A Multiscale Attention Network for Remote Sensing Scene Images Classification
The remote sensing scene images classification has been of great value to civil and military fields. Deep learning models, especially the convolutional neural network (CNN), have achieved great success in this task, however, they may suffer from two challenges: first, the sizes of the category objec...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2021-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9528049/ |
| _version_ | 1831762422380101632 |
|---|---|
| author | Guokai Zhang Weizhe Xu Wei Zhao Chenxi Huang Eddie Ng Yk Yongyong Chen Jian Su |
| author_facet | Guokai Zhang Weizhe Xu Wei Zhao Chenxi Huang Eddie Ng Yk Yongyong Chen Jian Su |
| author_sort | Guokai Zhang |
| collection | DOAJ |
| description | The remote sensing scene images classification has been of great value to civil and military fields. Deep learning models, especially the convolutional neural network (CNN), have achieved great success in this task, however, they may suffer from two challenges: first, the sizes of the category objects are usually different, but the conventional CNN extracts the features with fixed convolution extractor, which could cause the failure in learning the multiscale features; second, some image regions may not be useful during the feature learning process, therefore, how to guide the network to select and focus on the most relevant regions is crucially vital for remote sensing scene image classification. To address these two challenges, we propose a multiscale attention network (MSA-Network), which integrates a multiscale (MS) module and a channel and position attention (CPA) module to boost the performance of the remote sensing scene classification. The proposed MS module learns multiscale features by adopting various sizes of sliding windows from different depths’ layers and receptive fields. The CPA module is composed of two parts: the channel attention (CA) module and the position attention (PA) one. The CA module learns the global attention features from channel-level, and the PA module extracts the local attention features from pixel-level. Thus, fusing both of those two attention features, the network is apt to focus on the more critical and salient regions automatically. Extensive experiments on UC Merced, AID, NWPU-RESISC45 datasets demonstrate that the proposed MSA-Network outperforms several state-of-the-art methods. |
| first_indexed | 2024-12-22T05:06:43Z |
| format | Article |
| id | doaj.art-75554dff98284939a18c6a702cef2509 |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-12-22T05:06:43Z |
| publishDate | 2021-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-75554dff98284939a18c6a702cef25092022-12-21T18:38:06ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352021-01-01149530954510.1109/JSTARS.2021.31096619528049A Multiscale Attention Network for Remote Sensing Scene Images ClassificationGuokai Zhang0https://orcid.org/0000-0002-0952-8325Weizhe Xu1Wei Zhao2Chenxi Huang3https://orcid.org/0000-0002-2100-0259Eddie Ng Yk4https://orcid.org/0000-0002-5701-1080Yongyong Chen5https://orcid.org/0000-0003-1970-1993Jian Su6https://orcid.org/0000-0003-0634-4843School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai, ChinaSchool of Computer Science, University of Manchester, Manchester, U.K.School of Software Enginnering, Tongji University, Shanghai, ChinaSchool of Informatics, Xiamen University, Xiamen, ChinaSchool of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore, SingaporeSchool of Computer Science and Technology, Harbin Institute of Technology, Shenzhen, ChinaSchool of Computer and Software, Nanjing University of Information Science and Technology, Nanjing, ChinaThe remote sensing scene images classification has been of great value to civil and military fields. Deep learning models, especially the convolutional neural network (CNN), have achieved great success in this task, however, they may suffer from two challenges: first, the sizes of the category objects are usually different, but the conventional CNN extracts the features with fixed convolution extractor, which could cause the failure in learning the multiscale features; second, some image regions may not be useful during the feature learning process, therefore, how to guide the network to select and focus on the most relevant regions is crucially vital for remote sensing scene image classification. To address these two challenges, we propose a multiscale attention network (MSA-Network), which integrates a multiscale (MS) module and a channel and position attention (CPA) module to boost the performance of the remote sensing scene classification. The proposed MS module learns multiscale features by adopting various sizes of sliding windows from different depths’ layers and receptive fields. The CPA module is composed of two parts: the channel attention (CA) module and the position attention (PA) one. The CA module learns the global attention features from channel-level, and the PA module extracts the local attention features from pixel-level. Thus, fusing both of those two attention features, the network is apt to focus on the more critical and salient regions automatically. Extensive experiments on UC Merced, AID, NWPU-RESISC45 datasets demonstrate that the proposed MSA-Network outperforms several state-of-the-art methods.https://ieeexplore.ieee.org/document/9528049/Remote sensing scenemulti-scaleattentionfeature fusion |
| spellingShingle | Guokai Zhang Weizhe Xu Wei Zhao Chenxi Huang Eddie Ng Yk Yongyong Chen Jian Su A Multiscale Attention Network for Remote Sensing Scene Images Classification IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Remote sensing scene multi-scale attention feature fusion |
| title | A Multiscale Attention Network for Remote Sensing Scene Images Classification |
| title_full | A Multiscale Attention Network for Remote Sensing Scene Images Classification |
| title_fullStr | A Multiscale Attention Network for Remote Sensing Scene Images Classification |
| title_full_unstemmed | A Multiscale Attention Network for Remote Sensing Scene Images Classification |
| title_short | A Multiscale Attention Network for Remote Sensing Scene Images Classification |
| title_sort | multiscale attention network for remote sensing scene images classification |
| topic | Remote sensing scene multi-scale attention feature fusion |
| url | https://ieeexplore.ieee.org/document/9528049/ |
| work_keys_str_mv | AT guokaizhang amultiscaleattentionnetworkforremotesensingsceneimagesclassification AT weizhexu amultiscaleattentionnetworkforremotesensingsceneimagesclassification AT weizhao amultiscaleattentionnetworkforremotesensingsceneimagesclassification AT chenxihuang amultiscaleattentionnetworkforremotesensingsceneimagesclassification AT eddiengyk amultiscaleattentionnetworkforremotesensingsceneimagesclassification AT yongyongchen amultiscaleattentionnetworkforremotesensingsceneimagesclassification AT jiansu amultiscaleattentionnetworkforremotesensingsceneimagesclassification AT guokaizhang multiscaleattentionnetworkforremotesensingsceneimagesclassification AT weizhexu multiscaleattentionnetworkforremotesensingsceneimagesclassification AT weizhao multiscaleattentionnetworkforremotesensingsceneimagesclassification AT chenxihuang multiscaleattentionnetworkforremotesensingsceneimagesclassification AT eddiengyk multiscaleattentionnetworkforremotesensingsceneimagesclassification AT yongyongchen multiscaleattentionnetworkforremotesensingsceneimagesclassification AT jiansu multiscaleattentionnetworkforremotesensingsceneimagesclassification |