AI-Tracker to track animals of the road realtime to mitigate roadkill

Name of project leader : Liwan Liyanage

Project title: AI-Tracker to track animals of the road realtime to mitigate roadkill

Team members: Yasiru Kirindirachchi, Omar Mubin, Steen TfNSW, Ricky Spencer SoSc

Brief summary of your project (200 words maximum): Project Discription: This multidisciplinary research project aims to mitigate roadkill, protect endangered species, and improve road safety, thereby reducing accidents and contributing to the economy by lowering insurance claims and healthcare costs. The study introduces a reliable, efficient, and precise method for conducting ecological surveys of animals on and near roads across NSW, with potential national extension to reduce roadkill. Some Statistics: Roadkill, responsible for more than 5% of fatal crashes in Australia, poses the second biggest threat to Tasmanian devils' extinction and results in approximately 10 million animals being hit on Australian roads each year; furthermore, it incurs an annual cost of 19 million USD in Queensland alone. Approach Our approach leverages improved computer vision techniques to detect small objects in fast-moving videos, extending to real-time detection. As a case study, the research focuses on turtles, an endangered species prone to vehicle collisions in Australia, and will extend to other animals, including birds. Data is provided by TfNSW through GoPro cameras installed in vehicles during daily activities, making it cost-effective. A proof of concept has been completed, producing an AI-Tracker with over 90% accuracy and 80% confidence. We are now at the deployment stage, testing the AI-Tracker's effectiveness and comparing communication options for drivers to mitigate roadkill. Partner name/partner contribution (e.g. AusScope/$10000): TfNSW / Provide videos (in kind $20,000 already spend on the project on Gopro cameras and conducting the survey.)

CDMS contribution: $3190