Colleagues, I’d like to introduce you to finwave, a platform we have designed to help streamline the processes associated with dorsal fin photo-identification data management. Finwave does this first by combining all the steps associated with photo-ID workflow in one platform online. Each step from data submission, sorting, and cataloguing to individual identification and registration can be performed manually on finwave, however the system is set-up to automate these processes. For example, finwave incorporates machine learning (AI) algorithms for dorsal fin recognition initially developed by Bergler et al 2021 (see Sci. Rep. 11:23480) that can expedite the processes associated with individual identification. Model predictions are accurate to over 90% with sufficient training data and are continually being improved through incorporation of added layers of contextual knowledge. Any photo-ID data manager can create and administrate a population and associated community of data contributors on finwave, and there are several levels of usership and data access that can be set by administrators. For example, the (west coast transient) Bigg’s killer whale population has 2 administrators, 3 pros, and 8 experts who each have differing abilities to edit, export, and visualize data and view model predictions. There are also 148 other users who have joined this population, the majority of whom are active data contributors and can view all data but not edit any of it. Finwave is built and backed to house large quantities of images and their associated metadata. It currently has 12 years of Bigg’s killer whale photo-identification data back-loaded into the system including 153,680 images of 433 individuals from 2855 encounters. The built in algorithms have detected 207,593 dorsal fins in these images and made 175,083 identification predictions. In addition to being a digital data repository for the (west coast transient) Bigg’s killer whale population, photo-identification data managers of 4 other cetacean populations are currently onboarding to finwave. This system is continually under development but is now available for almost anyone to utilize. Populations can be made as openly accessible or private as administrators prefer and contributors can also set limits on how open their image data are by applying their preferred creative commons licenses to them. To request an invite to peruse finwave beyond the homepage at www.finwave.io, start or join a population, or if you have any questions please contact either or both of us at jrtow...@gmail.com and alexander.barnh...@fau.de Sincerely, Jared Towers Executive Director – Bay Cetology Alert Bay BC, Canada jrtow...@gmail.com Alex Barnhill PhD Student – Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen, Germany alexander.barnh...@fau.de |
_______________________________________________ MARMAM mailing list MARMAM@lists.uvic.ca https://lists.uvic.ca/mailman/listinfo/marmam
