AI-powered images to help predict Australian bushfires

A new citizen science initiative will help predict bushfires in the summer ahead, according to the University of the Sunshine Coast (UniSC).

The National Bushfire Resilience Network (NOBURN) project empowers people to use their mobile phones for information collection to help predict bushfire hotspots and minimise their impact, said Chief Investigator Dr Sam Van Holsbeeck.

“Fire season is approaching. After some very wet years with everything growing nicely, there’s a lot of fuel available,” Van Holsbeeck said.

“The NOBURN app encourages people out and about in their local forests to take photos and tell us more about the forest and fuels. That data is processed by artificial intelligence to help predict the probability, severity and burn area of potential bushfires.

“So what we want people to do is to go into the forest, snap a pic and help predict.”

The citizen science project will utilise images captured by the public to help predict hotspots.

The project is the culmination of two years’ research through an alliance of world-renowned researchers in artificial intelligence, forestry, human factors and science communication at the University of the Sunshine Coast and University of Adelaide’s Australian Institute for Machine Learning (AIML), in partnership with Noosa Shire Council and funded through the Federal Department of Industry, Science and Resources.

Professor Javen Qinfeng Shi from AIML said the AI developed for NOBURN is cutting edge.

“We are developing AI models to spot potential bushfire hazards and assess bushfire fuel load from the images captured by the NOBURN app. The algorithms behind these AI models are based on AIML’s world-leading expertise in computer vision, and machine learning,” Shi said.

NOBURN was developed in the wake of the 2019–20 bushfires that burned more than 10 million hectares of forest, destroyed 2000 homes and claimed dozens of lives.

Professor Mark Brown from UniSC’s Forest Research Institute said the app will help predict future disasters of that scale.

“While naturally occurring bushfires cannot be avoided, there is an opportunity with this project to predict their likelihood and implement strategies to minimise their impact on the environment, property and life,” Brown said.

Van Holsbeeck believes NOBURN will generate not only better-informed science about the risk of bushfires in Australia, but also better-informed communities.

“The NOBURN project is a unique opportunity to engage the community to collect nationwide forest fuel data while creating more awareness on the risk associated with fuels in our forests. It's a great way to learn how to be better prepared for any potential disasters or extreme bushfire events if they were to happen,” Van Holsbeeck said.