Training marine species object detectors with synthetic images and unsupervised domain adaptation

Visual surveys by autonomous underwater vehicles (AUVs) and other underwater platforms provide a valuable method for analysing and understanding the benthic environment. Scientists can measure the presence and abundance of benthic species by manually annotating survey images with online annotation s...

Full description

Saved in:
Bibliographic Details
Main Authors: Heather Doig, Oscar Pizarro, Stefan Williams
Format: Article
Language:English
Published: Frontiers Media S.A. 2025-07-01
Series:Frontiers in Marine Science
Subjects:
Online Access:https://www.frontiersin.org/articles/10.3389/fmars.2025.1581778/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850117760274661376
author Heather Doig
Oscar Pizarro
Stefan Williams
author_facet Heather Doig
Oscar Pizarro
Stefan Williams
author_sort Heather Doig
collection DOAJ
description Visual surveys by autonomous underwater vehicles (AUVs) and other underwater platforms provide a valuable method for analysing and understanding the benthic environment. Scientists can measure the presence and abundance of benthic species by manually annotating survey images with online annotation software or other tools. Neural network object detectors can reduce the effort involved in this process by locating and classifying species of interest in the images. However, accurate object detectors often rely on large numbers of annotated training images which are not currently available for many marine applications. To address this issue, we propose a novel pipeline for generating large amounts of synthetic annotated training data for a species of interest using 3D modelling and rendering software. The detector is trained with synthetic images and annotations along with real unlabelled images to improve performance through domain adaptation. Our method is demonstrated on a sea urchin detector trained only with synthetic data, achieving a performance slightly lower than an equivalent detector trained with manually labelled real images (AP50 of 84.3 vs 92.3). Using realistic synthetic data for species or objects with few or no annotations is a promising approach to reducing the manual effort required to analyse imaging survey data.
format Article
id doaj-art-620e5f18461645ffa5c79b861b374921
institution OA Journals
issn 2296-7745
language English
publishDate 2025-07-01
publisher Frontiers Media S.A.
record_format Article
series Frontiers in Marine Science
spelling doaj-art-620e5f18461645ffa5c79b861b3749212025-08-20T02:36:02ZengFrontiers Media S.A.Frontiers in Marine Science2296-77452025-07-011210.3389/fmars.2025.15817781581778Training marine species object detectors with synthetic images and unsupervised domain adaptationHeather Doig0Oscar Pizarro1Stefan Williams2Australian Centre for Robotics, University of Sydney, Sydney, NSW, AustraliaDepartment of Marine Technology, Norwegian University of Science and Technology, Trondheim, NorwayAustralian Centre for Robotics, University of Sydney, Sydney, NSW, AustraliaVisual surveys by autonomous underwater vehicles (AUVs) and other underwater platforms provide a valuable method for analysing and understanding the benthic environment. Scientists can measure the presence and abundance of benthic species by manually annotating survey images with online annotation software or other tools. Neural network object detectors can reduce the effort involved in this process by locating and classifying species of interest in the images. However, accurate object detectors often rely on large numbers of annotated training images which are not currently available for many marine applications. To address this issue, we propose a novel pipeline for generating large amounts of synthetic annotated training data for a species of interest using 3D modelling and rendering software. The detector is trained with synthetic images and annotations along with real unlabelled images to improve performance through domain adaptation. Our method is demonstrated on a sea urchin detector trained only with synthetic data, achieving a performance slightly lower than an equivalent detector trained with manually labelled real images (AP50 of 84.3 vs 92.3). Using realistic synthetic data for species or objects with few or no annotations is a promising approach to reducing the manual effort required to analyse imaging survey data.https://www.frontiersin.org/articles/10.3389/fmars.2025.1581778/fullbenthic monitoringobject detectionunsupervised domain adaptationsynthetic imagesbenthic imaging
spellingShingle Heather Doig
Oscar Pizarro
Stefan Williams
Training marine species object detectors with synthetic images and unsupervised domain adaptation
Frontiers in Marine Science
benthic monitoring
object detection
unsupervised domain adaptation
synthetic images
benthic imaging
title Training marine species object detectors with synthetic images and unsupervised domain adaptation
title_full Training marine species object detectors with synthetic images and unsupervised domain adaptation
title_fullStr Training marine species object detectors with synthetic images and unsupervised domain adaptation
title_full_unstemmed Training marine species object detectors with synthetic images and unsupervised domain adaptation
title_short Training marine species object detectors with synthetic images and unsupervised domain adaptation
title_sort training marine species object detectors with synthetic images and unsupervised domain adaptation
topic benthic monitoring
object detection
unsupervised domain adaptation
synthetic images
benthic imaging
url https://www.frontiersin.org/articles/10.3389/fmars.2025.1581778/full
work_keys_str_mv AT heatherdoig trainingmarinespeciesobjectdetectorswithsyntheticimagesandunsuperviseddomainadaptation
AT oscarpizarro trainingmarinespeciesobjectdetectorswithsyntheticimagesandunsuperviseddomainadaptation
AT stefanwilliams trainingmarinespeciesobjectdetectorswithsyntheticimagesandunsuperviseddomainadaptation