Combining computer vision and standardised protocols for improved measurement of live sea urchins for research and industry

Abstract

Abstract To allow sea urchin aquaculture to achieve its intended scale, efficient and precise methods for measuring large numbers of urchins in commercial‐scale operations are needed. Current protocols for measuring urchin test (shell) dimensions and mass are time‐consuming and prone to high measurement error, thus inconvenient in research and impractical in a commercial context. This study investigates and compares various measurement methods with a newly developed computer vision approach developed in this study, to establish a single protocol using precise, efficient and accessible methodology for measuring live urchins. We show that urchin wet mass can vary up to 8.73% depending on time out of water; this is significantly reduced to an average of 0.1% change by allowing urchins to drip‐dry for at least 90 s prior to weighing. We found the conventional vernier calliper method used to measure urchin dimensions to be both time‐consuming and imprecise (mean coefficient of variation (CV) of 2.41% for Tripneustes gratilla). Conversely, the computer vision programme we developed measures with higher precision (mean CV of 1.55% for T. gratilla) and is considerably faster. The software uses a series of hue saturation value filters, edge detection algorithms and distortions to measure the diameter of the test (excluding spines) of multiple urchins at once. The software is open‐source, and the protocol does not require specialised equipment (can be performed with a mobile phone camera). When the computer vision application is combined with the simple procedures described in this paper, to reduce measurement inaccuracies, urchin wet mass and diameter can be more efficiently and precisely determined. For a larger scale context, this software could easily be incorporated into various tools, such as a grading machine, to completely automate various farm processes. As such, this study has potential to assist urchin data collection in both research and commercial contexts.

Keywords

• computer vision
• machine vision
• methodology
• sea urchins
• wet weight