Computer programmers in California’s Silicon Valley are helping conservationists use facial recognition technology to study the majestic North American brown bear.
Brown bears lack distinctive spots or stripes and undergo extreme weight changes that make it difficult for researchers to identify specific bears.
But software developers and scientists have created a new computer program called BearID that uses artificial intelligence to allow scientists to monitor and study wild bears like never before, according to news reports.
BearID uses a computer algorithm to quickly recognize bears by comparing facial features in video images with a database of photos of known bears, according to a November 6 paper in the journal Ecology and Evolution.
The technology will help conservationists, who previously had to rely on personal knowledge of the habits of specific bears, to better track relationships and understand behavior.
Wildlife conservation is one of many fields to gain from American innovators’ advances in artificial intelligence. Computers programmed to think or respond like human beings are already used to diagnose diseases, drive cars and predict natural disasters.
The U.S. government in August invested $1 billion for research into AI and other advanced technologies, such as quantum computers, that U.S. companies are advancing to solve the world’s most pressing problems.
Additionally, the U.S. government spends more than $100 million on combating wildlife trafficking around the world each year. The funding helps foster innovative solutions for monitoring wildlife and combating wildlife trafficking, including through programs such as the State Department’s annual ZooHackathon, a global competition that fosters innovative technological solutions to wildlife trafficking.
The idea for BearID dates back to 2017, when software developers Ed Miller and Mary Nguyen were taking an online class in machine learning and spending free time watching livestreams of bears in Katmai National Park in Alaska, according to the New York Times. When Miller found himself wondering which bears he was watching at a given time, the seed for BearID was planted.
Miller and Nguyen teamed up with Melanie Clapham, a researcher at the University of Victoria, Canada, and Chris Darimont of the Raincoast Conservation Foundation, who were considering how AI could advance their conservation efforts.
Nguyen told the Times she and Miller reviewed over 4,000 photos of bears, identifying each bear by its eyes, ears and nose. The dataset taught the computer program how to identify landmarks on a bear’s face and use them to quickly identify known bears with 84 percent accuracy.
BearID’s creators have published the computer code so others may improve or adapt the program to recognize other hard-to-identify animals around the world.
“What we’d love is that one day we have somewhere where people can upload camera trap images and the system tells you not only what species you’ve seen, but also what individual you’ve seen,” Clapham told the Times.