Facial recognition may have a mixed reputation at the moment, but despite the kind of controversial implementation which makes privacy campaigners wince, it can undoubtedly be used for good. A case in point for this: researchers from Michigan State University have adapted facial recognition technology to identify primates with up to 93.76% accuracy.
Tracking animals is essential for conservationists, as it gives them real insights into how effective their efforts are proving. If you can track the animals you’re trying to protect, in other words, you can ensure your methods are best practice. But there’s more to it than that as well – if animals can be recognised by artificial intelligence, than those seized from traffickers can be traced back to their place of origin, helping to improve defences in future.
There are, of course, already ways of tracking animals, but something as non-invasive as facial recognition would be a huge improvement for a number of reasons. Not only do traditional tracking devices cost between $400 and $4,000, but the actual tagging process is both time consuming for conservationists and uncomfortable for the animals involved.
READ NEXT: The clever tech keeping South African rhinos alive
For this reason, the researchers trained their own software – PrimNet – with thousands of reference photographs of primates (3,000 lemur pics, 1,450 golden monkey snaps and 5,559 chimp photos). It’s not as simple as just putting a bunch of photographs in the system though, as the researchers outline in the paper: “Face detection also comes with some additional challenges due to the presence of variations in hair and fur, low contrast between eyes and background, and variation in eye colors across different individuals,” the paper explains.
In the end, they had to manually tag in the eyes and mouths, but this approach proved pretty successful, improving on standard off the shelf human facial recognition software. For lemurs, PrimNet achieved an accuracy of up to 93.76% and golden monkeys weren’t far behind on 90.36%. Chimpanzees, however, were some way behind with a maximum of 75.82%.
That’s still not perfect, but in the accompanying Android app – PrimID – aims to mitigate that problem a little. If you snap a photo of a lemur, and it isn’t 100% certain of a match, it will provide five other likely candidates.
“Moving forward, we plan to enlarge our primate datasets, develop a primate face detector, and share our efforts through open-source websites,” said Anil Jain, senior author on the study. If taken up by anti-trafficking agents, this could prove another useful technological weapon in conservationists’ fight against poachers.
Disclaimer: Some pages on this site may include an affiliate link. This does not effect our editorial in any way.
