Post-bushfire AI surveillance program monitoring wildlife one year on from the Black Summer

19 December 2020

AUSTRALIA – More than 600 cameras will be installed across the country to track the elusive movements of Australia’s bushfire-affected wildlife.

It is estimated almost 3 billion native animals were in the path of the devastating bushfires that ravaged parts of Australia.

Led by World Wide Fund For Nature Australia (WWF) and Conservation International, with a $US1 million grant from Google, An Eye on Recovery is a large-scale collaborative camera project.

The cameras will survey areas in the Blue Mountains, East Gippsland, Kangaroo Island and South-East Queensland to find where further recovery actions are needed.

‘Still traumatised’

Kangaroo Island Odysseys tour guide, Nikki Redman, was at her home in Kingscote when the island was on fire.

Ms Redman played an instrumental role when it became safe enough to start the recovery process.

“I was helping a friend manage the Hanson Bay Wildlife Sanctuary, so I was out volunteering helping find injured animals and help the surviving ones.”

Ms Redman said the project sounded like a “good” idea.

“There are animals coming back into these areas, but it would be great to see what’s actually happening in there,” she said.

But, Ms Redman said it was a “shame” it took a fire for there to be a stronger focus on the island’s wildlife.

“There has always been trouble getting funding for projects and it is a real shame for the researchers to try and get the funding to do this — it’s taken a devastating fire,” she said.

A technological venture

WWF Australia’s Head of Healthy Land and Seascapes, Darren Grover, said it would provide a good indication of how the bushfire-affected areas were recovering.

“By having more than 600 sets of eyes out there across Eastern Australia including Kangaroo Island, we’ll start to get an indication of how that wildlife is faring or how that wildlife has suffered,” he said.

Mr Grover said the use of artificial intelligence could “revolutionise” the way wildlife are monitored in Australia.

“If anyone has used camera traps or wildlife cameras, you know you end up with thousands and thousands of images and the thing you might be looking for might only be on one or two of those images,” he said.

“What this artificial intelligence can deliver is the ability to very quickly go through your images and catalogue the images accordingly.”

The pictures captured by the sensor cameras will be uploaded to an online database, then analysed by Wildlife Insights, which uses artificial intelligence.

This is the first time the technology will be used in Australia, so WWF Australia is asking the public to upload their sensor camera images of wildlife, especially in bushfire regions, to speed up the machine-learning process.

A difficult task

So far, 100 rechargeable battery-operated cameras have been set up on Kangaroo Island and the project is expected to last at least two years.

“We are installing what we call drift lines, which are little fences maybe not even 30 centimetres high, made of fly wire like you have on your windows,” he said.

“The drift lines are about 30 metres long and at each end we have a camera — the idea being, any small wildlife that’s wandering through will go past one of our cameras.”

The sensor cameras have already taken pictures of the elusive Kangaroo Island dunnart.

“They are a critically endangered, carnivorous marsupial about the size of a mouse and we had really grave concerns for them because 90–95 per cent of its habitat was burnt,” Mr Grover said.

Different approaches

Australian Wildlife Conservancy’s (AWC) Chief Science Officer, John Kanowski, said the organisation also had cameras on Kangaroo Island, but instead of relying on the public to make submissions to teach the AI technology to identify animals, the AWC was using researchers.

“Our approach has the benefit of having 60 full-time ecologists and 20 of them have a PhD,” he said.

“We’re starting with a very high-quality training set and we’re finding that’s a pretty critical part of the problem-solving.”

However, he said there were some shortfalls to using AI technology.

“We employ a coder as well and we’re working with Microsoft in the United States and we’ve gone through between 4 and 5 million images already.

“There are a number of people trying to solve this problem.”

19. December 2020/by GFMCadmin