Creative Commons source reading · Machine learning for performance
Testing the feasibility of deep learning approaches to enhance monitoring of marine macroinvertebrates: Insights from a case study using the gastropod Peringia ulvae
Jurado IC; Earp HS; Vanschoren J; Sugden H
Originally published in Environ Monit Assess · 2026-07-14
From the article
Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.
Abstract
AbstractCoastal ecosystems, while vital for their ecosystem services, face growing threats from climate change and human activities. Traditional methods for monitoring these systems, such as morphological species identification and laboratory analysis, are time-consuming and resource-intensive, restricting their scalability. Recent advances in Deep Learning have demonstrated significant potential for automating organism identification and counting from images, streamlining ecological research efforts. Using a common marine gastropod, Peringia ulvae, as a model species, we evaluate the potential for state-of-the-art computer vision methods incorporating self-supervised and few-shot learning techniques, to quantify this marine macroinvertebrate. We tested five machine learning models (CounTR, Segment Anything, Training-Free Object…
Discussion
Deep learning approaches are becoming increasingly common place in scientific research, particularly to identify and enumerate organisms within images (Farjon et al., 2023; Singh et al., 2023). Here we advanced the use of computer vision models, using them to classify and count microscopic a common marine macroinvertebrate, P. ulvae, in images. We found that sample orientation, organism density and level of aggregation, and the choice of computer vision model all influenced digital count accuracy, with values obtained by computer models generally lower than those obtained by human counts — highlighting the importance of benchmarking procedures. We also assessed the strengths and limitations of computer vision techniques compared to traditional laboratory procedures for analyzing macroinvertebrate samples. Collectively, our results…
Conclusion
This study demonstrates the potential of machine learning approaches to enhance the efficiency and scalability of marine macroinvertebrate monitoring, highlighting that some models can be trained to correctly identify and count a common gastropod species across a range of imaging conditions. By benchmarking five computer vision models against traditional laboratory techniques, we found that CounTR and Deep Data Space achieved a comparable accuracy to human counts, particularly when organisms are aggregated together. However, performance limitations emerged at greater organism densities, with most models plateauing around 400 individuals per image, underscoring the need for further refinement in handling high-density scenarios, which are common in many marine monitoring programmes. While machine learning techniques are not a substitute…
Editorial reading guide
- What data created and tested the model, and was evaluation independent?
- What is the cost of a false positive, false negative or out-of-distribution decision?
