Automated matching algorithms are an increasingly valuable tool in marine mammal research. A multispecies matching algorithm, originally described in Patton et al. analyzes distinctive morphological characteristics on cetaceans using scarring, pigmentation patterns, and notches along the trailing edge of the dorsal fin, which can pose challenges in traditional eye-based matching due to individual variation over time. While the algorithm was trained on 24 different cetacean species, Risso’s dolphins were not included. Here, we assess its effectiveness using Cascadia Research’s Risso’s dolphin photo-identification catalog, which consists of 1165 unique individuals from Southern California documented across 701 sightings from 2010 to 2021. The multispecies algorithm identified 90 missed matches within the catalog and was subsequently used to update the Cascadia Research catalog more efficiently with sightings collected through 2024. Our results indicate that the algorithm outperformed experienced human matchers, particularly for individuals with less distinctive features, streamlining the identification process and supporting the long-term monitoring of Risso’s dolphins while showing strong potential for application to other species with similar identification challenges.
Citation:
Vanderzee, A., T. Cheeseman, P.T. Patton, and J. Calambokidis. 2026. Evaluating the Effectiveness of a Multispecies Matching Algorithm Applied to a Novel Species, Risso’s Dolphins (Grampus griseus). Marine Mammal Science 42(2): e70159. doi: 10.1111/mms.70159
Link:
https://onlinelibrary.wiley.com/doi/10.1111/mms.70159