A team of researchers is using facial recognition to identify brown bears in the wild. The BearID app is able to identify individual bears with 83.9 percent accuracy, and can at least detect the presence of a bear’s face 98 percent of the time.
The facial recognition algorithm was trained and tested with 4,674 images of 132 documented brown bears in Alaska and British Columbia. The researchers were attempting to use computer vision to help study animal species that do not have unique body markings. In the past, such methods have primarily been used to identify primates, which have more human features.
BearID uses landmarks like the snout, the ears, the eyes, and the shape of the forehead to identify individual bears. The researchers are hoping that the technology will help them study bear behavior and support conservation efforts, and suggested that the technology could potentially be modified to recognize other species.
This is not the first time that facial recognition has been applied to non-human species. Conservationists are using the technology to stop the illegal sale of chimpanzees, while the China Conservation and Research Center has developed an app that can identify giant pandas. Facial recognition has also been used to identify cats and dogs for insurance purposes.
The BearID study was published in Ecology and Evolution.
November 12, 2020 – by Eric Weiss