Dual-stream detection and segmentation framework for vision based unmanned ground vehicle pothole perception on unstructured roads

Abstract Reliable perception of road surface damage is considered essential for ensuring safe and autonomous operation of Unmanned Ground Vehicles (UGVs) on unstructured roads, where irregular textures, blurred boundaries, and environmental interference are often encountered. To overcome the limitat...

Full description

Saved in:
Bibliographic Details
Main Authors: Chenyuan He, He Yang, Zhouyu Zhang, Hai Wang, Yingfeng Cai, Long Chen, Can Zhong, Yiqun Zhang
Format: Article
Language:English
Published: Elsevier 2025-08-01
Series:Journal of King Saud University: Computer and Information Sciences
Subjects:
Online Access:https://doi.org/10.1007/s44443-025-00236-7
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849225847903354880
author Chenyuan He
He Yang
Zhouyu Zhang
Hai Wang
Yingfeng Cai
Long Chen
Can Zhong
Yiqun Zhang
author_facet Chenyuan He
He Yang
Zhouyu Zhang
Hai Wang
Yingfeng Cai
Long Chen
Can Zhong
Yiqun Zhang
author_sort Chenyuan He
collection DOAJ
description Abstract Reliable perception of road surface damage is considered essential for ensuring safe and autonomous operation of Unmanned Ground Vehicles (UGVs) on unstructured roads, where irregular textures, blurred boundaries, and environmental interference are often encountered. To overcome the limitations of existing methods, a dual-stream detection–segmentation framework is presented, in which object-level localization and pixel-level boundary extraction are decoupled and independently optimized. Specifically, the detection stream adopts an enhanced YOLOv10+ network equipped with a frequency-aware fusion module (FreqFusion) in the neck to improve semantic–spatial alignment and robustness to texture variation. The segmentation stream introduces GAL-DeepLabv3+plus, which integrates a Dense Atrous Spatial Pyramid Pooling (DenseASPP) module and a Graph Attention Layer (GAL) into the standard DeepLabv3+ architecture, thereby enhancing contextual reasoning and boundary refinement. Extensive experiments are conducted on a self-constructed dataset comprising 3,000 annotated images of unstructured roads. Quantitative results demonstrate that the proposed framework achieves an F1-score of 93.0% and recall of 93.2% in detection, and an IoU of 92.5% and F1-score of 96.1% in segmentation. Ablation studies and environmental condition tests further confirm its component effectiveness and real-world applicability. Compared with state-of-the-art baselines such as YOLOv8 and DeepLabv3+, the proposed method achieves +5.1% and +3.4% improvements in detection F1-score and segmentation IoU, respectively, highlighting its superior performance and practical value in complex terrain perception scenarios.
format Article
id doaj-art-8099b17420f0439fba235a22d8d98ff7
institution Kabale University
issn 1319-1578
2213-1248
language English
publishDate 2025-08-01
publisher Elsevier
record_format Article
series Journal of King Saud University: Computer and Information Sciences
spelling doaj-art-8099b17420f0439fba235a22d8d98ff72025-08-24T11:53:33ZengElsevierJournal of King Saud University: Computer and Information Sciences1319-15782213-12482025-08-0137713110.1007/s44443-025-00236-7Dual-stream detection and segmentation framework for vision based unmanned ground vehicle pothole perception on unstructured roadsChenyuan He0He Yang1Zhouyu Zhang2Hai Wang3Yingfeng Cai4Long Chen5Can Zhong6Yiqun Zhang7School of Automotive and Traffic Engineering, Jiangsu UniversitySchool of Automotive and Traffic Engineering, Jiangsu UniversitySchool of Automotive and Traffic Engineering, Jiangsu UniversitySchool of Automotive and Traffic Engineering, Jiangsu UniversityAutomotive Engineering Research Institute, Jiangsu UniversityAutomotive Engineering Research Institute, Jiangsu UniversityBeijing Engineering Research Center of Aerial Intelligent Remote Sensing EquipmentsTopXGun (Nanjing) Robotics Company LimitedAbstract Reliable perception of road surface damage is considered essential for ensuring safe and autonomous operation of Unmanned Ground Vehicles (UGVs) on