Coordinate-Aware Mask R-CNN with Group Normalization: A Underwater Marine Animal Instance Segmentation Framework

Dewei Yi; Hasan Bayarov  Ahmedov; Shouyong Jiang; Yiren Li; Sean Joseph  Flinn; Paul G. Fernandes

doi:10.1016/j.neucom.2024.127488

Coordinate-Aware Mask R-CNN with Group Normalization: A Underwater Marine Animal Instance Segmentation Framework

Dewei Yi^* (Corresponding Author), Hasan Bayarov Ahmedov, Shouyong Jiang, Yiren Li, Sean Joseph Flinn, Paul G. Fernandes

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

Abstract

Unsustainable fishing, driven by bycatch and discards, harms marine ecosystems. Addressing this, we propose a Coordinate-Aware Mask R-CNN (CAM-RCNN) method to enhance fish detection in commercial trawls. Leveraging CoordConv and Group Normalization, our approach improves generalization and stability. To tackle class imbalance, a compound Dice and cross-entropy loss is employed, and image data are enhanced through multi-scale retinex and color restoration. Evaluating on two fishing datasets, CAM-RCNN excels in accuracy and generalization, achieving the best Average Precision (AP) for instance mask and BBOX prediction in both source (39.7%, 40.2%) and target domains (24.4%, 24.2%). This method promotes sustainable fishing by selectively capturing desired fish, reducing harm to non-target species.

Original language	English
Article number	127488
Journal	Neurocomputing
Early online date	6 Mar 2024
DOIs	https://doi.org/10.1016/j.neucom.2024.127488
Publication status	E-pub ahead of print - 6 Mar 2024

Bibliographical note

This work was supported in part by the Fisheries Innovation & Sustainability and U.K. Department for Environment, Food & Rural Affairs under Grant numbers: FIS039 and FIS045A.

Keywords

Instance segmentation
Convolutional neural network (CNN)
Underwater dataset
Generalisability

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1016/j.neucom.2024.127488Licence: CC BY

Cite this

@article{3888ca29c7f1491dbe5785422878814e,

title = "Coordinate-Aware Mask R-CNN with Group Normalization: A Underwater Marine Animal Instance Segmentation Framework",

abstract = "Unsustainable fishing, driven by bycatch and discards, harms marine ecosystems. Addressing this, we propose a Coordinate-Aware Mask R-CNN (CAM-RCNN) method to enhance fish detection in commercial trawls. Leveraging CoordConv and Group Normalization, our approach improves generalization and stability. To tackle class imbalance, a compound Dice and cross-entropy loss is employed, and image data are enhanced through multi-scale retinex and color restoration. Evaluating on two fishing datasets, CAM-RCNN excels in accuracy and generalization, achieving the best Average Precision (AP) for instance mask and BBOX prediction in both source (39.7%, 40.2%) and target domains (24.4%, 24.2%). This method promotes sustainable fishing by selectively capturing desired fish, reducing harm to non-target species.",

keywords = "Instance segmentation, Convolutional neural network (CNN), Underwater dataset, Generalisability",

author = "Dewei Yi and Ahmedov, {Hasan Bayarov} and Shouyong Jiang and Yiren Li and Flinn, {Sean Joseph} and Fernandes, {Paul G.}",

note = "This work was supported in part by the Fisheries Innovation & Sustainability and U.K. Department for Environment, Food & Rural Affairs under Grant numbers: FIS039 and FIS045A. ",

year = "2024",

month = mar,

day = "6",

doi = "10.1016/j.neucom.2024.127488",

language = "English",

journal = "Neurocomputing",

issn = "0925-2312",

publisher = "ELSEVIER SCIENCE BV",

}

TY - JOUR

T1 - Coordinate-Aware Mask R-CNN with Group Normalization

T2 - A Underwater Marine Animal Instance Segmentation Framework

AU - Yi, Dewei

AU - Ahmedov, Hasan Bayarov

AU - Jiang, Shouyong

AU - Li, Yiren

AU - Flinn, Sean Joseph

AU - Fernandes, Paul G.

N1 - This work was supported in part by the Fisheries Innovation & Sustainability and U.K. Department for Environment, Food & Rural Affairs under Grant numbers: FIS039 and FIS045A.

PY - 2024/3/6

Y1 - 2024/3/6

N2 - Unsustainable fishing, driven by bycatch and discards, harms marine ecosystems. Addressing this, we propose a Coordinate-Aware Mask R-CNN (CAM-RCNN) method to enhance fish detection in commercial trawls. Leveraging CoordConv and Group Normalization, our approach improves generalization and stability. To tackle class imbalance, a compound Dice and cross-entropy loss is employed, and image data are enhanced through multi-scale retinex and color restoration. Evaluating on two fishing datasets, CAM-RCNN excels in accuracy and generalization, achieving the best Average Precision (AP) for instance mask and BBOX prediction in both source (39.7%, 40.2%) and target domains (24.4%, 24.2%). This method promotes sustainable fishing by selectively capturing desired fish, reducing harm to non-target species.

AB - Unsustainable fishing, driven by bycatch and discards, harms marine ecosystems. Addressing this, we propose a Coordinate-Aware Mask R-CNN (CAM-RCNN) method to enhance fish detection in commercial trawls. Leveraging CoordConv and Group Normalization, our approach improves generalization and stability. To tackle class imbalance, a compound Dice and cross-entropy loss is employed, and image data are enhanced through multi-scale retinex and color restoration. Evaluating on two fishing datasets, CAM-RCNN excels in accuracy and generalization, achieving the best Average Precision (AP) for instance mask and BBOX prediction in both source (39.7%, 40.2%) and target domains (24.4%, 24.2%). This method promotes sustainable fishing by selectively capturing desired fish, reducing harm to non-target species.

KW - Instance segmentation

KW - Convolutional neural network (CNN)

KW - Underwater dataset

KW - Generalisability

U2 - 10.1016/j.neucom.2024.127488

DO - 10.1016/j.neucom.2024.127488

M3 - Article

SN - 0925-2312

JO - Neurocomputing

JF - Neurocomputing

M1 - 127488

ER -

Coordinate-Aware Mask R-CNN with Group Normalization: A Underwater Marine Animal Instance Segmentation Framework

Abstract

Bibliographical note

Keywords

UN SDGs

Access to Document

Fingerprint

Cite this