地理空间信息2025，Vol.23Issue(5)：1-4,38,5.DOI:10.3969/j.issn.1672-4623.2025.05.001
DAR-Net:双空间注意力与边界细化的遥感影像滑坡提取网络
DAR-Net:A Dual Spatial Attention and Boundary Refinement Network for Landslide Extraction from Remote Sensing Images
摘要
Abstract
Rapid and accurate landslide extraction from remote sensing images is of great significance for disaster prevention,reduction,and emergency response.Existing remote sensing image landslide extraction algorithms have problems such as blurred boundaries,internal holes,and overall instance discontinuity.In response to these issues,we proposed a landslide extraction network called DAR-Net.This network introduces the dual spatial attention module and coarse-to-fine boundary refinement module on the basis of U-Net,which are respectively used to capture global and local features and refine boundaries.Compared with the semantic segmentation algorithms,this algorithm produces the best performance on the open source dataset.关键词
滑坡提取/遥感影像/全局与局部注意力/边界细化Key words
landslide extraction/remote sensing image/global and local attention/boundary refinement分类
天文与地球科学引用本文复制引用
严炳杨,冯志荣,鲍灵辉,黄智勇,唐俊,郭寅虎,张晋博,陈敏..DAR-Net:双空间注意力与边界细化的遥感影像滑坡提取网络[J].地理空间信息,2025,23(5):1-4,38,5.基金项目
西成铁路客运专线陕西有限责任公司科技研究重点开发计划资助项目(西康高铁合(2021)21号) （西康高铁合(2021）
四川省科技计划资助项目(2023NSFSC0247). （2023NSFSC0247）
