Can automatic measuring replace humans when evaluating a shrimp fishery?

Abstract

Size measurements of fish and crustacean species play a critical role in fishery stock assessments, fishing gear size selectivity studies, and monitoring compliance with fisheries management regulations. One such example is from shrimp fisheries where samples of trawl-caught shrimps are frequently collected and size measured. However, the manual measurement of hundreds of small shrimps per sample is time-consuming and exhausting. Therefore, this study evaluates whether an automatic measuring procedure using off-the-shelf camera technology and a general-purpose artificial intelligence algorithm can replace manual measurements of deep-water shrimp (Pandalus borealis). Despite some deviations between manual and automatic measurements for individual shrimps, the automatic method proved sufficiently accurate for stock, gear selectivity and compliance assessment. Furthermore, this study demonstrated how a use-case driven approach can be applied when evaluating whether a new measuring technology can replace an existing.