WildObs: Australian scientists are trying to create a huge database of wildlife selfies

Around the country, there are thousands of cameras hidden in the wild, just waiting for an unsuspecting potoroo, dingo, or feral cat to snap a selfie.

These cameras, and the massive number of images and data they capture, are currently mostly isolated – potentially used for a research project or two, and then deleted to make room on the hard drive for something else.

However, a team of Australian researchers from the University of Queensland, Sydney and Melbourne, as well as James Cook University and others, are trying to create something much more far reaching. The team is in the early stages of building a central database, which they’ve named the Wildlife Observatory of Australia (WildObs), to collect, store and tag this huge quantity of data.

“We’re creating a national standard for wildlife monitoring so researchers can, for the first time, systematically monitor wildlife across the Australian continent,” says UQ ecologist, Dr Matthew Luskin.

The idea for WildObs is to have a national ‘one stop shop’ for researchers. Eventually, the team hopes it will be treasure trove of information for researchers looking for certain animals or location data even without entering the field.

But aware that they’re unlikely to convince researchers to give up their hard earned data without reward, the team is creating helpful tools for ecologists in the field, to make it attractive to use the system.  

“We want to provide tools for practitioners that set camera traps on the field, provide tools to rapidly sort out images using artificial intelligence and then once those images are all sorted … we aim to provide some nifty tools for users to then analyse that camera trap data,” UQ ecologist Zachary Amir told Cosmos.

The AI system which WildObs is using is called Wildlife Insights, and has already sorted over 54 million camera trap records from around the world.

The researchers stress that this is still very much early days. Some of the team – like Amir – have been creating new datasets for the system, having recently finished taking 130,000 photos on 25 cameras on K’gari (Fraser Island) monitoring potoroos. Now, he’s just starting a new camera survey across Queensland’s tropical rainforests, hoping to get camera data across nine national parks.

For him, he’s happy for the data to be out there for other researchers to analyse as soon as possible, although he explains that there the system will have embargoes in place if the researchers want to get a first shot at their data before it goes to the wider research community.

“My interest is looking at species interactions, and I want to understand how the abundance of feral cats in one park might affect the abundance of musky rat kangaroos in a different park,” he says.

“The more people that have more data, the better decisions we can all make. It’s just it’s better for all parties.”

The WildObs team are hoping to be as transparent and open as possible with their data, although there are still some kinks to work out. For example, sensitive species might need their coordinates blurred, while hunters could use their data for more nefarious reasons.

“There will still be checks in place to make sure it’s not pig hunter Joe, who uses the data to find out where feral pigs are,” says Amir.

“But if you’re a researcher and you’re going to use the data for science, or if you’re a land manager trying to make better decisions – there should be no boundaries … We are aiming for pretty open transparency and availability of these data sources.”