Uncertainty-Guided Segmentation Network for Geospatial Object Segmentation
Geospatial objects pose significant challenges, including dense distribution, substantial interclass variations, and minimal intraclass variations. These complexities make achieving precise foreground object segmentation in high-resolution remote sensing images highly challenging. Current segmentati...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10418974/ |
| _version_ | 1797243400108376064 |
|---|---|
| author | Hongyu Jia Wenwu Yang Lin Wang Haolin Li |
| author_facet | Hongyu Jia Wenwu Yang Lin Wang Haolin Li |
| author_sort | Hongyu Jia |
| collection | DOAJ |
| description | Geospatial objects pose significant challenges, including dense distribution, substantial interclass variations, and minimal intraclass variations. These complexities make achieving precise foreground object segmentation in high-resolution remote sensing images highly challenging. Current segmentation approaches often rely on the standard encoder–decoder architecture to extract object-related information, but overlook the inherent uncertainty issues that arise during the process. In this article, we aim to enhance segmentation by introducing an uncertainty-guided decoding mechanism and propose the uncertainty-guided segmentation network (UGSNet). Specifically, building upon the conventional encoder–decoder architecture, we initially employ the pyramid vision transformer to extract multilevel features containing extensive long-range information. We then introduce an uncertainty-guided decoding mechanism, addressing both epistemic and aleatoric uncertainties, to progressively refine segmentation with higher certainty at each level. With this uncertainty-guided decoding mechanism, our UGSNet achieves accurate geospatial object segmentation. To validate the effectiveness of UGSNet, we conduct extensive experiments on the large-scale ISAID dataset, and the results unequivocally demonstrate the superiority of our method over other state-of-the-art segmentation methods. |
| first_indexed | 2024-04-24T18:54:30Z |
| format | Article |
| id | doaj.art-e248be5221f24d4fa072bef2b5993b9b |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-04-24T18:54:30Z |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-e248be5221f24d4fa072bef2b5993b9b2024-03-26T17:46:31ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352024-01-01175824583310.1109/JSTARS.2024.336169310418974Uncertainty-Guided Segmentation Network for Geospatial Object SegmentationHongyu Jia0https://orcid.org/0009-0004-9479-3482Wenwu Yang1https://orcid.org/0000-0002-5245-6857Lin Wang2https://orcid.org/0009-0003-3413-7144Haolin Li3https://orcid.org/0009-0003-7200-1432School of Maritime Economics and Management, Dalian Maritime University, Dalian, ChinaSchool of Maritime Economics and Management, Dalian Maritime University, Dalian, ChinaSchool of Maritime Economics and Management, Dalian Maritime University, Dalian, ChinaSchool of Maritime Economics and Management, Dalian Maritime University, Dalian, ChinaGeospatial objects pose significant challenges, including dense distribution, substantial interclass variations, and minimal intraclass variations. These complexities make achieving precise foreground object segmentation in high-resolution remote sensing images highly challenging. Current segmentation approaches often rely on the standard encoder–decoder architecture to extract object-related information, but overlook the inherent uncertainty issues that arise during the process. In this article, we aim to enhance segmentation by introducing an uncertainty-guided decoding mechanism and propose the uncertainty-guided segmentation network (UGSNet). Specifically, building upon the conventional encoder–decoder architecture, we initially employ the pyramid vision transformer to extract multilevel features containing extensive long-range information. We then introduce an uncertainty-guided decoding mechanism, addressing both epistemic and aleatoric uncertainties, to progressively refine segmentation with higher certainty at each level. With this uncertainty-guided decoding mechanism, our UGSNet achieves accurate geospatial object segmentation. To validate the effectiveness of UGSNet, we conduct extensive experiments on the large-scale ISAID dataset, and the results unequivocally demonstrate the superiority of our method over other state-of-the-art segmentation methods.https://ieeexplore.ieee.org/document/10418974/Geospatial object segmentationremote sensing (RS)semantic segmentationuncertainty decoding mechanism |
| spellingShingle | Hongyu Jia Wenwu Yang Lin Wang Haolin Li Uncertainty-Guided Segmentation Network for Geospatial Object Segmentation IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Geospatial object segmentation remote sensing (RS) semantic segmentation uncertainty decoding mechanism |
| title | Uncertainty-Guided Segmentation Network for Geospatial Object Segmentation |
| title_full | Uncertainty-Guided Segmentation Network for Geospatial Object Segmentation |
| title_fullStr | Uncertainty-Guided Segmentation Network for Geospatial Object Segmentation |
| title_full_unstemmed | Uncertainty-Guided Segmentation Network for Geospatial Object Segmentation |
| title_short | Uncertainty-Guided Segmentation Network for Geospatial Object Segmentation |
| title_sort | uncertainty guided segmentation network for geospatial object segmentation |
| topic | Geospatial object segmentation remote sensing (RS) semantic segmentation uncertainty decoding mechanism |
| url | https://ieeexplore.ieee.org/document/10418974/ |
| work_keys_str_mv | AT hongyujia uncertaintyguidedsegmentationnetworkforgeospatialobjectsegmentation AT wenwuyang uncertaintyguidedsegmentationnetworkforgeospatialobjectsegmentation AT linwang uncertaintyguidedsegmentationnetworkforgeospatialobjectsegmentation AT haolinli uncertaintyguidedsegmentationnetworkforgeospatialobjectsegmentation |