Advanced drone-based systems are enabling the surveillance and location of threatened fauna with the highest-accuracy achievable up to now. QUT researchers have employed these devices to detect and locate koalas by conducting flight campaigns in New South Wales, and areas of neighbouring Queensland, including Logan, the Gold Coast, and the Australia Zoo Wildlife Hospital on the Sunshine Coast.

The research team has developed a unique combination of unmanned aerial vehicles (UAV) with thermal imaging, statistical modelling, and artificial intelligence (AI).

“We’ve found thermal imaging allows the detection of even well-camouflaged koalas effectively and our counting and tracking algorithms discriminate the shape of a koala from a possum, birds or other animals,” says Associate Professor Felipe Gonzalez, who is leading a UAV research team in collaboration with statisticians from the ARC Centre of Excellence for Mathematical & Statistical Frontiers (ACEMS) and biosecurity officers from the Australia Zoo Wildlife Hospital.

The project has already proved the technology can save councils valuable time. In a single test, it takes biosecurity experts over two hours to conduct roughly the same survey a UAV took just 30 minutes to complete.

The technology not only counts koalas and displays their location in georeferenced maps, but it can also monitor their movements and population fluctuations over time.

“This project is primarily focused on assessing koala populations, but this technology can easily be adjusted to monitor other species however, in conservation or pest-cataloguing such as feral cats, wild pigs and dingoes,” Felipe says.

Researchers and council delegates expect that having access to this information would contribute to supporting councils and ecologists into more efficient planning programmes for koala conservation.