Vessel Detection from Optical Remote Sensing Images with Deep Learning Methods

Vessel detection from remote sensing images is becoming exponentially crucial component in marine surveillance applications including maritime traffic control, anti-illegal fishing applications, oil discharge control, marine pollution and safety. Applying deep learning methods to vessel detection ap...

Full description

Saved in:
Bibliographic Details
Published in2023 10th International Conference on Recent Advances in Air and Space Technologies (RAST) pp. 1 - 5
Main Authors Buyukkanber, Furkan, Yanalak, Mustafa, Musaoglu, Nebiye
Format Conference Proceeding
LanguageEnglish
Published IEEE 07.06.2023
Subjects
Adaptive optics
convolutional neural networks
Deep learning
Optical imaging
optical remote sensing images
ship dataset
Space technology
Surveillance
Traffic control
Training
vessel detection
YOLO series
Online AccessGet full text

Cover

More Information
Summary:Vessel detection from remote sensing images is becoming exponentially crucial component in marine surveillance applications including maritime traffic control, anti-illegal fishing applications, oil discharge control, marine pollution and safety. Applying deep learning methods to vessel detection applications ineluctably improve the detection results and overcome unforeseen errors that could be made by analysts. Publicly available datasets play vital role for development and evaluation process of deep learning models. In this paper, open source DOTA dataset has been revised and trained with single-staged deep learning methods. The results show that YOLOv8 model has the most efficient value on detecting ships and fastest to detect instances from given test images on inference.
DOI:10.1109/RAST57548.2023.10197948