A Transfer Learning Approach for Landslide Semantic Segmentation Based on Visual Foundation Model

Landslides are one of the most destructive natural disasters in the world, threatening human life and safety. With excellent performance as a foundation model for image segmentation, the segment anything model (SAM) has provided a novel paradigm for semantic segmentation research. However, the lack...

Full description

Saved in:
Bibliographic Details
Main Authors: Changhong Hou, Junchuan Yu, Daqing Ge, Liu Yang, Laidian Xi, Yunxuan Pang, Yi Wen
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10962290/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849322339072737280
author Changhong Hou
Junchuan Yu
Daqing Ge
Liu Yang
Laidian Xi
Yunxuan Pang
Yi Wen
author_facet Changhong Hou
Junchuan Yu
Daqing Ge
Liu Yang
Laidian Xi
Yunxuan Pang
Yi Wen
author_sort Changhong Hou
collection DOAJ
description Landslides are one of the most destructive natural disasters in the world, threatening human life and safety. With excellent performance as a foundation model for image segmentation, the segment anything model (SAM) has provided a novel paradigm for semantic segmentation research. However, the lack of remote sensing images in the SAM training data limits its ability to recognize landslides. In addition, despite the transfer learning approach can transfer SAM feature extraction capability to the landslide segmentation task, but it will consume a lot of computational resources and training time. In order to solve these challenges, this study proposes a TransLandSeg model that transfers the segmentation capability of SAM while learning landslide features at a low training cost. To limit model training parameters, the adaptive transfer learning (ATL) module is purposely designed, the image encoder is frozen during model training, only the ATL module and mask decoder are trained, and the knowledge learned from the ATL module is input into the original network. Moreover, to select the best ATL module, we also designed 9 kinds of ATL modules and analyzed the accuracy of the TransLandSeg model with different ATL modules. We selected the Bijie landslide dataset and the Landslide4Sense dataset for model training and testing. The experiment results show that the TransLandSeg model increases the mean intersection over union by 1.48% –13.01% compared to other state-of-the-art semantic segmentation models. In addition, TransLandSeg requires only 1.3% of SAM parameters to enable SAM's powerful capabilities to transfer to landslide segmentation.
format Article
id doaj-art-ab4355fd25a04e1e8d1cc3b3ed6a3193
institution Kabale University
issn 1939-1404
2151-1535
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
spelling doaj-art-ab4355fd25a04e1e8d1cc3b3ed6a31932025-08-20T03:49:23ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-0118115611157210.1109/JSTARS.2025.355988410962290A Transfer Learning Approach for Landslide Semantic Segmentation Based on Visual Foundation ModelChanghong Hou0https://orcid.org/0009-0002-9324-812XJunchuan Yu1https://orcid.org/0000-0003-2987-0504Daqing Ge2https://orcid.org/0009-0005-2779-8854Liu Yang3Laidian Xi4Yunxuan Pang5Yi Wen6College of Geoscience and Surveying Engineering, China University of Mining and Technology-Beijing, Beijing, ChinaDepartment of Satellite Application Research, China Aero Geophysical Survey and Remote Sensing Center for Natural Resources, Beijing, ChinaDepartment of Satellite Application Research, China Aero Geophysical Survey and Remote Sensing Center for Natural Resources, Beijing, ChinaCollege of Geoscience and Surveying Engineering, China University of Mining and Technology-Beijing, Beijing, ChinaSchool of Earth Sciences and Resources, China University of Geosciences, Beijing, ChinaSchool of Earth Sciences and Resources, China University of Geosciences, Beijing, ChinaCollege of Geoscience and Surveying Engineering, China University of Mining and Technology-Beijing, Beijing, ChinaLandslides are one of the most destructive natural disasters in the world, threatening human life and safety. With excellent performance as a foundation model for image segmentation, the segment anything model (SAM) has provided a novel paradigm for semantic segmentation research. However, the lack of remote sensing images in the SAM training data limits its ability to recognize landslides. In addition, despite the transfer learning approach can transfer SAM feature extraction capability to the landslide segmentation task, but it will consume a lot of computational resources and training time. In order to solve these challenges, this study proposes a TransLandSeg model that transfers the segmentation capability of SAM while learning landslide features at a low training cost. To limit model training parameters, the adaptive transfer learning (ATL) module is purposely designed, the image encoder is frozen during model training, only the ATL module and mask decoder are trained, and the knowledge learned from the ATL module is input into the original network. Moreover, to select the best ATL module, we also designed 9 kinds of ATL modules and analyzed the accuracy of the TransLandSeg model with different ATL modules. We selected the Bijie landslide dataset and the Landslide4Sense dataset for model training and testing. The experiment results show that the TransLandSeg model increases the mean intersection over union by 1.48% –13.01% compared to other state-of-the-art semantic segmentation models. In addition, TransLandSeg requires only 1.3% of SAM parameters to enable SAM's powerful capabilities to transfer to landslide segmentation.https://ieeexplore.ieee.org/document/10962290/Foundation modellandslide recognitionsemantic segmentationtransfer learning
spellingShingle Changhong Hou
Junchuan Yu
Daqing Ge
Liu Yang
Laidian Xi
Yunxuan Pang
Yi Wen
A Transfer Learning Approach for Landslide Semantic Segmentation Based on Visual Foundation Model
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Foundation model
landslide recognition
semantic segmentation
transfer learning
title A Transfer Learning Approach for Landslide Semantic Segmentation Based on Visual Foundation Model
title_full A Transfer Learning Approach for Landslide Semantic Segmentation Based on Visual Foundation Model
title_fullStr A Transfer Learning Approach for Landslide Semantic Segmentation Based on Visual Foundation Model
title_full_unstemmed A Transfer Learning Approach for Landslide Semantic Segmentation Based on Visual Foundation Model
title_short A Transfer Learning Approach for Landslide Semantic Segmentation Based on Visual Foundation Model
title_sort transfer learning approach for landslide semantic segmentation based on visual foundation model
topic Foundation model
landslide recognition
semantic segmentation
transfer learning
url https://ieeexplore.ieee.org/document/10962290/
work_keys_str_mv AT changhonghou atransferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT junchuanyu atransferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT daqingge atransferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT liuyang atransferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT laidianxi atransferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT yunxuanpang atransferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT yiwen atransferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT changhonghou transferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT junchuanyu transferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT daqingge transferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT liuyang transferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT laidianxi transferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT yunxuanpang transferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel
AT yiwen transferlearningapproachforlandslidesemanticsegmentationbasedonvisualfoundationmodel