IMIS

Publications | Institutes | Persons | Datasets | Projects | Maps
[ report an error in this record ]basket (0): add | show Print this page

Performance of iSharkFin in the identification of wet dorsal fins from priority shark species
Barone, M.; Mollen, F.H.; Giles, J.L.; Marshall, L.J.; Villate-Moreno, M.; Mazzoldi, C.; Pérez-Costas, E.; Heine, J.; Guisande, C. (2022). Performance of iSharkFin in the identification of wet dorsal fins from priority shark species. Ecological Informatics 68: 101514. https://dx.doi.org/10.1016/j.ecoinf.2021.101514
In: Ecological Informatics. Elsevier: Amsterdam. ISSN 1574-9541; e-ISSN 1878-0512, more
Peer reviewed article  

Available in  Authors 

Keyword
    Marine/Coastal
Author keywords
    Fisheries; Conservation; Monitoring; Compliance; CITES; Seafood; Species identification

Authors  Top 
  • Barone, M.
  • Mollen, F.H., more
  • Giles, J.L.
  • Marshall, L.J.
  • Villate-Moreno, M.
  • Mazzoldi, C.
  • Pérez-Costas, E.
  • Heine, J.
  • Guisande, C.

Abstract
    The past decade has seen a considerable rise in international concern regarding the conservation status of sharks and rays. The demand for highly prized shark commodities continues to fuel the international trade and gives fisheries incentive to use these resources, which have a low intrinsic capability to recover. Recognising the urgency for regulation, many countries voted to include more shark and ray species in the Appendices of the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES). However, the identification of fins in fisheries landings before they enter international trade is a major limitation for CITES compliance. This study reports the current performance of the iSharkFin system, a machine learning technology which aims to allow users to identify the species of a wet shark dorsal fin from its image. Photographs of 1147 wet dorsal fins from 39 shark species, collected in 12 countries, were used to train the algorithm over a four-year period. As new cohorts of images were used to test the performance of the learning algorithm, the accuracy of species assignments of known specimens was variable but did increase, reaching 85.3% and 59.1% at genus and species level respectively. The accuracy in predicting CITES-listed sharks versus unlisted sharks was 94.0% based on the 39 species currently represented in the baseline. Our results suggest that if supplied with high data inputs for specific fisheries assemblages and accompanied by user training, iSharkFin has promise for site-specific development as a rapid field identification tool in fisheries monitoring, and as a screening tool alongside traditional field morphology to detect potential CITES specimens for fisheries compliance and enforcement.

All data in the Integrated Marine Information System (IMIS) is subject to the VLIZ privacy policy Top | Authors