A Decoupled Segmentation-Classification Strategy Based on Semantic-SAM for Precise Semantic Segmentation in Coal Mine Areas

To address complex semantic segmentation in coal mine areas, this study proposes the SAM-SEF (SAM-based Semantic Enhancement Framework). It integrates Semantic-SAM’s zero-shot segmentation capability with specialized deep learning models through a “segmentation-classification-u...

Full description

Saved in:
Bibliographic Details
Main Authors: Libing Wang, Jihong Dong, Lei Wang, Feng Liu, Qingke Wen, Bekir Genc
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/11023034/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849421440025100288
author Libing Wang
Jihong Dong
Lei Wang
Feng Liu
Qingke Wen
Bekir Genc
author_facet Libing Wang
Jihong Dong
Lei Wang
Feng Liu
Qingke Wen
Bekir Genc
author_sort Libing Wang
collection DOAJ
description To address complex semantic segmentation in coal mine areas, this study proposes the SAM-SEF (SAM-based Semantic Enhancement Framework). It integrates Semantic-SAM’s zero-shot segmentation capability with specialized deep learning models through a “segmentation-classification-update” strategy. Semantic-SAM initially provides category-agnostic multi-granularity segmentations; specialized deep learning models then perform semantic understanding of these regions, and a category mapping mechanism generates precise results. Evaluating 18 models, SAM-SEF-ENet-B0 achieved the highest IoUs for Open-pit Coal Mining Sites (OCMS) (93.68%) and Composite Coal-related Sites (CCS) (81.52%), with an mIoU of 87.60%, significantly outperforming traditional models. A Semantic-SAM-based semi-automatic annotation tool reduced manual annotation time by approximately 60%. A multi-scale evaluation system (based on 33% and 66% target area percentiles) revealed a positive correlation between target scale and segmentation performance, with small-scale target segmentation being the primary challenge (IoUs of 0.79 for OCMS and 0.67 for CCS in small scales). Systematic experiments validated Semantic-SAM’s superiority over SAM (mIoU 54.03%) and SAM2 (mIoU 45.33%) in semantic understanding, architectural adaptability, and computational efficiency. The study also found that classification networks within the SAM-SEF framework outperformed specially designed segmentation networks. This demonstrates the effectiveness of a “classification-driven segmentation” paradigm, where lightweight classifiers operating on purified targets from a frozen foundation model can surpass traditional end-to-end segmentation networks in specific, complex domains, offering new technical solutions for remote sensing monitoring.
format Article
id doaj-art-9cc225b59a3244e5a1822c2148b0f7e2
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-9cc225b59a3244e5a1822c2148b0f7e22025-08-20T03:31:27ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-0118148201484210.1109/JSTARS.2025.357628511023034A Decoupled Segmentation-Classification Strategy Based on Semantic-SAM for Precise Semantic Segmentation in Coal Mine AreasLibing Wang0https://orcid.org/0000-0001-8382-8425Jihong Dong1https://orcid.org/0009-0004-7851-3276Lei Wang2Feng Liu3Qingke Wen4Bekir Genc5https://orcid.org/0000-0002-3943-5103School of Environment and Spatial Informatics, China University of Mining and Technology and Engineering Research Center of Mine Ecological Restoration, Ministry of Education, Xuzhou, ChinaSchool of Environment and Spatial Informatics, China University of Mining and Technology, Xuzhou, ChinaChina Coal Society, Beijing, ChinaChina Coal Society, Beijing, ChinaNational Engineering Research Center for Geomatics, Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing, ChinaSchool of Mining Engineering, University of the Witwatersrand, Johannesburg, South AfricaTo address complex semantic segmentation in coal mine areas, this study proposes the SAM-SEF (SAM-based Semantic Enhancement Framework). It integrates Semantic-SAM’s zero-shot segmentation capability with specialized deep learning models through a “segmentation-classification-update” strategy. Semantic-SAM initially provides category-agnostic multi-granularity segmentations; specialized deep learning models then perform semantic understanding of these regions, and a category mapping mechanism generates precise results. Evaluating 18 models, SAM-SEF-ENet-B0 achieved the highest IoUs for Open-pit Coal Mining Sites (OCMS) (93.68%) and Composite Coal-related Sites (CCS) (81.52%), with an mIoU of 87.60%, significantly outperforming traditional models. A Semantic-SAM-based semi-automatic annotation tool reduced manual annotation time by approximately 60%. A multi-scale evaluation system (based on 33% and 66% target area percentiles) revealed a positive correlation between target scale and segmentation performance, with small-scale target segmentation being the primary challenge (IoUs of 0.79 for OCMS and 0.67 for CCS in small scales). Systematic experiments validated Semantic-SAM’s superiority over SAM (mIoU 54.03%) and SAM2 (mIoU 45.33%) in semantic understanding, architectural adaptability, and computational efficiency. The study also found that classification networks within the SAM-SEF framework outperformed specially designed segmentation networks. This demonstrates the effectiveness of a “classification-driven segmentation” paradigm, where lightweight classifiers operating on purified targets from a frozen foundation model can surpass traditional end-to-end segmentation networks in specific, complex domains, offering new technical solutions for remote sensing monitoring.https://ieeexplore.ieee.org/document/11023034/Coal mine site segmentationdeep learningmultiscale evaluationremote sensing image segmentationsemantic-SAMsemi-automatic annotation
spellingShingle Libing Wang
Jihong Dong
Lei Wang
Feng Liu
Qingke Wen
Bekir Genc
A Decoupled Segmentation-Classification Strategy Based on Semantic-SAM for Precise Semantic Segmentation in Coal Mine Areas
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Coal mine site segmentation
deep learning
multiscale evaluation
remote sensing image segmentation
semantic-SAM
semi-automatic annotation
title A Decoupled Segmentation-Classification Strategy Based on Semantic-SAM for Precise Semantic Segmentation in Coal Mine Areas
title_full A Decoupled Segmentation-Classification Strategy Based on Semantic-SAM for Precise Semantic Segmentation in Coal Mine Areas
title_fullStr A Decoupled Segmentation-Classification Strategy Based on Semantic-SAM for Precise Semantic Segmentation in Coal Mine Areas
title_full_unstemmed A Decoupled Segmentation-Classification Strategy Based on Semantic-SAM for Precise Semantic Segmentation in Coal Mine Areas
title_short A Decoupled Segmentation-Classification Strategy Based on Semantic-SAM for Precise Semantic Segmentation in Coal Mine Areas
title_sort decoupled segmentation classification strategy based on semantic sam for precise semantic segmentation in coal mine areas
topic Coal mine site segmentation
deep learning
multiscale evaluation
remote sensing image segmentation
semantic-SAM
semi-automatic annotation
url https://ieeexplore.ieee.org/document/11023034/
work_keys_str_mv AT libingwang adecoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT jihongdong adecoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT leiwang adecoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT fengliu adecoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT qingkewen adecoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT bekirgenc adecoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT libingwang decoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT jihongdong decoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT leiwang decoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT fengliu decoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT qingkewen decoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas
AT bekirgenc decoupledsegmentationclassificationstrategybasedonsemanticsamforprecisesemanticsegmentationincoalmineareas