Improved Pseudomasks Generation for Weakly Supervised Building Extraction From High-Resolution Remote Sensing Imagery
Benefiting from free labeling pixel-level samples, weakly supervised semantic segmentation (WSSS) is making progress in automatically extracting building from high-resolution (HR) remote sensing (RS) imagery. For WSSS methods, generating high-quality pseudomasks is crucial for accurate building extr...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9684996/ |
| _version_ | 1818296580327866368 |
|---|---|
| author | Fang Fang Daoyuan Zheng Shengwen Li Yuanyuan Liu Linyun Zeng Jiahui Zhang Bo Wan |
| author_facet | Fang Fang Daoyuan Zheng Shengwen Li Yuanyuan Liu Linyun Zeng Jiahui Zhang Bo Wan |
| author_sort | Fang Fang |
| collection | DOAJ |
| description | Benefiting from free labeling pixel-level samples, weakly supervised semantic segmentation (WSSS) is making progress in automatically extracting building from high-resolution (HR) remote sensing (RS) imagery. For WSSS methods, generating high-quality pseudomasks is crucial for accurate building extraction.To improve the performance of generating pseudomasks by using image-level labels, this article proposes a weakly supervised building extraction method by combining adversarial climbing and gated convolution. The proposed method optimizes class activation maps (CAMs) by using adversarial climbing strategy, generates accurate class boundary maps by introducing a gated convolution module, and further refines building pseudomasks by fusing pairing semantic affinities and CAMs with a random walk strategy. Experimental results on three datasets—two ISPRS datasets and a self-annotated dataset—demonstrate that the proposed approach outperformed SOTA WSSS methods, leading to improvement of building extraction from HR RS imager. This article provides a new approach for optimizing pseudomasks generation, and a methodological reference for the applications of weakly supervised on RS images. |
| first_indexed | 2024-12-13T04:05:48Z |
| format | Article |
| id | doaj.art-fb07a123cd944cb3a275e821669dcd39 |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-12-13T04:05:48Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-fb07a123cd944cb3a275e821669dcd392022-12-22T00:00:12ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352022-01-01151629164210.1109/JSTARS.2022.31441769684996Improved Pseudomasks Generation for Weakly Supervised Building Extraction From High-Resolution Remote Sensing ImageryFang Fang0Daoyuan Zheng1https://orcid.org/0000-0003-0344-1760Shengwen Li2https://orcid.org/0000-0002-1829-4006Yuanyuan Liu3Linyun Zeng4Jiahui Zhang5Bo Wan6School of Computer Science, China University of Geosciences, Wuhan, ChinaSchool of Computer Science, China University of Geosciences, Wuhan, ChinaSchool of Computer Science, China University of Geosciences, Wuhan, ChinaSchool of Computer Science, China University of Geosciences, Wuhan, ChinaSchool of Computer Science, China University of Geosciences, Wuhan, ChinaNational Engineering Research Center of Geographic Information System, China University of Geosciences, Wuhan, ChinaSchool of Computer Science, China University of Geosciences, Wuhan, ChinaBenefiting from free labeling pixel-level samples, weakly supervised semantic segmentation (WSSS) is making progress in automatically extracting building from high-resolution (HR) remote sensing (RS) imagery. For WSSS methods, generating high-quality pseudomasks is crucial for accurate building extraction.To improve the performance of generating pseudomasks by using image-level labels, this article proposes a weakly supervised building extraction method by combining adversarial climbing and gated convolution. The proposed method optimizes class activation maps (CAMs) by using adversarial climbing strategy, generates accurate class boundary maps by introducing a gated convolution module, and further refines building pseudomasks by fusing pairing semantic affinities and CAMs with a random walk strategy. Experimental results on three datasets—two ISPRS datasets and a self-annotated dataset—demonstrate that the proposed approach outperformed SOTA WSSS methods, leading to improvement of building extraction from HR RS imager. This article provides a new approach for optimizing pseudomasks generation, and a methodological reference for the applications of weakly supervised on RS images.https://ieeexplore.ieee.org/document/9684996/Adversarial climbing (AC)building extractiongated convolutionhigh-resolution (HR) remote sensing (RS) imageryweakly supervised semantic segmentation (WSSS) |
| spellingShingle | Fang Fang Daoyuan Zheng Shengwen Li Yuanyuan Liu Linyun Zeng Jiahui Zhang Bo Wan Improved Pseudomasks Generation for Weakly Supervised Building Extraction From High-Resolution Remote Sensing Imagery IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Adversarial climbing (AC) building extraction gated convolution high-resolution (HR) remote sensing (RS) imagery weakly supervised semantic segmentation (WSSS) |
| title | Improved Pseudomasks Generation for Weakly Supervised Building Extraction From High-Resolution Remote Sensing Imagery |
| title_full | Improved Pseudomasks Generation for Weakly Supervised Building Extraction From High-Resolution Remote Sensing Imagery |
| title_fullStr | Improved Pseudomasks Generation for Weakly Supervised Building Extraction From High-Resolution Remote Sensing Imagery |
| title_full_unstemmed | Improved Pseudomasks Generation for Weakly Supervised Building Extraction From High-Resolution Remote Sensing Imagery |
| title_short | Improved Pseudomasks Generation for Weakly Supervised Building Extraction From High-Resolution Remote Sensing Imagery |
| title_sort | improved pseudomasks generation for weakly supervised building extraction from high resolution remote sensing imagery |
| topic | Adversarial climbing (AC) building extraction gated convolution high-resolution (HR) remote sensing (RS) imagery weakly supervised semantic segmentation (WSSS) |
| url | https://ieeexplore.ieee.org/document/9684996/ |
| work_keys_str_mv | AT fangfang improvedpseudomasksgenerationforweaklysupervisedbuildingextractionfromhighresolutionremotesensingimagery AT daoyuanzheng improvedpseudomasksgenerationforweaklysupervisedbuildingextractionfromhighresolutionremotesensingimagery AT shengwenli improvedpseudomasksgenerationforweaklysupervisedbuildingextractionfromhighresolutionremotesensingimagery AT yuanyuanliu improvedpseudomasksgenerationforweaklysupervisedbuildingextractionfromhighresolutionremotesensingimagery AT linyunzeng improvedpseudomasksgenerationforweaklysupervisedbuildingextractionfromhighresolutionremotesensingimagery AT jiahuizhang improvedpseudomasksgenerationforweaklysupervisedbuildingextractionfromhighresolutionremotesensingimagery AT bowan improvedpseudomasksgenerationforweaklysupervisedbuildingextractionfromhighresolutionremotesensingimagery |