Ocean Ruler—an image-based, AI-driven approach to small-scale fisheries monitoring and catch size estimation
Many global fisheries lack sufficient monitoring, yet simple length‑based data can offer critical insights. This study evaluates Ocean Ruler, a web‑based computer‑vision tool that measures fish lengths from user‑submitted images. Comparing software‑derived measurements with hand measurements across commercial, recreational and artisanal fisheries in California and Baja California, the tool showed minimal bias in most groups, with slight overestimation for finfish. Although some measurement error remains, accuracy was generally acceptable. By integrating machine learning with citizen and community science, Ocean Ruler demonstrates strong potential for generating management‑relevant size‑structure data, especially in small‑scale, data‑limited fisheries.
Subject Tags
- Data Science and Artificial Intelligence
- Fisheries
Abstract
The ubiquity of data-limited, data-absent and unmanaged fisheries around the world illustrates a significant need for enhanced monitoring of living marine resources beyond conventional agency-led programs. While quantitative stock assessments represent the gold standard for fisheries management, simple length-based datasets alone can provide important insights into fishery status and can be collected by citizen and community scientists. Here, we demonstrate the performance of Ocean Ruler, a web-based tool that uses computer vision software and digital edge detection to measure individual lengths of harvested catch from images submitted by users (e.g., fishers, scientists, fisheries managers). Specifically, we compared software-derived measurements to conventional hand measurements to estimate rates of software bias and measurement error across four fishery groups in commercial, recreational, and artisanal fisheries that operate along the coast of California, USA and Baja California, Mexico. Through collaboration with local fishing communities, we demonstrate minimal software bias across three out of the four fishery groups, with minor, yet consistent overestimation observed while using the tool to estimate finfish lengths. We also note that efforts must be made to reduce software measurement error, despite achieving acceptable levels of accuracy on average in many cases. Nevertheless, we believe our efforts represent early successes in integrating machine learning tools with citizen and community science to generate management-relevant fishery size-structure data. Such approaches, when implemented effectively, have the potential to directly support management of living marine resources, particularly in small-scale, data-limited contexts.
Citation
Elstner, J., Bellquist, L., Hurd, F., Yocum, D., Schmuckal, C., Gleason, M., ... & Semmens, B. (2025). Ocean Ruler–an image-based, AI-driven approach to small-scale fisheries monitoring and catch size estimation. Marine Biology, 172(6), 90. https://doi.org/10.1007/s00227-025-04650-3
TNC Authors
-
Lyall Bellquist
The Nature Conservancy -
Frank Hurd
Oceans Project Director
The Nature Conservancy
Email: frank.hurd@tnc.org -
Mary Gleason
The Nature Conservancy -
Tom Dempsey
Director, California Oceans Program
The Nature Conservancy
Email: thomas.dempsey@tnc.org
TNC Authors
-
Kate Kauer
Assoc Director, Oceans Program, California
The Nature Conservancy
Email: kate.kauer@tnc.org -
Alexis Jackson
The Nature Conservancy