |
DataCite Metadata Store (German National Library of Science and Technology) |
op_collection_id |
ftdatacite |
language |
unknown |
topic |
Computer Vision and Pattern Recognition cs.CV Artificial Intelligence cs.AI FOS Computer and information sciences |
spellingShingle |
Computer Vision and Pattern Recognition cs.CV Artificial Intelligence cs.AI FOS Computer and information sciences Wu, Bowen Wu, Huaming Zhang, Monica M. Y. Scalable Angular Discriminative Deep Metric Learning for Face Recognition |
topic_facet |
Computer Vision and Pattern Recognition cs.CV Artificial Intelligence cs.AI FOS Computer and information sciences |
description |
With the development of deep learning, Deep Metric Learning (DML) has achieved great improvements in face recognition. Specifically, the widely used softmax loss in the training process often bring large intra-class variations, and feature normalization is only exploited in the testing process to compute the pair similarities. To bridge the gap, we impose the intra-class cosine similarity between the features and weight vectors in softmax loss larger than a margin in the training step, and extend it from four aspects. First, we explore the effect of a hard sample mining strategy. To alleviate the human labor of adjusting the margin hyper-parameter, a self-adaptive margin updating strategy is proposed. Then, a normalized version is given to take full advantage of the cosine similarity constraint. Furthermore, we enhance the former constraint to force the intra-class cosine similarity larger than the mean inter-class cosine similarity with a margin in the exponential feature projection space. Extensive experiments on Labeled Face in the Wild (LFW), Youtube Faces (YTF) and IARPA Janus Benchmark A (IJB-A) datasets demonstrate that the proposed methods outperform the mainstream DML methods and approach the state-of-the-art performance. |
format |
Report |
author |
Wu, Bowen Wu, Huaming Zhang, Monica M. Y. |
author_facet |
Wu, Bowen Wu, Huaming Zhang, Monica M. Y. |
author_sort |
Wu, Bowen |
title |
Scalable Angular Discriminative Deep Metric Learning for Face Recognition |
title_short |
Scalable Angular Discriminative Deep Metric Learning for Face Recognition |
title_full |
Scalable Angular Discriminative Deep Metric Learning for Face Recognition |
title_fullStr |
Scalable Angular Discriminative Deep Metric Learning for Face Recognition |
title_full_unstemmed |
Scalable Angular Discriminative Deep Metric Learning for Face Recognition |
title_sort |
scalable angular discriminative deep metric learning for face recognition |
publisher |
arXiv |
publishDate |
2018 |
url |
https://dx.doi.org/10.48550/arxiv.1804.10899 https://arxiv.org/abs/1804.10899 |
long_lat |
ENVELOPE(163.100,163.100,-71.067,-71.067) |
geographic |
Janus |
geographic_facet |
Janus |
genre |
DML |
genre_facet |
DML |
op_rights |
arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
op_doi |
https://doi.org/10.48550/arxiv.1804.10899 |
_version_ |
1766397436385820672 |