YOLOv8-Based Deep Learning System for Underground Pipeline Detection Using GPR B-Scans

Maha Yaghi; Taimur Hassan; Ahmad Yahia Al-Hashaikeh; Mohammad Faek Abdallah; Mohammed Ghazal; Mohammed Mahmoud; Jawad Yousaf

doi:10.1007/978-3-032-19978-2_24

YOLOv8-Based Deep Learning System for Underground Pipeline Detection Using GPR B-Scans

Maha Yaghi
, Taimur Hassan
, Ahmad Yahia Al-Hashaikeh
, Mohammad Faek Abdallah
, Mohammed Ghazal
, Mohammed Mahmoud
, Jawad Yousaf^*

^*Corresponding author for this work

Abu Dhabi University
Liwa University
Emirates Global Aluminum

Research output: Chapter in Book/Published conference output › Conference publication

Abstract

This study presents a smart deep learning integrated system for the detection and localization of different types of underground pipelines using ground penetration radar (GPR) B-scan images. The presence of a pipeline is determined from the created hyperbolic pattern in the scanned image. A dataset of around 2900 GPR B-scans of more than 1500 water and gas pipeline targets in different terrains is annotated using Roboflow. The dataset is trained using the YOLO (You Only Look Once) v8 model for the different configurations for the effective detection of pipelines based on their distinct hyperbolic anomalies in GPR scans. The optimized model effectively detects and localizes the multiple pipelines with a precision and recall of 95.2% and 95.1%, respectively. The findings show that combining radar sensing with deep learning greatly improves detection accuracy to localize a variety of pipelines while lowering costs and efforts associated with manual inspections.

Original language	English
Title of host publication	Internet of Things, Smart Spaces, and Next Generation Networks and Systems
Subtitle of host publication	25th International Conference, NEW2AN 2025, and 18th Conference, ruSMART 2025, Abu Dhabi, United Arab Emirates, November 10–12, 2025, Proceedings, Part I
Editors	Yevgeni Koucheryavy, Neeraj Kumar, Ammar Muthanna, Sachin Sharma
Publisher	Springer, Cham
Pages	332-341
Number of pages	10
ISBN (Electronic)	9783032199782
ISBN (Print)	9783032199775 (pbk)
DOIs	https://doi.org/10.1007/978-3-032-19978-2_24
Publication status	Published - 11 May 2026
Event	25th International Conference on Next Generation Wired/Wireless Networks and Systems, NEW2AN 2025 and the 18th Conference on Internet of Things and Smart Spaces, ruSMART 2025 - Abu Dhabi, United Arab Emirates Duration: 10 Nov 2025 → 12 Nov 2025

Publication series

Name	Lecture Notes in Computer Science (LNCS)
Volume	16461
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	25th International Conference on Next Generation Wired/Wireless Networks and Systems, NEW2AN 2025 and the 18th Conference on Internet of Things and Smart Spaces, ruSMART 2025
Country/Territory	United Arab Emirates
City	Abu Dhabi
Period	10/11/25 → 12/11/25

Keywords

Deep learning
Ground Penetrating Radar (GPR)
Underground pipeline
YOLOv8

Access to Document

10.1007/978-3-032-19978-2_24

Cite this

Yaghi, M., Hassan, T., Al-Hashaikeh, A. Y., Abdallah, M. F., Ghazal, M., Mahmoud, M., & Yousaf, J. (2026). YOLOv8-Based Deep Learning System for Underground Pipeline Detection Using GPR B-Scans. In Y. Koucheryavy, N. Kumar, A. Muthanna, & S. Sharma (Eds.), Internet of Things, Smart Spaces, and Next Generation Networks and Systems: 25th International Conference, NEW2AN 2025, and 18th Conference, ruSMART 2025, Abu Dhabi, United Arab Emirates, November 10–12, 2025, Proceedings, Part I (pp. 332-341). (Lecture Notes in Computer Science (LNCS); Vol. 16461). Springer, Cham. https://doi.org/10.1007/978-3-032-19978-2_24

Yaghi, Maha ; Hassan, Taimur ; Al-Hashaikeh, Ahmad Yahia et al. / YOLOv8-Based Deep Learning System for Underground Pipeline Detection Using GPR B-Scans. Internet of Things, Smart Spaces, and Next Generation Networks and Systems: 25th International Conference, NEW2AN 2025, and 18th Conference, ruSMART 2025, Abu Dhabi, United Arab Emirates, November 10–12, 2025, Proceedings, Part I. editor / Yevgeni Koucheryavy ; Neeraj Kumar ; Ammar Muthanna ; Sachin Sharma. Springer, Cham, 2026. pp. 332-341 (Lecture Notes in Computer Science (LNCS)).

