Extraction of Cropland Based on Multi-Source Remote Sensing and an Improved Version of the Deep Learning-Based Segment Anything Model (SAM)

Fine extraction of cropland parcels is an essential prerequisite for achieving precision agriculture. Remote sensing technology, due to its large-scale and multi-dimensional characteristics, can effectively enhance the efficiency of collecting information on agricultural land parcels. Currently, sem...

Full description

Saved in:
Bibliographic Details
Main Authors: Kunjian Tao, He Li, Chong Huang, Qingsheng Liu, Junyan Zhang, Ruoqi Du
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Agronomy
Subjects:
Online Access:https://www.mdpi.com/2073-4395/15/5/1139
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849327708304048128
author Kunjian Tao
He Li
Chong Huang
Qingsheng Liu
Junyan Zhang
Ruoqi Du
author_facet Kunjian Tao
He Li
Chong Huang
Qingsheng Liu
Junyan Zhang
Ruoqi Du
author_sort Kunjian Tao
collection DOAJ
description Fine extraction of cropland parcels is an essential prerequisite for achieving precision agriculture. Remote sensing technology, due to its large-scale and multi-dimensional characteristics, can effectively enhance the efficiency of collecting information on agricultural land parcels. Currently, semantic segmentation models based on high-resolution remote sensing imagery utilize limited spectral information and rely heavily on a large amount of fine data annotation, while pixel classification models based on medium-to-low-resolution multi-temporal remote sensing imagery are limited by the mixed pixel problem. To address this, the study utilizes GF-2 high-resolution imagery and Sentinel-2 multi-temporal data, in conjunction with the basic image segmentation model SAM, by additionally introducing a prompt generation module (Box module and Auto module) to achieve automatic fine extraction of cropland parcels. The research results indicate the following: (1) The <i>mIoU</i> of SAM with the Box module is 0.711, and the <i>OA</i> is 0.831, showing better performance, while the <i>mIoU</i> of SAM with the Auto module is 0.679, and the <i>OA</i> is 0.81, yielding higher-quality cropland masks; (2) The combination of various prompts (box, point, and mask), along with the hierarchical extraction strategy, can effectively improve the performance of Box module SAM; (3) Employing a more accurate prompt data source can significantly boost model performance. The <i>mIoU</i> of the superior-performing Box module SAM is increased to 0.920, and the <i>OA</i> is raised to 0.958. Overall, the improved SAM, while reducing the demand for mask annotation and model training, can achieve high-precision extraction results for cropland parcels.
format Article
id doaj-art-3ad2dfc042954bcc9cd3bf195ab59a8c
institution Kabale University
issn 2073-4395
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Agronomy
spelling doaj-art-3ad2dfc042954bcc9cd3bf195ab59a8c2025-08-20T03:47:48ZengMDPI AGAgronomy2073-43952025-05-01155113910.3390/agronomy15051139Extraction of Cropland Based on Multi-Source Remote Sensing and an Improved Version of the Deep Learning-Based Segment Anything Model (SAM)Kunjian Tao0He Li1Chong Huang2Qingsheng Liu3Junyan Zhang4Ruoqi Du5State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, ChinaState Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, ChinaState Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, ChinaState Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, ChinaState Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, ChinaState Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, ChinaFine extraction of cropland parcels is an essential prerequisite for achieving precision agriculture. Remote sensing technology, due to its large-scale and multi-dimensional characteristics, can effectively enhance the efficiency of collecting information on agricultural land parcels. Currently, semantic segmentation models based on high-resolution remote sensing imagery utilize limited spectral information and rely heavily on a large amount of fine data annotation, while pixel classification models based on medium-to-low-resolution multi-temporal remote sensing imagery are limited by the mixed pixel problem. To address this, the study utilizes GF-2 high-resolution imagery and Sentinel-2 multi-temporal data, in conjunction with the basic image segmentation model SAM, by additionally introducing a prompt generation module (Box module and Auto module) to achieve automatic fine extraction of cropland parcels. The research results indicate the following: (1) The <i>mIoU</i> of SAM with the Box module is 0.711, and the <i>OA</i> is 0.831, showing better performance, while the <i>mIoU</i> of SAM with the Auto module is 0.679, and the <i>OA</i> is 0.81, yielding higher-quality cropland masks; (2) The combination of various prompts (box, point, and mask), along with the hierarchical extraction strategy, can effectively improve the performance of Box module SAM; (3) Employing a more accurate prompt data source can significantly boost model performance. The <i>mIoU</i> of the superior-performing Box module SAM is increased to 0.920, and the <i>OA</i> is raised to 0.958. Overall, the improved SAM, while reducing the demand for mask annotation and model training, can achieve high-precision extraction results for cropland parcels.https://www.mdpi.com/2073-4395/15/5/1139croplandremote sensingsemantic segmentationSAMprompt generation modulemultimodal satellite data
spellingShingle Kunjian Tao
He Li
Chong Huang
Qingsheng Liu
Junyan Zhang
Ruoqi Du
Extraction of Cropland Based on Multi-Source Remote Sensing and an Improved Version of the Deep Learning-Based Segment Anything Model (SAM)
Agronomy
cropland
remote sensing
semantic segmentation
SAM
prompt generation module
multimodal satellite data
title Extraction of Cropland Based on Multi-Source Remote Sensing and an Improved Version of the Deep Learning-Based Segment Anything Model (SAM)
title_full Extraction of Cropland Based on Multi-Source Remote Sensing and an Improved Version of the Deep Learning-Based Segment Anything Model (SAM)
title_fullStr Extraction of Cropland Based on Multi-Source Remote Sensing and an Improved Version of the Deep Learning-Based Segment Anything Model (SAM)
title_full_unstemmed Extraction of Cropland Based on Multi-Source Remote Sensing and an Improved Version of the Deep Learning-Based Segment Anything Model (SAM)
title_short Extraction of Cropland Based on Multi-Source Remote Sensing and an Improved Version of the Deep Learning-Based Segment Anything Model (SAM)
title_sort extraction of cropland based on multi source remote sensing and an improved version of the deep learning based segment anything model sam
topic cropland
remote sensing
semantic segmentation
SAM
prompt generation module
multimodal satellite data
url https://www.mdpi.com/2073-4395/15/5/1139
work_keys_str_mv AT kunjiantao extractionofcroplandbasedonmultisourceremotesensingandanimprovedversionofthedeeplearningbasedsegmentanythingmodelsam
AT heli extractionofcroplandbasedonmultisourceremotesensingandanimprovedversionofthedeeplearningbasedsegmentanythingmodelsam
AT chonghuang extractionofcroplandbasedonmultisourceremotesensingandanimprovedversionofthedeeplearningbasedsegmentanythingmodelsam
AT qingshengliu extractionofcroplandbasedonmultisourceremotesensingandanimprovedversionofthedeeplearningbasedsegmentanythingmodelsam
AT junyanzhang extractionofcroplandbasedonmultisourceremotesensingandanimprovedversionofthedeeplearningbasedsegmentanythingmodelsam
AT ruoqidu extractionofcroplandbasedonmultisourceremotesensingandanimprovedversionofthedeeplearningbasedsegmentanythingmodelsam