The words “facial recognition” may inspire visions of a dystopian future. But this technology isn’t just for humans—it can also identify animals. And a non-profit called BearID believes that facial recognition may revolutionize wildlife conservation.
BearID is operated by a small team. The software is developed by Mary Nguyen and Ed Miller (who also happens to be principal engineer at ARM), while the conservation expertise comes from Melanie Clapham, a PhD in conservation biology.
In some ways, this is a very straightforward project. The BearID team wants to develop a real-time facial recognition network for bears. Such a system will save a ton of labor—scientists won’t need to go out and count the number of bears in a given area (or tag them, which is obviously a bit dangerous). But it will also provide instant information on a bear’s whereabouts, its path, and its habits.
From a conversation standpoint, this data is pretty invaluable. It also has the potential to be a huge money-saver, as scientists can collect a wealth of data without sending people to manually observe bears. Of course, BearID believes that facial recognition technology could track other animals.
The technology behind BearID is still very hands-on. Basically, images are shot in the field, uploaded to a cloud server, and processed using AI. Ed Miller hopes to improve and automate this system through a small ARM computer, such as a Raspberry Pi 4B, that can shoot and process images of bears locally. A low-power transmitter in the computer can then send finished data back to a lab. (In any case, the computer will need to have onboard AI capabilities, which may prove to be a challenge. Maybe they should use Google Pixel phones?)
For more information on BearID, I suggest visiting the non-profit’s website. You can also check out The Raspberry Pi Foundation’s writeup on BearID, which contains some unique images, videos, and information.
Source: BearID via The Raspberry Pi Foundation
