Assisting human annotation of marine images with foundation models

Marine scientists have been leveraging supervised machine learning algorithms to analyze image and video data for nearly two decades. There have been many advances, but the cost of generating expert human annotations to train new models remains extremely high. There is broad recognition both in comp...

Full description

Saved in:
Bibliographic Details
Main Authors: Eric C. Orenstein, Benjamin Woodward, Lonny Lundsten, Kevin Barnard, Brian Schlining, Kakani Katjia
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.2025.1469396/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850067842691497984
author Eric C. Orenstein
Eric C. Orenstein
Benjamin Woodward
Lonny Lundsten
Kevin Barnard
Brian Schlining
Kakani Katjia
author_facet Eric C. Orenstein
Eric C. Orenstein
Benjamin Woodward
Lonny Lundsten
Kevin Barnard
Brian Schlining
Kakani Katjia
author_sort Eric C. Orenstein
collection DOAJ
description Marine scientists have been leveraging supervised machine learning algorithms to analyze image and video data for nearly two decades. There have been many advances, but the cost of generating expert human annotations to train new models remains extremely high. There is broad recognition both in computer and domain sciences that generating training data remains the major bottleneck when developing ML models for targeted tasks. Increasingly, computer scientists are not attempting to produce highly-optimized models from general annotation frameworks, instead focusing on adaptation strategies to tackle new data challenges. Taking inspiration from large language models, computer vision researchers are now thinking in terms of “foundation models” that can yield reasonable zero- and few-shot detection and segmentation performance with human prompting. Here we consider the utility of this approach for ocean imagery, leveraging Meta’s Segment Anything Model to enrich ocean image annotations based on existing labels. This workflow yields promising results, especially for modernizing existing data repositories. Moreover, it suggests that future human annotation efforts could use foundation models to speed progress toward a sufficient training set to address domain specific problems.
format Article
id doaj-art-12c7e49938d44a0492247d00a3526e9e
institution DOAJ
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-12c7e49938d44a0492247d00a3526e9e2025-08-20T02:48:12ZengFrontiers Media S.A.Frontiers in Marine Science2296-77452025-07-011210.3389/fmars.2025.14693961469396Assisting human annotation of marine images with foundation modelsEric C. Orenstein0Eric C. Orenstein1Benjamin Woodward2Lonny Lundsten3Kevin Barnard4Brian Schlining5Kakani Katjia6Research and Development, Information and Technology Dissemination, Monterey Bay Aquarium Research Institute, Moss Landing, CA, United StatesResearch and Development, National Oceanography Centre, Southampton, United KingdomResearch and Development, CVision AI, Medford, MA, United StatesResearch and Development, Information and Technology Dissemination, Monterey Bay Aquarium Research Institute, Moss Landing, CA, United StatesResearch and Development, Information and Technology Dissemination, Monterey Bay Aquarium Research Institute, Moss Landing, CA, United StatesResearch and Development, Information and Technology Dissemination, Monterey Bay Aquarium Research Institute, Moss Landing, CA, United StatesResearch and Development, Information and Technology Dissemination, Monterey Bay Aquarium Research Institute, Moss Landing, CA, United StatesMarine scientists have been leveraging supervised machine learning algorithms to analyze image and video data for nearly two decades. There have been many advances, but the cost of generating expert human annotations to train new models remains extremely high. There is broad recognition both in computer and domain sciences that generating training data remains the major bottleneck when developing ML models for targeted tasks. Increasingly, computer scientists are not attempting to produce highly-optimized models from general annotation frameworks, instead focusing on adaptation strategies to tackle new data challenges. Taking inspiration from large language models, computer vision researchers are now thinking in terms of “foundation models” that can yield reasonable zero- and few-shot detection and segmentation performance with human prompting. Here we consider the utility of this approach for ocean imagery, leveraging Meta’s Segment Anything Model to enrich ocean image annotations based on existing labels. This workflow yields promising results, especially for modernizing existing data repositories. Moreover, it suggests that future human annotation efforts could use foundation models to speed progress toward a sufficient training set to address domain specific problems.https://www.frontiersin.org/articles/10.3389/fmars.2025.1469396/fullfoundation modelmarine imagerysegmentationobject detectionhuman-in-the-loop
spellingShingle Eric C. Orenstein
Eric C. Orenstein
Benjamin Woodward
Lonny Lundsten
Kevin Barnard
Brian Schlining
Kakani Katjia
Assisting human annotation of marine images with foundation models
Frontiers in Marine Science
foundation model
marine imagery
segmentation
object detection
human-in-the-loop
title Assisting human annotation of marine images with foundation models
title_full Assisting human annotation of marine images with foundation models
title_fullStr Assisting human annotation of marine images with foundation models
title_full_unstemmed Assisting human annotation of marine images with foundation models
title_short Assisting human annotation of marine images with foundation models
title_sort assisting human annotation of marine images with foundation models
topic foundation model
marine imagery
segmentation
object detection
human-in-the-loop
url https://www.frontiersin.org/articles/10.3389/fmars.2025.1469396/full
work_keys_str_mv AT ericcorenstein assistinghumanannotationofmarineimageswithfoundationmodels
AT ericcorenstein assistinghumanannotationofmarineimageswithfoundationmodels
AT benjaminwoodward assistinghumanannotationofmarineimageswithfoundationmodels
AT lonnylundsten assistinghumanannotationofmarineimageswithfoundationmodels
AT kevinbarnard assistinghumanannotationofmarineimageswithfoundationmodels
AT brianschlining assistinghumanannotationofmarineimageswithfoundationmodels
AT kakanikatjia assistinghumanannotationofmarineimageswithfoundationmodels