Researchers and wildlife officials have often struggled to identify bears. Now Silicon Valley software developers and biologists at the University of Victoria have joined forces to roll out an artificial intelligence (AI) solution.
BearID for Grizzly Bears
The AI, called BearID, can use features like the distance between the bear’s eyes, nose, ears and forehead to pick out an individual bear. The system was originally designed for grizzlies, the Vancouver Sun reports. But researchers hope to apply it to other bear species, and maybe even to animals like endangered caribou.
“The software takes all kinds of measurements of the face in multiple dimensions and learns which features are stable in individuals,” University of Victoria postdoctoral fellow Melanie Clapham told the Sun. “We need to be able to recognize them in the fall and in the spring and our initial analysis suggests that it can.”
The bears gain significant amounts of weight in the autumn. So it can be tough to follow the same bear through the seasons.
The technology has an 84% accuracy rate right now, the CBC reports. It can follow bears through their environment much in the same way the authorities track humans through airports.
Hopes for Broader Use
Clapham linked up with the software developers, Ed Miller and Mary Nyugen, in late 2017, according to the CBC. They met on Wildlabs.net. The website helps engineers and conservationists connect. And then they published a paper together about BearID in the journal Ecology and Evolution this month.
The scientists and software developers hope their invention will contribute to conservation.
“Learning about individual animals and their life stories can have really positive effects on public engagement and really help with conservation efforts,” Clapham told the Sun.
The group hopes local governments and nonprofits will pick up BearID. It has many practical uses, they say. For example, it could help people figure out whether a bear raiding a trash can was one culprit or many.
The technology should also help to reduce human-bear interactions. It diminishes the need to capture, tag and get samples from bears.