AUTOMATIC DETECTION OF VARIOUS STINGRAY SPECIES IN TAMPA, FLORIDA

Studies on stingrays are lacking in elasmobranch research, leading to a higher risk of extinction of the species. Using remote sensing techniques will lead to thorough stingray research. Unoccupied aerial vehicles (UAVs), or drones, are a non-invasive method to gather in-situ data on stingrays including behavior, which can define the stingray’s ecological niche. Incorporating deep learning (DL) into ecological research is beneficial because it automatically detects the subject of the research being done. This study aims to use UAVs to collect footage of stingrays in their habitat and train a DL program to correctly identify three stingray species within Tampa Bay, Florida; spotted eagle rays, southern stingrays, and Atlantic stingrays. Drone flights will be conducted between November 2024 and January 2025. UAV footage is analyzed and notated using the Labelbox program. GPS coordinates recorded during each flight are used to estimate an individual’s location. The annotated footage will train the DL program in TensorFlow and will provide valuable insight into the usability and effectiveness of DL in identifying aquatic species. Ultimately, the methods developed throughout this study will help monitor the populations of other at-risk elasmobranch species and help improve species identification, population monitoring, and management efforts.

Presentation Preference: Either

Primary Presenter: Josephine Hoffman, University of Tampa (josiehoffman12@gmail.com)

Authors:

Caitlyn Walley, University of Tampa  (caitlyn.walley@spartans.ut.edu)

Edna Fernandez-Figueroa, University of Tampa  (efernandezfigueroa@ut.edu)