Automated Detection and Counting of Flying-Foxes from Thermal Imagery

Fruit bats, particularly the endangered spectacled flying-fox (Pteropus conspicillatus), play a crucial role in maintaining healthy ecosystems in Australia’s Wet Tropics by pollinating plants and dispersing seeds. Reliable population monitoring is essential for conservation, but traditional counting methods are time-consuming, labour-intensive, and difficult to scale across large or densely vegetated roost sites.

This project explores how artificial intelligence (AI) and computer vision can improve the way flying-fox populations are monitored. Using thermal infrared cameras mounted on drones, bats can be detected based on their body heat, enabling surveys even in low-light conditions without disturbing the animals. However, analysing thermal imagery is challenging because the images are often noisy, low resolution, and bats appear as very small objects.

The project is developing an AI-powered system that can automatically detect and count bats from drone-captured thermal images. This includes building annotated datasets, designing deep learning models tailored to thermal imagery, and evaluating performance to ensure reliable counting results.

By enabling faster, more scalable, and consistent monitoring, this work demonstrates how AI and data science can support wildlife conservation and provide valuable insights for ecological research and management in the Wet Tropics region.

Project Team and Collaborators:

Emmeline Norris, Xi Zhou, Susan Laurance, Tao (Kevin) Huang