Saliency guided data augmentation strategy for maximally utilizing an object's visual information.
Among the various types of data augmentation strategies, the mixup-based approach has been particularly studied. However, in existing mixup-based approaches, object loss and label mismatching can occur if random patches are utilized when constructing augmented images, and additionally, patches that...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Public Library of Science (PLoS)
2022-01-01
|
Series: | PLoS ONE |
Online Access: | https://doi.org/10.1371/journal.pone.0274767 |
_version_ | 1811182186458513408 |
---|---|
author | Junhyeok An Soojin Jang Junehyoung Kwon Kyohoon Jin YoungBin Kim |
author_facet | Junhyeok An Soojin Jang Junehyoung Kwon Kyohoon Jin YoungBin Kim |
author_sort | Junhyeok An |
collection | DOAJ |
description | Among the various types of data augmentation strategies, the mixup-based approach has been particularly studied. However, in existing mixup-based approaches, object loss and label mismatching can occur if random patches are utilized when constructing augmented images, and additionally, patches that do not contain objects might be included, which degrades performance. In this paper, we propose a novel augmentation method that mixes patches in a non-overlapping manner after they are extracted from the salient regions in an image. The suggested method can make effective use of object characteristics, because the constructed image consists only of visually important regions and is robust to noise. Since the patches do not occlude each other, the semantically meaningful information in the salient regions can be fully utilized. Additionally, our method is more robust to adversarial attack than the conventional augmentation method. In the experimental results, when Wide ResNet was trained on the public datasets, CIFAR-10, CIFAR-100 and STL-10, the top-1 accuracy was 97.26%, 83.99% and 82.40% respectively, which surpasses other augmentation methods. |
first_indexed | 2024-04-11T09:27:46Z |
format | Article |
id | doaj.art-4c9aef747b914205a3f504f8d3199850 |
institution | Directory Open Access Journal |
issn | 1932-6203 |
language | English |
last_indexed | 2024-04-11T09:27:46Z |
publishDate | 2022-01-01 |
publisher | Public Library of Science (PLoS) |
record_format | Article |
series | PLoS ONE |
spelling | doaj.art-4c9aef747b914205a3f504f8d31998502022-12-22T04:31:58ZengPublic Library of Science (PLoS)PLoS ONE1932-62032022-01-011710e027476710.1371/journal.pone.0274767Saliency guided data augmentation strategy for maximally utilizing an object's visual information.Junhyeok AnSoojin JangJunehyoung KwonKyohoon JinYoungBin KimAmong the various types of data augmentation strategies, the mixup-based approach has been particularly studied. However, in existing mixup-based approaches, object loss and label mismatching can occur if random patches are utilized when constructing augmented images, and additionally, patches that do not contain objects might be included, which degrades performance. In this paper, we propose a novel augmentation method that mixes patches in a non-overlapping manner after they are extracted from the salient regions in an image. The suggested method can make effective use of object characteristics, because the constructed image consists only of visually important regions and is robust to noise. Since the patches do not occlude each other, the semantically meaningful information in the salient regions can be fully utilized. Additionally, our method is more robust to adversarial attack than the conventional augmentation method. In the experimental results, when Wide ResNet was trained on the public datasets, CIFAR-10, CIFAR-100 and STL-10, the top-1 accuracy was 97.26%, 83.99% and 82.40% respectively, which surpasses other augmentation methods.https://doi.org/10.1371/journal.pone.0274767 |
spellingShingle | Junhyeok An Soojin Jang Junehyoung Kwon Kyohoon Jin YoungBin Kim Saliency guided data augmentation strategy for maximally utilizing an object's visual information. PLoS ONE |
title | Saliency guided data augmentation strategy for maximally utilizing an object's visual information. |
title_full | Saliency guided data augmentation strategy for maximally utilizing an object's visual information. |
title_fullStr | Saliency guided data augmentation strategy for maximally utilizing an object's visual information. |
title_full_unstemmed | Saliency guided data augmentation strategy for maximally utilizing an object's visual information. |
title_short | Saliency guided data augmentation strategy for maximally utilizing an object's visual information. |
title_sort | saliency guided data augmentation strategy for maximally utilizing an object s visual information |
url | https://doi.org/10.1371/journal.pone.0274767 |
work_keys_str_mv | AT junhyeokan saliencyguideddataaugmentationstrategyformaximallyutilizinganobjectsvisualinformation AT soojinjang saliencyguideddataaugmentationstrategyformaximallyutilizinganobjectsvisualinformation AT junehyoungkwon saliencyguideddataaugmentationstrategyformaximallyutilizinganobjectsvisualinformation AT kyohoonjin saliencyguideddataaugmentationstrategyformaximallyutilizinganobjectsvisualinformation AT youngbinkim saliencyguideddataaugmentationstrategyformaximallyutilizinganobjectsvisualinformation |