WaterSAM: Adapting SAM for Underwater Object Segmentation

Object segmentation, a key type of image segmentation, focuses on detecting and delineating individual objects within an image, essential for applications like robotic vision and augmented reality. Despite advancements in deep learning improving object segmentation, underwater object segmentation re...

Full description

Saved in:
Bibliographic Details
Main Authors: Yang Hong, Xiaowei Zhou, Ruzhuang Hua, Qingxuan Lv, Junyu Dong
Format: Article
Language:English
Published: MDPI AG 2024-09-01
Series:Journal of Marine Science and Engineering
Subjects:
Online Access:https://www.mdpi.com/2077-1312/12/9/1616
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850260771397697536
author Yang Hong
Xiaowei Zhou
Ruzhuang Hua
Qingxuan Lv
Junyu Dong
author_facet Yang Hong
Xiaowei Zhou
Ruzhuang Hua
Qingxuan Lv
Junyu Dong
author_sort Yang Hong
collection DOAJ
description Object segmentation, a key type of image segmentation, focuses on detecting and delineating individual objects within an image, essential for applications like robotic vision and augmented reality. Despite advancements in deep learning improving object segmentation, underwater object segmentation remains challenging due to unique underwater complexities such as turbulence diffusion, light absorption, noise, low contrast, uneven illumination, and intricate backgrounds. The scarcity of underwater datasets further complicates these challenges. The Segment Anything Model (SAM) has shown potential in addressing these issues, but its adaptation for underwater environments, AquaSAM, requires fine-tuning all parameters, demanding more labeled data and high computational costs. In this paper, we propose WaterSAM, an adapted model for underwater object segmentation. Inspired by Low-Rank Adaptation (LoRA), WaterSAM incorporates trainable rank decomposition matrices into the Transformer’s layers, specifically enhancing the image encoder. This approach significantly reduces the number of trainable parameters to 6.7% of SAM’s parameters, lowering computational costs. We validated WaterSAM on three underwater image datasets: COD10K, SUIM, and UIIS. Results demonstrate that WaterSAM significantly outperforms pre-trained SAM in underwater segmentation tasks, contributing to advancements in marine biology, underwater archaeology, and environmental monitoring.
format Article
id doaj-art-2e49cab6eacd4c0181bf36801671ea05
institution OA Journals
issn 2077-1312
language English
publishDate 2024-09-01
publisher MDPI AG
record_format Article
series Journal of Marine Science and Engineering
spelling doaj-art-2e49cab6eacd4c0181bf36801671ea052025-08-20T01:55:34ZengMDPI AGJournal of Marine Science and Engineering2077-13122024-09-01129161610.3390/jmse12091616WaterSAM: Adapting SAM for Underwater Object SegmentationYang Hong0Xiaowei Zhou1Ruzhuang Hua2Qingxuan Lv3Junyu Dong4School of Computer Science and Technology, West Coast Campus, Ocean University of China, No. 1299 Sansha Road, Binhai Street, Huangdao District, Qingdao 266100, ChinaSchool of Computer Science and Technology, West Coast Campus, Ocean University of China, No. 1299 Sansha Road, Binhai Street, Huangdao District, Qingdao 266100, ChinaSchool of Computer Science and Technology, West Coast Campus, Ocean University of China, No. 1299 Sansha Road, Binhai Street, Huangdao District, Qingdao 266100, ChinaSchool of Computer Science and Technology, West Coast Campus, Ocean University of China, No. 1299 Sansha Road, Binhai Street, Huangdao District, Qingdao 266100, ChinaSchool of Computer Science and Technology, West Coast Campus, Ocean University of China, No. 1299 Sansha Road, Binhai Street, Huangdao District, Qingdao 266100, ChinaObject segmentation, a key type of image segmentation, focuses on detecting and delineating individual objects within an image, essential for applications like robotic vision and augmented reality. Despite advancements in deep learning improving object segmentation, underwater object segmentation remains challenging due to unique underwater complexities such as turbulence diffusion, light absorption, noise, low contrast, uneven illumination, and intricate backgrounds. The scarcity of underwater datasets further complicates these challenges. The Segment Anything Model (SAM) has shown potential in addressing these issues, but its adaptation for underwater environments, AquaSAM, requires fine-tuning all parameters, demanding more labeled data and high computational costs. In this paper, we propose WaterSAM, an adapted model for underwater object segmentation. Inspired by Low-Rank Adaptation (LoRA), WaterSAM incorporates trainable rank decomposition matrices into the Transformer’s layers, specifically enhancing the image encoder. This approach significantly reduces the number of trainable parameters to 6.7% of SAM’s parameters, lowering computational costs. We validated WaterSAM on three underwater image datasets: COD10K, SUIM, and UIIS. Results demonstrate that WaterSAM significantly outperforms pre-trained SAM in underwater segmentation tasks, contributing to advancements in marine biology, underwater archaeology, and environmental monitoring.https://www.mdpi.com/2077-1312/12/9/1616underwater object segmentationunderwater imageSegment Anything Model (SAM)
spellingShingle Yang Hong
Xiaowei Zhou
Ruzhuang Hua
Qingxuan Lv
Junyu Dong
WaterSAM: Adapting SAM for Underwater Object Segmentation
Journal of Marine Science and Engineering
underwater object segmentation
underwater image
Segment Anything Model (SAM)
title WaterSAM: Adapting SAM for Underwater Object Segmentation
title_full WaterSAM: Adapting SAM for Underwater Object Segmentation
title_fullStr WaterSAM: Adapting SAM for Underwater Object Segmentation
title_full_unstemmed WaterSAM: Adapting SAM for Underwater Object Segmentation
title_short WaterSAM: Adapting SAM for Underwater Object Segmentation
title_sort watersam adapting sam for underwater object segmentation
topic underwater object segmentation
underwater image
Segment Anything Model (SAM)
url https://www.mdpi.com/2077-1312/12/9/1616
work_keys_str_mv AT yanghong watersamadaptingsamforunderwaterobjectsegmentation
AT xiaoweizhou watersamadaptingsamforunderwaterobjectsegmentation
AT ruzhuanghua watersamadaptingsamforunderwaterobjectsegmentation
AT qingxuanlv watersamadaptingsamforunderwaterobjectsegmentation
AT junyudong watersamadaptingsamforunderwaterobjectsegmentation