Swin-GAT Fusion Dual-Stream Hybrid Network for High-Resolution Remote Sensing Road Extraction

This paper introduces a novel dual-stream collaborative architecture for remote sensing road segmentation, designed to overcome multi-scale feature conflicts, limited dynamic adaptability, and compromised topological integrity. Our network employs a parallel “local–global” encoding scheme: the local...

Full description

Saved in:
Bibliographic Details
Main Authors: Hongkai Zhang, Hongxuan Yuan, Minghao Shao, Junxin Wang, Suhong Liu
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/17/13/2238
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850115907849814016
author Hongkai Zhang
Hongxuan Yuan
Minghao Shao
Junxin Wang
Suhong Liu
author_facet Hongkai Zhang
Hongxuan Yuan
Minghao Shao
Junxin Wang
Suhong Liu
author_sort Hongkai Zhang
collection DOAJ
description This paper introduces a novel dual-stream collaborative architecture for remote sensing road segmentation, designed to overcome multi-scale feature conflicts, limited dynamic adaptability, and compromised topological integrity. Our network employs a parallel “local–global” encoding scheme: the local stream uses depth-wise separable convolutions to capture fine-grained details, while the global stream integrates a Swin-Transformer with a graph-attention module (Swin-GAT) to model long-range contextual and topological relationships. By decoupling detailed feature extraction from global context modeling, the proposed framework more faithfully represents complex road structures. Comprehensive experiments on multiple aerial datasets demonstrate that our approach outperforms conventional baselines—especially under shadow occlusion and for thin-road delineation—while achieving real-time inference at 31 FPS. Ablation studies further confirm the critical roles of the Swin Transformer and GAT components in preserving topological continuity. Overall, this dual-stream dynamic-fusion network sets a new benchmark for remote sensing road extraction and holds promise for real-world, real-time applications.
format Article
id doaj-art-8008a9c3b90a42d3aa098348dfbf05d1
institution OA Journals
issn 2072-4292
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj-art-8008a9c3b90a42d3aa098348dfbf05d12025-08-20T02:36:27ZengMDPI AGRemote Sensing2072-42922025-06-011713223810.3390/rs17132238Swin-GAT Fusion Dual-Stream Hybrid Network for High-Resolution Remote Sensing Road ExtractionHongkai Zhang0Hongxuan Yuan1Minghao Shao2Junxin Wang3Suhong Liu4Faculty of Arts and Sciences, Beijing Normal University, Zhuhai 519087, ChinaFaculty of Arts and Sciences, Beijing Normal University, Zhuhai 519087, ChinaFaculty of Arts and Sciences, Beijing Normal University, Zhuhai 519087, ChinaFaculty of Arts and Sciences, Beijing Normal University, Zhuhai 519087, ChinaDepartment of Geography, Faculty of Arts and Sciences, Beijing Normal University, Zhuhai 519087, ChinaThis paper introduces a novel dual-stream collaborative architecture for remote sensing road segmentation, designed to overcome multi-scale feature conflicts, limited dynamic adaptability, and compromised topological integrity. Our network employs a parallel “local–global” encoding scheme: the local stream uses depth-wise separable convolutions to capture fine-grained details, while the global stream integrates a Swin-Transformer with a graph-attention module (Swin-GAT) to model long-range contextual and topological relationships. By decoupling detailed feature extraction from global context modeling, the proposed framework more faithfully represents complex road structures. Comprehensive experiments on multiple aerial datasets demonstrate that our approach outperforms conventional baselines—especially under shadow occlusion and for thin-road delineation—while achieving real-time inference at 31 FPS. Ablation studies further confirm the critical roles of the Swin Transformer and GAT components in preserving topological continuity. Overall, this dual-stream dynamic-fusion network sets a new benchmark for remote sensing road extraction and holds promise for real-world, real-time applications.https://www.mdpi.com/2072-4292/17/13/2238remote sensing road extractiondual-stream networkSwin-GATdepthwise separable convolutiondynamic feature fusion
spellingShingle Hongkai Zhang
Hongxuan Yuan
Minghao Shao
Junxin Wang
Suhong Liu
Swin-GAT Fusion Dual-Stream Hybrid Network for High-Resolution Remote Sensing Road Extraction
Remote Sensing
remote sensing road extraction
dual-stream network
Swin-GAT
depthwise separable convolution
dynamic feature fusion
title Swin-GAT Fusion Dual-Stream Hybrid Network for High-Resolution Remote Sensing Road Extraction
title_full Swin-GAT Fusion Dual-Stream Hybrid Network for High-Resolution Remote Sensing Road Extraction
title_fullStr Swin-GAT Fusion Dual-Stream Hybrid Network for High-Resolution Remote Sensing Road Extraction
title_full_unstemmed Swin-GAT Fusion Dual-Stream Hybrid Network for High-Resolution Remote Sensing Road Extraction
title_short Swin-GAT Fusion Dual-Stream Hybrid Network for High-Resolution Remote Sensing Road Extraction
title_sort swin gat fusion dual stream hybrid network for high resolution remote sensing road extraction
topic remote sensing road extraction
dual-stream network
Swin-GAT
depthwise separable convolution
dynamic feature fusion
url https://www.mdpi.com/2072-4292/17/13/2238
work_keys_str_mv AT hongkaizhang swingatfusiondualstreamhybridnetworkforhighresolutionremotesensingroadextraction
AT hongxuanyuan swingatfusiondualstreamhybridnetworkforhighresolutionremotesensingroadextraction
AT minghaoshao swingatfusiondualstreamhybridnetworkforhighresolutionremotesensingroadextraction
AT junxinwang swingatfusiondualstreamhybridnetworkforhighresolutionremotesensingroadextraction
AT suhongliu swingatfusiondualstreamhybridnetworkforhighresolutionremotesensingroadextraction