HSRoadNet: Hard-Swish Activation Function and Improved Squeeze–Excitation Module Network for Road Extraction Using Satellite Remote Sensing Imagery

Road information plays an essential role in many fields. To prevent failed extraction of heterogeneous regions, fracture of extracted roads and others resulted from vehicles and trees when using very high resolution remote sensing images; a remote sensing image road extraction method based on Hard-S...

Full description

Saved in:
Bibliographic Details
Main Authors: Xunqiang Gong, Yingjie Ma, Ailong Ma, Zhaoyang Hou, Meng Zhang, Yanfei Zhong
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/10850767/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1823857160731230208
author Xunqiang Gong
Yingjie Ma
Ailong Ma
Zhaoyang Hou
Meng Zhang
Yanfei Zhong
author_facet Xunqiang Gong
Yingjie Ma
Ailong Ma
Zhaoyang Hou
Meng Zhang
Yanfei Zhong
author_sort Xunqiang Gong
collection DOAJ
description Road information plays an essential role in many fields. To prevent failed extraction of heterogeneous regions, fracture of extracted roads and others resulted from vehicles and trees when using very high resolution remote sensing images; a remote sensing image road extraction method based on Hard-Swish Squeeze&#x2013;Excitation RoadNet is proposed in this article. First, road extraction task is divided into three correlated subtasks to reduce the impact of vehicles and trees in road extracting. Second, a normalization layer is adopted to prevent gradient levels from vanishing and exploring and avoid fracture of the extracted road. Then, adopting Hard-Swish activation function to improve the accuracy of road extracting, and then finally, using the improved squeeze&#x2013;excitation module to make the trained net a full use of the characteristic information of the road that do not increase excessive capacity. Comparison experimental results indicate that, in various indicators, the proposed method performs serviceably, it, respectively, increased by 16.8&#x0025;, 2.2&#x0025;, 1.5&#x0025;, and 8.5&#x0025; over the suboptimal in <italic>F</italic>-score, global accuracy, class average accuracy, and recall ratio. The mean intersection over union (MIoU) value of the proposed method was the suboptimum with a disparity of 0.2&#x0025; from the optimal. Ablation experiments show that the proposed method performs best in various indices, and the global accuracy, MIoU, class average accuracy, and recall rate are improved by 0.5&#x0025;, 0.1&#x0025;, 0.5&#x0025;, and 0.2&#x0025;, respectively, compared with the suboptimal method. The <italic>F</italic>-score is suboptimal, with a 0.3&#x0025; difference from the best.
format Article
id doaj-art-cac276d07faa4671970865f266b5940e
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-cac276d07faa4671970865f266b5940e2025-02-12T00:00:59ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-01184907492010.1109/JSTARS.2025.353319610850767HSRoadNet: Hard-Swish Activation Function and Improved Squeeze&#x2013;Excitation Module Network for Road Extraction Using Satellite Remote Sensing ImageryXunqiang Gong0https://orcid.org/0000-0002-6700-2241Yingjie Ma1https://orcid.org/0009-0004-1694-3932Ailong Ma2https://orcid.org/0000-0003-3692-6473Zhaoyang Hou3https://orcid.org/0000-0001-9894-4795Meng Zhang4Yanfei Zhong5https://orcid.org/0000-0001-9446-5850Key Laboratory of Mine Environmental Monitoring and Improving around Poyang Lake, Ministry of Natural Resources, East China University of Technology, Nanchang, ChinaKey Laboratory of Mine Environmental Monitoring and Improving around Poyang Lake, Ministry of Natural Resources, East China University of Technology, Nanchang, ChinaState Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University, Wuhan, ChinaKey Laboratory of Mine Environmental Monitoring and Improving around Poyang Lake, Ministry of Natural Resources, East China University of Technology, Nanchang, ChinaJiangxi Academy of Eco-Environmental Sciences and Planning, Nanchang, ChinaState Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University, Wuhan, ChinaRoad information plays an essential role in many fields. To prevent failed extraction of heterogeneous regions, fracture of extracted roads and others resulted