@inproceedings{e28ddcaacd914285afae3a9232bd9c2f,

title = "YOLOv8-Based Deep Learning System for Underground Pipeline Detection Using GPR B-Scans",

abstract = "This study presents a smart deep learning integrated system for the detection and localization of different types of underground pipelines using ground penetration radar (GPR) B-scan images. The presence of a pipeline is determined from the created hyperbolic pattern in the scanned image. A dataset of around 2900 GPR B-scans of more than 1500 water and gas pipeline targets in different terrains is annotated using Roboflow. The dataset is trained using the YOLO (You Only Look Once) v8 model for the different configurations for the effective detection of pipelines based on their distinct hyperbolic anomalies in GPR scans. The optimized model effectively detects and localizes the multiple pipelines with a precision and recall of 95.2\% and 95.1\%, respectively. The findings show that combining radar sensing with deep learning greatly improves detection accuracy to localize a variety of pipelines while lowering costs and efforts associated with manual inspections.",

keywords = "Deep learning, Ground Penetrating Radar (GPR), Underground pipeline, YOLOv8",

author = "Maha Yaghi and Taimur Hassan and Al-Hashaikeh, \{Ahmad Yahia\} and Abdallah, \{Mohammad Faek\} and Mohammed Ghazal and Mohammed Mahmoud and Jawad Yousaf",

year = "2026",

month = may,

day = "11",

doi = "10.1007/978-3-032-19978-2\_24",

language = "English",

isbn = "9783032199775 (pbk)",

series = "Lecture Notes in Computer Science (LNCS)",

publisher = "Springer, Cham",

pages = "332--341",

editor = "Yevgeni Koucheryavy and Neeraj Kumar and Ammar Muthanna and Sachin Sharma",

booktitle = "Internet of Things, Smart Spaces, and Next Generation Networks and Systems",

note = "25th International Conference on Next Generation Wired/Wireless Networks and Systems, NEW2AN 2025 and the 18th Conference on Internet of Things and Smart Spaces, ruSMART 2025 ; Conference date: 10-11-2025 Through 12-11-2025",

}

Yaghi, M, Hassan, T, Al-Hashaikeh, AY, Abdallah, MF, Ghazal, M, Mahmoud, M & Yousaf, J 2026, YOLOv8-Based Deep Learning System for Underground Pipeline Detection Using GPR B-Scans. in Y Koucheryavy, N Kumar, A Muthanna & S Sharma (eds), Internet of Things, Smart Spaces, and Next Generation Networks and Systems: 25th International Conference, NEW2AN 2025, and 18th Conference, ruSMART 2025, Abu Dhabi, United Arab Emirates, November 10–12, 2025, Proceedings, Part I. Lecture Notes in Computer Science (LNCS), vol. 16461, Springer, Cham, pp. 332-341, 25th International Conference on Next Generation Wired/Wireless Networks and Systems, NEW2AN 2025 and the 18th Conference on Internet of Things and Smart Spaces, ruSMART 2025, Abu Dhabi, United Arab Emirates, 10/11/25. https://doi.org/10.1007/978-3-032-19978-2_24

YOLOv8-Based Deep Learning System for Underground Pipeline Detection Using GPR B-Scans. / Yaghi, Maha; Hassan, Taimur; Al-Hashaikeh, Ahmad Yahia et al.
Internet of Things, Smart Spaces, and Next Generation Networks and Systems: 25th International Conference, NEW2AN 2025, and 18th Conference, ruSMART 2025, Abu Dhabi, United Arab Emirates, November 10–12, 2025, Proceedings, Part I. ed. / Yevgeni Koucheryavy; Neeraj Kumar; Ammar Muthanna; Sachin Sharma. Springer, Cham, 2026. p. 332-341 (Lecture Notes in Computer Science (LNCS); Vol. 16461).

