An explainable CNN-based approach for maritime search and rescue on edge

Gelayol Golcarenarenji; Alaa Mohasseb

doi:10.1109/ICARAI67046.2025.11137856

An explainable CNN-based approach for maritime search and rescue on edge

Gelayol Golcarenarenji^*, Alaa Mohasseb

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

22 Downloads (Pure)

Abstract

Saving lives at sea remains central to maritime search and rescue (SAR) missions. Traditional methods such as aerial and marine visual searches, helicopter, radar and sonar systems are inefficient, costly, and less effective when dealing with small or hard-to-detect objects. Unmanned aerial vehicles (UAVs) have emerged as a powerful tool to improve response times to save more lives. In this work, a custom convolutional neural network (CNN) was developed and trained on the SeaDronesSee dataset to detect stranded people or boats in UAV-captured video over the sea. Our model obtained an accuracy of 68.4 percent the challenging SeaDronesee dataset with real-time performance required for low-powered computers such as Jetson Orin. When deployed on the Jetson AGX Orin platform operating at 50W, the model achieved a speed of 32 frames per second.

Original language	English
Title of host publication	Proceedings of 2025 International Conference Automatics, Robotics and Artificial Intelligence (ICARAI) 13 - 15 June 2025
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781665465663
ISBN (Print)	9781665465670
DOIs	https://doi.org/10.1109/ICARAI67046.2025.11137856
Publication status	Published - 3 Sept 2025
Event	International Conference Automatics, Robotics & Artificial Intelligence - Faculty of Automatics of the Technical Universiy of Sofia, Sozopol, Bulgaria Duration: 13 Jun 2025 → 15 Jun 2025 https://icarai.tu-sofia.bg/

Conference

Conference	International Conference Automatics, Robotics & Artificial Intelligence
Abbreviated title	ICARAI
Country/Territory	Bulgaria
City	Sozopol
Period	13/06/25 → 15/06/25
Internet address	https://icarai.tu-sofia.bg/

Keywords

maritime search and rescue
edge device
explainability
YOLOv8
deep learning

Access to Document

10.1109/ICARAI67046.2025.11137856

An Explainable CNN-based ApproachAccepted author manuscript (Post-print), 1.76 MBLicence: CC BY

Cite this

@inproceedings{5e624537ad0c44d6939de1bce7e3114b,

title = "An explainable CNN-based approach for maritime search and rescue on edge",

abstract = "Saving lives at sea remains central to maritime search and rescue (SAR) missions. Traditional methods such as aerial and marine visual searches, helicopter, radar and sonar systems are inefficient, costly, and less effective when dealing with small or hard-to-detect objects. Unmanned aerial vehicles (UAVs) have emerged as a powerful tool to improve response times to save more lives. In this work, a custom convolutional neural network (CNN) was developed and trained on the SeaDronesSee dataset to detect stranded people or boats in UAV-captured video over the sea. Our model obtained an accuracy of 68.4 percent the challenging SeaDronesee dataset with real-time performance required for low-powered computers such as Jetson Orin. When deployed on the Jetson AGX Orin platform operating at 50W, the model achieved a speed of 32 frames per second.",

keywords = "maritime search and rescue, edge device, explainability, YOLOv8, deep learning",

author = "Gelayol Golcarenarenji and Alaa Mohasseb",

note = "No ISSN; International Conference Automatics, Robotics \& Artificial Intelligence, ICARAI ; Conference date: 13-06-2025 Through 15-06-2025",

year = "2025",

month = sep,

day = "3",

doi = "10.1109/ICARAI67046.2025.11137856",

language = "English",

isbn = "9781665465670",

booktitle = "Proceedings of 2025 International Conference Automatics, Robotics and Artificial Intelligence (ICARAI) 13 - 15 June 2025",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

address = "United States",

url = "https://icarai.tu-sofia.bg/",

}

Golcarenarenji, G & Mohasseb, A 2025, An explainable CNN-based approach for maritime search and rescue on edge. in Proceedings of 2025 International Conference Automatics, Robotics and Artificial Intelligence (ICARAI) 13 - 15 June 2025. Institute of Electrical and Electronics Engineers Inc., International Conference Automatics, Robotics & Artificial Intelligence, Sozopol, Bulgaria, 13/06/25. https://doi.org/10.1109/ICARAI67046.2025.11137856

An explainable CNN-based approach for maritime search and rescue on edge. / Golcarenarenji, Gelayol ; Mohasseb, Alaa.
Proceedings of 2025 International Conference Automatics, Robotics and Artificial Intelligence (ICARAI) 13 - 15 June 2025. Institute of Electrical and Electronics Engineers Inc., 2025.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - An explainable CNN-based approach for maritime search and rescue on edge

AU - Golcarenarenji, Gelayol

AU - Mohasseb, Alaa

N1 - No ISSN

PY - 2025/9/3

Y1 - 2025/9/3

N2 - Saving lives at sea remains central to maritime search and rescue (SAR) missions. Traditional methods such as aerial and marine visual searches, helicopter, radar and sonar systems are inefficient, costly, and less effective when dealing with small or hard-to-detect objects. Unmanned aerial vehicles (UAVs) have emerged as a powerful tool to improve response times to save more lives. In this work, a custom convolutional neural network (CNN) was developed and trained on the SeaDronesSee dataset to detect stranded people or boats in UAV-captured video over the sea. Our model obtained an accuracy of 68.4 percent the challenging SeaDronesee dataset with real-time performance required for low-powered computers such as Jetson Orin. When deployed on the Jetson AGX Orin platform operating at 50W, the model achieved a speed of 32 frames per second.

AB - Saving lives at sea remains central to maritime search and rescue (SAR) missions. Traditional methods such as aerial and marine visual searches, helicopter, radar and sonar systems are inefficient, costly, and less effective when dealing with small or hard-to-detect objects. Unmanned aerial vehicles (UAVs) have emerged as a powerful tool to improve response times to save more lives. In this work, a custom convolutional neural network (CNN) was developed and trained on the SeaDronesSee dataset to detect stranded people or boats in UAV-captured video over the sea. Our model obtained an accuracy of 68.4 percent the challenging SeaDronesee dataset with real-time performance required for low-powered computers such as Jetson Orin. When deployed on the Jetson AGX Orin platform operating at 50W, the model achieved a speed of 32 frames per second.

KW - maritime search and rescue

KW - edge device

KW - explainability

KW - YOLOv8

KW - deep learning

U2 - 10.1109/ICARAI67046.2025.11137856

DO - 10.1109/ICARAI67046.2025.11137856

M3 - Conference contribution

SN - 9781665465670

BT - Proceedings of 2025 International Conference Automatics, Robotics and Artificial Intelligence (ICARAI) 13 - 15 June 2025

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - International Conference Automatics, Robotics & Artificial Intelligence

Y2 - 13 June 2025 through 15 June 2025

ER -

An explainable CNN-based approach for maritime search and rescue on edge

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this