Conv-trans dual network for landslide detection of multi-channel optical remote sensing images
Landslide detection is crucial for disaster management and prevention. With the advent of multi-channel optical remote sensing technology, detecting landslides have become more accessible and more accurate. Although the use of the convolutional neural network (CNN) has significantly increased the ac...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2023-05-01
|
| Series: | Frontiers in Earth Science |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/feart.2023.1182145/full |
| _version_ | 1797819855585411072 |
|---|---|
| author | Xin Chen Mingzhe Liu Dongfen Li Jiaru Jia Aiqing Yang Wenfeng Zheng Lirong Yin |
| author_facet | Xin Chen Mingzhe Liu Dongfen Li Jiaru Jia Aiqing Yang Wenfeng Zheng Lirong Yin |
| author_sort | Xin Chen |
| collection | DOAJ |
| description | Landslide detection is crucial for disaster management and prevention. With the advent of multi-channel optical remote sensing technology, detecting landslides have become more accessible and more accurate. Although the use of the convolutional neural network (CNN) has significantly increased the accuracy of landslide detection on multi-channel optical remote sensing images, most previous methods using CNN lack the ability to obtain global context information due to the structural limitations of the convolution operation. Motivated by the powerful global modeling capability of the Swin transformer, we propose a new Conv-Trans Dual Network (CTDNet) based on Swin-Unet. First, we propose a dual-stream module (CTDBlock) that combines the advantages of ConvNeXt and Swin transformer, which can establish pixel-level connections and global dependencies from the CNN hierarchy to enhance the ability of the model to extract spatial information. Second, we apply an additional gating module (AGM) to effectively fuse the low-level information extracted by the shallow network and the high-level information extracted by the deep network and minimize the loss of detailed information when propagating. In addition, We conducted extensive subjective and objective comparison and ablation experiments on the Landslide4Sense dataset. Experimental results demonstrate that our proposed CTDNet outperforms other models currently applied in our experiments. |
| first_indexed | 2024-03-13T09:28:46Z |
| format | Article |
| id | doaj.art-fec7e256fbda4292be4b4580805972b0 |
| institution | Directory Open Access Journal |
| issn | 2296-6463 |
| language | English |
| last_indexed | 2024-03-13T09:28:46Z |
| publishDate | 2023-05-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Earth Science |
| spelling | doaj.art-fec7e256fbda4292be4b4580805972b02023-05-26T04:43:44ZengFrontiers Media S.A.Frontiers in Earth Science2296-64632023-05-011110.3389/feart.2023.11821451182145Conv-trans dual network for landslide detection of multi-channel optical remote sensing imagesXin Chen0Mingzhe Liu1Dongfen Li2Jiaru Jia3Aiqing Yang4Wenfeng Zheng5Lirong Yin6State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu, ChinaState Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu, ChinaState Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu, ChinaState Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu, ChinaState Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu, ChinaSchool of Automation, University of Electronic Science and Technology of China, Chengdu, ChinaDepartment of Geography and Anthropology, Louisiana State University, Baton Rouge, LA, United StatesLandslide detection is crucial for disaster management and prevention. With the advent of multi-channel optical remote sensing technology, detecting landslides have become more accessible and more accurate. Although the use of the convolutional neural network (CNN) has significantly increased the accuracy of landslide detection on multi-channel optical remote sensing images, most previous methods using CNN lack the ability to obtain global context information due to the structural limitations of the convolution operation. Motivated by the powerful global modeling capability of the Swin transformer, we propose a new Conv-Trans Dual Network (CTDNet) based on Swin-Unet. First, we propose a dual-stream module (CTDBlock) that combines the advantages of ConvNeXt and Swin transformer, which can establish pixel-level connections and global dependencies from the CNN hierarchy to enhance the ability of the model to extract spatial information. Second, we apply an additional gating module (AGM) to effectively fuse the low-level information extracted by the shallow network and the high-level information extracted by the deep network and minimize the loss of detailed information when propagating. In addition, We conducted extensive subjective and objective comparison and ablation experiments on the Landslide4Sense dataset. Experimental results demonstrate that our proposed CTDNet outperforms other models currently applied in our experiments.https://www.frontiersin.org/articles/10.3389/feart.2023.1182145/fulllandslide detectionswin transformerconvolutional neural network (CNN)remote sensing (RS)landslide |
| spellingShingle | Xin Chen Mingzhe Liu Dongfen Li Jiaru Jia Aiqing Yang Wenfeng Zheng Lirong Yin Conv-trans dual network for landslide detection of multi-channel optical remote sensing images Frontiers in Earth Science landslide detection swin transformer convolutional neural network (CNN) remote sensing (RS) landslide |
| title | Conv-trans dual network for landslide detection of multi-channel optical remote sensing images |
| title_full | Conv-trans dual network for landslide detection of multi-channel optical remote sensing images |
| title_fullStr | Conv-trans dual network for landslide detection of multi-channel optical remote sensing images |
| title_full_unstemmed | Conv-trans dual network for landslide detection of multi-channel optical remote sensing images |
| title_short | Conv-trans dual network for landslide detection of multi-channel optical remote sensing images |
| title_sort | conv trans dual network for landslide detection of multi channel optical remote sensing images |
| topic | landslide detection swin transformer convolutional neural network (CNN) remote sensing (RS) landslide |
| url | https://www.frontiersin.org/articles/10.3389/feart.2023.1182145/full |
| work_keys_str_mv | AT xinchen convtransdualnetworkforlandslidedetectionofmultichannelopticalremotesensingimages AT mingzheliu convtransdualnetworkforlandslidedetectionofmultichannelopticalremotesensingimages AT dongfenli convtransdualnetworkforlandslidedetectionofmultichannelopticalremotesensingimages AT jiarujia convtransdualnetworkforlandslidedetectionofmultichannelopticalremotesensingimages AT aiqingyang convtransdualnetworkforlandslidedetectionofmultichannelopticalremotesensingimages AT wenfengzheng convtransdualnetworkforlandslidedetectionofmultichannelopticalremotesensingimages AT lirongyin convtransdualnetworkforlandslidedetectionofmultichannelopticalremotesensingimages |