Work

Automated detection of sea cucumbers in turbid subtidal marine habitats: An explainable approach

Year: 2025

Type: article

Abstract: Deposit-feeding sea cucumbers perform important ecosystem functions, enhancing nutrient cycling and maintaining ecosystem health. Understanding this function over large spatial scales requires population surveys which can be logistically constrained, especially in turbid subtidal environments where sea cucumbers exhibit patchy distributions. While deep learning has shown promise for sea cucumber detection, factors like object occlusions, limited ambient light, and turbid environmental conditions hamper the model generalizability. Yet, detection models that can be generalised under a broad range of environmental conditions are key to conducting large-scale in-situ population surveys of sea cucumbers, particularly in highly developed coastal regions. Here, we demonstrate the pipeline using YOLOv8 for large-scale population surveys of subtidal organisms. Specifically, we use sea cucumbers as a model species to showcase a series of photometric data augmentation processes tailored for small datasets, collected under various turbid environmental conditions with a consumer grade waterproof camera (GoPro Hero9 Black). We further interpret the model detection sensitivity with Eigenvalue-based class activation map (Eigen-CAM) to visualise the model perceptual fields on in-situ image samples. Our sea cucumber detection model achieved 96.41% average precision, with the augmentation processes effectively enabling generalisation under turbid environmental conditions despite limited training data. Visualisations from Eigen-CAM empowered model interpretability and transparency on how the spatial arrangement of features affect the decision-making processes during model prediction. Overall, this study demonstrates a practical solution for efficient and accurate surveys of subtidal species distribution, especially under low-visibility environmental conditions.

Source: Ecological Informatics

DOI: 10.1016/j.ecoinf.2025.103342

Authors: Cheryl Chu, Yi-Fei Gu, Adrian Wong, Bayden D. Russell

Institution: University of Hong Kong

Language: English

Topic: Advanced Neural Network Applications

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

SDG: Life below water

Open Access status: gold

Estimated APC paid: $2,770