Multi-Level Metric Learning Network for Fine-Grained Classification
The application of fine-grained image classification can be problematic due to subtle differences between classes. The existing global feature-based methods have worse accuracies than regional feature-based methods, because regional feature-based methods focus on the determination of differentiated...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2019-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/8903259/ |
_version_ | 1818349414822969344 |
---|---|
author | Jiabao Wang Yang Li Zhuang Miao Xun Zhao Zhang Rui |
author_facet | Jiabao Wang Yang Li Zhuang Miao Xun Zhao Zhang Rui |
author_sort | Jiabao Wang |
collection | DOAJ |
description | The application of fine-grained image classification can be problematic due to subtle differences between classes. The existing global feature-based methods have worse accuracies than regional feature-based methods, because regional feature-based methods focus on the determination of differentiated features within local regions. To learn more discriminative global features, in this paper, we proposed the use of L2 normalization to tackle a neglected conflict between the widely used metric loss (triplet loss) and classification loss (softmax loss) in global feature-based methods. Furthermore, a multi-level metric learning network (MMLN) is proposed for fine-grained image classification based on global features. In the MMLN, multi-level metric learning objectives and classification objectives are present at multiple high-level layers. The multi-level metric learning objectives work together to supervise the network in order to learn highly discriminative features. In addition, a new probability aggregation strategy (PAS) is proposed to produce a fused prediction by combining the multi-level predictive probabilities. Experiments were conducted on three standard fine-grained classification datasets (CUB-200-2011, Stanford Cars, and FGVC-Aircraft). Results demonstrated that our MMLN achieved accuracies of 88.0%, 94.6% and 92.4% respectively and outperformed state-of-the-art methods, substantially improving fine-grained classification tasks. Besides, gradient-weighted class activation mapping (Grad-CAM) shows that the MMLN is able to pay more attention to the discriminative local regions due to the application of multi-level metric learning. |
first_indexed | 2024-12-13T18:05:35Z |
format | Article |
id | doaj.art-aee958a1cf1c4867810ccbf49bc9e710 |
institution | Directory Open Access Journal |
issn | 2169-3536 |
language | English |
last_indexed | 2024-12-13T18:05:35Z |
publishDate | 2019-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Access |
spelling | doaj.art-aee958a1cf1c4867810ccbf49bc9e7102022-12-21T23:36:05ZengIEEEIEEE Access2169-35362019-01-01716639016639710.1109/ACCESS.2019.29539578903259Multi-Level Metric Learning Network for Fine-Grained ClassificationJiabao Wang0https://orcid.org/0000-0002-3706-9912Yang Li1https://orcid.org/0000-0003-1682-0284Zhuang Miao2https://orcid.org/0000-0003-2289-4589Xun Zhao3https://orcid.org/0000-0002-1767-6520Zhang Rui4https://orcid.org/0000-0002-3214-5330Command and Control Engineering College, Army Engineering University of PLA, Nanjing, ChinaCommand and Control Engineering College, Army Engineering University of PLA, Nanjing, ChinaCommand and Control Engineering College, Army Engineering University of PLA, Nanjing, ChinaCommand and Control Engineering College, Army Engineering University of PLA, Nanjing, ChinaCommand and Control Engineering College, Army Engineering University of PLA, Nanjing, ChinaThe application of fine-grained image classification can be problematic due to subtle differences between classes. The existing global feature-based methods have worse accuracies than regional feature-based methods, because regional feature-based methods focus on the determination of differentiated features within local regions. To learn more discriminative global features, in this paper, we proposed the use of L2 normalization to tackle a neglected conflict between the widely used metric loss (triplet loss) and classification loss (softmax loss) in global feature-based methods. Furthermore, a multi-level metric learning network (MMLN) is proposed for fine-grained image classification based on global features. In the MMLN, multi-level metric learning objectives and classification objectives are present at multiple high-level layers. The multi-level metric learning objectives work together to supervise the network in order to learn highly discriminative features. In addition, a new probability aggregation strategy (PAS) is proposed to produce a fused prediction by combining the multi-level predictive probabilities. Experiments were conducted on three standard fine-grained classification datasets (CUB-200-2011, Stanford Cars, and FGVC-Aircraft). Results demonstrated that our MMLN achieved accuracies of 88.0%, 94.6% and 92.4% respectively and outperformed state-of-the-art methods, substantially improving fine-grained classification tasks. Besides, gradient-weighted class activation mapping (Grad-CAM) shows that the MMLN is able to pay more attention to the discriminative local regions due to the application of multi-level metric learning.https://ieeexplore.ieee.org/document/8903259/Fine-grained recognitionmetric learningmulti-level objectivesclassification |
spellingShingle | Jiabao Wang Yang Li Zhuang Miao Xun Zhao Zhang Rui Multi-Level Metric Learning Network for Fine-Grained Classification IEEE Access Fine-grained recognition metric learning multi-level objectives classification |
title | Multi-Level Metric Learning Network for Fine-Grained Classification |
title_full | Multi-Level Metric Learning Network for Fine-Grained Classification |
title_fullStr | Multi-Level Metric Learning Network for Fine-Grained Classification |
title_full_unstemmed | Multi-Level Metric Learning Network for Fine-Grained Classification |
title_short | Multi-Level Metric Learning Network for Fine-Grained Classification |
title_sort | multi level metric learning network for fine grained classification |
topic | Fine-grained recognition metric learning multi-level objectives classification |
url | https://ieeexplore.ieee.org/document/8903259/ |
work_keys_str_mv | AT jiabaowang multilevelmetriclearningnetworkforfinegrainedclassification AT yangli multilevelmetriclearningnetworkforfinegrainedclassification AT zhuangmiao multilevelmetriclearningnetworkforfinegrainedclassification AT xunzhao multilevelmetriclearningnetworkforfinegrainedclassification AT zhangrui multilevelmetriclearningnetworkforfinegrainedclassification |