AI tells beavers apart by the ‘fingerprint’ patterns on their tails

Beavers rely on their leathery tails to steer while swimming and to loudly smack the water as an alarm call. A covering of lizardlike scales makes these tails so handy. It also provides a way to tell the animals apart. According to a study published this week in Ecology and Evolution , a computer algorithm can accurately identify individual beavers by the pattern of scales on their tails , a bit like human fingerprints. The advance could be good news for the Eurasian beaver ( Castor fiber ), which was nearly hunted to extinction in the 19th century. As the species recovers, researchers have been estimating the size of populations using ear tags and radio collars. But capturing beavers can stress them, so scientists trained a pattern-learning type of artificial intelligence on images of tails from 100 Eurasian beavers that had been hunted in Norway. The program was 96% accurate at telling the animals apart. The scientists analyzed photographs taken in a laboratory under good lighting, but the approach should be feasible with images from the wild, they say. In another study, published in the European Journal of Wildlife Research last year, the same team showed that adding a small plastic lens to automatic cameras in the field provided a fourfold boost in high-quality photos —good enough to help computers identify some of nature’s most famously industrious animals.
Source: ScienceNOW - Category: Science Source Type: news