unstructured roads, where irregular textures, blurred boundaries, and environmental interference are often encountered. To overcome the limitations of existing methods, a dual-stream detection–segmentation framework is presented, in which object-level localization and pixel-level boundary extraction are decoupled and independently optimized. Specifically, the detection stream adopts an enhanced YOLOv10+ network equipped with a frequency-aware fusion module (FreqFusion) in the neck to improve semantic–spatial alignment and robustness to texture variation. The segmentation stream introduces GAL-DeepLabv3+plus, which integrates a Dense Atrous Spatial Pyramid Pooling (DenseASPP) module and a Graph Attention Layer (GAL) into the standard DeepLabv3+ architecture, thereby enhancing contextual reasoning and boundary refinement. Extensive experiments are conducted on a self-constructed dataset comprising 3,000 annotated images of unstructured roads. Quantitative results demonstrate that the proposed framework achieves an F1-score of 93.0% and recall of 93.2% in detection, and an IoU of 92.5% and F1-score of 96.1% in segmentation. Ablation studies and environmental condition tests further confirm its component effectiveness and real-world applicability. Compared with state-of-the-art baselines such as YOLOv8 and DeepLabv3+, the proposed method achieves +5.1% and +3.4% improvements in detection F1-score and segmentation IoU, respectively, highlighting its superior performance and practical value in complex terrain perception scenarios.https://doi.org/10.1007/s44443-025-00236-7Unmanned Ground Vehicle (UGV)Pothole perceptionDeep learningMulti-task learningDetection and segmentation
spellingShingle Chenyuan He
He Yang
Zhouyu Zhang
Hai Wang
Yingfeng Cai
Long Chen
Can Zhong
Yiqun Zhang
Dual-stream detection and segmentation framework for vision based unmanned ground vehicle pothole perception on unstructured roads
Journal of King Saud University: Computer and Information Sciences
Unmanned Ground Vehicle (UGV)
Pothole perception
Deep learning
Multi-task learning
Detection and segmentation
title Dual-stream detection and segmentation framework for vision based unmanned ground vehicle pothole perception on unstructured roads
title_full Dual-stream detection and segmentation framework for vision based unmanned ground vehicle pothole perception on unstructured roads
title_fullStr Dual-stream detection and segmentation framework for vision based unmanned ground vehicle pothole perception on unstructured roads
title_full_unstemmed Dual-stream detection and segmentation framework for vision based unmanned ground vehicle pothole perception on unstructured roads
title_short Dual-stream detection and segmentation framework for vision based unmanned ground vehicle pothole perception on unstructured roads
title_sort dual stream detection and segmentation framework for vision based unmanned ground vehicle pothole perception on unstructured roads
topic Unmanned Ground Vehicle (UGV)
Pothole perception
Deep learning
Multi-task learning
Detection and segmentation
url https://doi.org/10.1007/s44443-025-00236-7
work_keys_str_mv AT chenyuanhe dualstreamdetectionandsegmentationframeworkforvisionbasedunmannedgroundvehiclepotholeperceptiononunstructuredroads
AT heyang dualstreamdetectionandsegmentationframeworkforvisionbasedunmannedgroundvehiclepotholeperceptiononunstructuredroads
AT zhouyuzhang dualstreamdetectionandsegmentationframeworkforvisionbasedunmannedgroundvehiclepotholeperceptiononunstructuredroads
AT haiwang dualstreamdetectionandsegmentationframeworkforvisionbasedunmannedgroundvehiclepotholeperceptiononunstructuredroads
AT yingfengcai dualstreamdetectionandsegmentationframeworkforvisionbasedunmannedgroundvehiclepotholeperceptiononunstructuredroads
AT longchen dualstreamdetectionandsegmentationframeworkforvisionbasedunmannedgroundvehiclepotholeperceptiononunstructuredroads
AT canzhong dualstreamdetectionandsegmentationframeworkforvisionbasedunmannedgroundvehiclepotholeperceptiononunstructuredroads
AT yiqunzhang dualstreamdetectionandsegmentationframeworkforvisionbasedunmannedgroundvehiclepotholeperceptiononunstructuredroads