Researcher(s)
- Caitlyn Davis, Wildlife Ecology and Conservation, University of Delaware
Faculty Mentor(s)
- Jeffrey Buler, Wildlife Ecology and Entomology, University of Delaware
Abstract
This summer I interned for the College of Agriculture and Natural Resources on a new Unidel foundation project called the Living Classroom. The goal of the Living Classroom is to implement modern environmental sensors and data processing tools to display the connections between soil, water, plant conditions, and wildlife on the UD farm in real-time.
My objectives were to 1) deploy all the wildlife sensors on the UD farm, including building weatherproof power supplies, 2) maintain these sensors over the course of the summer, and 3) help to develop Python script to automate data processing of camera images to detect and classify animals using the AI tools MegaDetector and SpeciesNet.
Within agricultural fields, wetlands, and forests on the farm, I deployed five Tactacam Reveal 3.0 Pro cameras that automatically capture and upload photos of wildlife, five Birdweather bioacoustics platforms, called PUCs, that automatically record bird, amphibian and insect calls along with environmental measures (e.g., temperature, light levels, airborne volatile organic compounds), and a network of 11 automated telemetry nodes that will track the movements of radio-tagged birds within the Roland Roth Ecology Woods. The PUCs Identify animal calls in audio clips with the BirdNET classifier list of over 6,000 species globally. The processed audio data is being visualized through a dashboard made by grad student Reshmitha Uppala that will eventually display species present on the farm on a public website.
Since the sensors have been deployed, camera traps have photographed 5 mammalian species and the birdweather PUCs recorded 37 various bird, insect, or amphibian species around the UD farm. It is with the hope that several classes across the agricultural college will utilize this living dataset for research questions and further learning.