Research output: Chapter in Book/Published conference output › Conference publication

TY - GEN

T1 - YOLOv8-Based Deep Learning System for Underground Pipeline Detection Using GPR B-Scans

AU - Yaghi, Maha

AU - Hassan, Taimur

AU - Al-Hashaikeh, Ahmad Yahia

AU - Abdallah, Mohammad Faek

AU - Ghazal, Mohammed

AU - Mahmoud, Mohammed

AU - Yousaf, Jawad

PY - 2026/5/11

Y1 - 2026/5/11

N2 - This study presents a smart deep learning integrated system for the detection and localization of different types of underground pipelines using ground penetration radar (GPR) B-scan images. The presence of a pipeline is determined from the created hyperbolic pattern in the scanned image. A dataset of around 2900 GPR B-scans of more than 1500 water and gas pipeline targets in different terrains is annotated using Roboflow. The dataset is trained using the YOLO (You Only Look Once) v8 model for the different configurations for the effective detection of pipelines based on their distinct hyperbolic anomalies in GPR scans. The optimized model effectively detects and localizes the multiple pipelines with a precision and recall of 95.2% and 95.1%, respectively. The findings show that combining radar sensing with deep learning greatly improves detection accuracy to localize a variety of pipelines while lowering costs and efforts associated with manual inspections.

AB - This study presents a smart deep learning integrated system for the detection and localization of different types of underground pipelines using ground penetration radar (GPR) B-scan images. The presence of a pipeline is determined from the created hyperbolic pattern in the scanned image. A dataset of around 2900 GPR B-scans of more than 1500 water and gas pipeline targets in different terrains is annotated using Roboflow. The dataset is trained using the YOLO (You Only Look Once) v8 model for the different configurations for the effective detection of pipelines based on their distinct hyperbolic anomalies in GPR scans. The optimized model effectively detects and localizes the multiple pipelines with a precision and recall of 95.2% and 95.1%, respectively. The findings show that combining radar sensing with deep learning greatly improves detection accuracy to localize a variety of pipelines while lowering costs and efforts associated with manual inspections.

KW - Deep learning

KW - Ground Penetrating Radar (GPR)

KW - Underground pipeline

KW - YOLOv8

UR - https://link.springer.com/chapter/10.1007/978-3-032-19978-2_24

UR - https://www.scopus.com/pages/publications/105040645144

U2 - 10.1007/978-3-032-19978-2_24

DO - 10.1007/978-3-032-19978-2_24

M3 - Conference publication

AN - SCOPUS:105040645144

SN - 9783032199775 (pbk)

T3 - Lecture Notes in Computer Science (LNCS)

SP - 332

EP - 341

BT - Internet of Things, Smart Spaces, and Next Generation Networks and Systems

A2 - Koucheryavy, Yevgeni

A2 - Kumar, Neeraj

A2 - Muthanna, Ammar

A2 - Sharma, Sachin

PB - Springer, Cham

T2 - 25th International Conference on Next Generation Wired/Wireless Networks and Systems, NEW2AN 2025 and the 18th Conference on Internet of Things and Smart Spaces, ruSMART 2025

Y2 - 10 November 2025 through 12 November 2025

ER -

Yaghi M, Hassan T, Al-Hashaikeh AY, Abdallah MF, Ghazal M, Mahmoud M et al. YOLOv8-Based Deep Learning System for Underground Pipeline Detection Using GPR B-Scans. In Koucheryavy Y, Kumar N, Muthanna A, Sharma S, editors, Internet of Things, Smart Spaces, and Next Generation Networks and Systems: 25th International Conference, NEW2AN 2025, and 18th Conference, ruSMART 2025, Abu Dhabi, United Arab Emirates, November 10–12, 2025, Proceedings, Part I. Springer, Cham. 2026. p. 332-341. (Lecture Notes in Computer Science (LNCS)). Epub 2025 Nov 10. doi: 10.1007/978-3-032-19978-2_24