Road Extraction from Remote Sensing Imagery with Spatial Attention Based on Swin Transformer
Road extraction is a crucial aspect of remote sensing imagery processing that plays a significant role in various remote sensing applications, including automatic driving, urban planning, and path navigation. However, accurate road extraction is a challenging task due to factors such as high road de...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-03-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/16/7/1183 |
| _version_ | 1797212022517006336 |
|---|---|
| author | Xianhong Zhu Xiaohui Huang Weijia Cao Xiaofei Yang Yunfei Zhou Shaokai Wang |
| author_facet | Xianhong Zhu Xiaohui Huang Weijia Cao Xiaofei Yang Yunfei Zhou Shaokai Wang |
| author_sort | Xianhong Zhu |
| collection | DOAJ |
| description | Road extraction is a crucial aspect of remote sensing imagery processing that plays a significant role in various remote sensing applications, including automatic driving, urban planning, and path navigation. However, accurate road extraction is a challenging task due to factors such as high road density, building occlusion, and complex traffic environments. In this study, a Spatial Attention Swin Transformer (SASwin Transformer) architecture is proposed to create a robust encoder capable of extracting roads from remote sensing imagery. In this architecture, we have developed a spatial self-attention (SSA) module that captures efficient and rich spatial information through spatial self-attention to reconstruct the feature map. Following this, the module performs residual connections with the input, which helps reduce interference from unrelated regions. Additionally, we designed a Spatial MLP (SMLP) module to aggregate spatial feature information from multiple branches while simultaneously reducing computational complexity. Two public road datasets, the Massachusetts dataset and the DeepGlobe dataset, were used for extensive experiments. The results show that our proposed model has an improved overall performance compared to several state-of-the-art algorithms. In particular, on the two datasets, our model outperforms D-LinkNet with an increase in Intersection over Union (IoU) metrics of 1.88% and 1.84%, respectively. |
| first_indexed | 2024-04-24T10:35:46Z |
| format | Article |
| id | doaj.art-994c3c5a5125402d8377b85c666f0373 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-04-24T10:35:46Z |
| publishDate | 2024-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-994c3c5a5125402d8377b85c666f03732024-04-12T13:25:33ZengMDPI AGRemote Sensing2072-42922024-03-01167118310.3390/rs16071183Road Extraction from Remote Sensing Imagery with Spatial Attention Based on Swin TransformerXianhong Zhu0Xiaohui Huang1Weijia Cao2Xiaofei Yang3Yunfei Zhou4Shaokai Wang5School of Information Engineering, East China Jiaotong University, Nanchang 330013, ChinaSchool of Information Engineering, East China Jiaotong University, Nanchang 330013, ChinaAerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, ChinaSchool of Electronic and Communication Engineering, Guangzhou University, Guangzhou 511370, ChinaSchool of Information Engineering, East China Jiaotong University, Nanchang 330013, ChinaSchool of Information Engineering, East China Jiaotong University, Nanchang 330013, ChinaRoad extraction is a crucial aspect of remote sensing imagery processing that plays a significant role in various remote sensing applications, including automatic driving, urban planning, and path navigation. However, accurate road extraction is a challenging task due to factors such as high road density, building occlusion, and complex traffic environments. In this study, a Spatial Attention Swin Transformer (SASwin Transformer) architecture is proposed to create a robust encoder capable of extracting roads from remote sensing imagery. In this architecture, we have developed a spatial self-attention (SSA) module that captures efficient and rich spatial information through spatial self-attention to reconstruct the feature map. Following this, the module performs residual connections with the input, which helps reduce interference from unrelated regions. Additionally, we designed a Spatial MLP (SMLP) module to aggregate spatial feature information from multiple branches while simultaneously reducing computational complexity. Two public road datasets, the Massachusetts dataset and the DeepGlobe dataset, were used for extensive experiments. The results show that our proposed model has an improved overall performance compared to several state-of-the-art algorithms. In particular, on the two datasets, our model outperforms D-LinkNet with an increase in Intersection over Union (IoU) metrics of 1.88% and 1.84%, respectively.https://www.mdpi.com/2072-4292/16/7/1183remote sensing applicationsroad extractionspatial self-attentionSpatial MLP |
| spellingShingle | Xianhong Zhu Xiaohui Huang Weijia Cao Xiaofei Yang Yunfei Zhou Shaokai Wang Road Extraction from Remote Sensing Imagery with Spatial Attention Based on Swin Transformer Remote Sensing remote sensing applications road extraction spatial self-attention Spatial MLP |
| title | Road Extraction from Remote Sensing Imagery with Spatial Attention Based on Swin Transformer |
| title_full | Road Extraction from Remote Sensing Imagery with Spatial Attention Based on Swin Transformer |
| title_fullStr | Road Extraction from Remote Sensing Imagery with Spatial Attention Based on Swin Transformer |
| title_full_unstemmed | Road Extraction from Remote Sensing Imagery with Spatial Attention Based on Swin Transformer |
| title_short | Road Extraction from Remote Sensing Imagery with Spatial Attention Based on Swin Transformer |
| title_sort | road extraction from remote sensing imagery with spatial attention based on swin transformer |
| topic | remote sensing applications road extraction spatial self-attention Spatial MLP |
| url | https://www.mdpi.com/2072-4292/16/7/1183 |
| work_keys_str_mv | AT xianhongzhu roadextractionfromremotesensingimagerywithspatialattentionbasedonswintransformer AT xiaohuihuang roadextractionfromremotesensingimagerywithspatialattentionbasedonswintransformer AT weijiacao roadextractionfromremotesensingimagerywithspatialattentionbasedonswintransformer AT xiaofeiyang roadextractionfromremotesensingimagerywithspatialattentionbasedonswintransformer AT yunfeizhou roadextractionfromremotesensingimagerywithspatialattentionbasedonswintransformer AT shaokaiwang roadextractionfromremotesensingimagerywithspatialattentionbasedonswintransformer |