from vehicles and trees when using very high resolution remote sensing images; a remote sensing image road extraction method based on Hard-Swish Squeeze&#x2013;Excitation RoadNet is proposed in this article. First, road extraction task is divided into three correlated subtasks to reduce the impact of vehicles and trees in road extracting. Second, a normalization layer is adopted to prevent gradient levels from vanishing and exploring and avoid fracture of the extracted road. Then, adopting Hard-Swish activation function to improve the accuracy of road extracting, and then finally, using the improved squeeze&#x2013;excitation module to make the trained net a full use of the characteristic information of the road that do not increase excessive capacity. Comparison experimental results indicate that, in various indicators, the proposed method performs serviceably, it, respectively, increased by 16.8&#x0025;, 2.2&#x0025;, 1.5&#x0025;, and 8.5&#x0025; over the suboptimal in <italic>F</italic>-score, global accuracy, class average accuracy, and recall ratio. The mean intersection over union (MIoU) value of the proposed method was the suboptimum with a disparity of 0.2&#x0025; from the optimal. Ablation experiments show that the proposed method performs best in various indices, and the global accuracy, MIoU, class average accuracy, and recall rate are improved by 0.5&#x0025;, 0.1&#x0025;, 0.5&#x0025;, and 0.2&#x0025;, respectively, compared with the suboptimal method. The <italic>F</italic>-score is suboptimal, with a 0.3&#x0025; difference from the best.https://ieeexplore.ieee.org/document/10850767/Activation functionimproved squeeze-excitation (SE) modulenormalization layerroad extraction of remote sensing imageRoadNet
spellingShingle Xunqiang Gong
Yingjie Ma
Ailong Ma
Zhaoyang Hou
Meng Zhang
Yanfei Zhong
HSRoadNet: Hard-Swish Activation Function and Improved Squeeze&#x2013;Excitation Module Network for Road Extraction Using Satellite Remote Sensing Imagery
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Activation function
improved squeeze-excitation (SE) module
normalization layer
road extraction of remote sensing image
RoadNet
title HSRoadNet: Hard-Swish Activation Function and Improved Squeeze&#x2013;Excitation Module Network for Road Extraction Using Satellite Remote Sensing Imagery
title_full HSRoadNet: Hard-Swish Activation Function and Improved Squeeze&#x2013;Excitation Module Network for Road Extraction Using Satellite Remote Sensing Imagery
title_fullStr HSRoadNet: Hard-Swish Activation Function and Improved Squeeze&#x2013;Excitation Module Network for Road Extraction Using Satellite Remote Sensing Imagery
title_full_unstemmed HSRoadNet: Hard-Swish Activation Function and Improved Squeeze&#x2013;Excitation Module Network for Road Extraction Using Satellite Remote Sensing Imagery
title_short HSRoadNet: Hard-Swish Activation Function and Improved Squeeze&#x2013;Excitation Module Network for Road Extraction Using Satellite Remote Sensing Imagery
title_sort hsroadnet hard swish activation function and improved squeeze x2013 excitation module network for road extraction using satellite remote sensing imagery
topic Activation function
improved squeeze-excitation (SE) module
normalization layer
road extraction of remote sensing image
RoadNet
url https://ieeexplore.ieee.org/document/10850767/
work_keys_str_mv AT xunqianggong hsroadnethardswishactivationfunctionandimprovedsqueezex2013excitationmodulenetworkforroadextractionusingsatelliteremotesensingimagery
AT yingjiema hsroadnethardswishactivationfunctionandimprovedsqueezex2013excitationmodulenetworkforroadextractionusingsatelliteremotesensingimagery
AT ailongma hsroadnethardswishactivationfunctionandimprovedsqueezex2013excitationmodulenetworkforroadextractionusingsatelliteremotesensingimagery
AT zhaoyanghou hsroadnethardswishactivationfunctionandimprovedsqueezex2013excitationmodulenetworkforroadextractionusingsatelliteremotesensingimagery
AT mengzhang hsroadnethardswishactivationfunctionandimprovedsqueezex2013excitationmodulenetworkforroadextractionusingsatelliteremotesensingimagery
AT yanfeizhong hsroadnethardswishactivationfunctionandimprovedsqueezex2013excitationmodulenetworkforroadextractionusingsatelliteremotesensingimagery