AI camera traps recognize the faces of mountain lions

same puma

Researchers have provide you with a method to take headshots of Mountain Lions within the wild after which categorize the person because of synthetic intelligence (AI) facial recognition.

The fascinating research makes use of digicam traps that make noises to curiosity the passing Concolore Puma, also referred to as cougars or mountain lions, in order that they flip and face the digicam permitting researchers to get a transparent image. The group can then analyze the pictures utilizing an internet AI app.

As one of the crucial secluded large cats on the planet, Mountain Lions are extraordinarily tough to look at firsthand. And in contrast to the distinctive stripes on a tiger, for instance, the dearth of distinctive options on a mountain lion’s physique makes it tough for researchers to establish and observe particular person animals.

Nevertheless, it is a completely different story in relation to mountain lion’s facial markings – which is what led researcher Peter Alexander and his group to develop non-invasive, facial-recognition-enhanced digicam traps to seize close-up images of the animals’ secretive faces.

same puma

In an interview with Scientific AmericanAnalysis biologist Peter Alexander, a resident of the Better Yellowstone Nationwide Park space, describes the difficulties of acquiring details about mountain lions with Conventional digicam traps utilized by researchers.

AI facial recognition digicam traps

Digital camera traps, that are concerning the measurement of a shoebox or perhaps a espresso cup, are connected to one thing alongside the animal’s typical path, reminiscent of a tree {that a} territorial puma scrapes. When movement is detected, the entice is triggered, leading to a shot of a mountain lion because it roams. The cameras even have an infrared flash in order that night time images are taken with out disturbing the animal.

Normal Digital camera Lure with Caller Alert.

Researchers all over the world use the sort of software to estimate inhabitants numbers and the general abundance of a species. They comb the pictures, typically utilizing machine studying algorithms, and analyze them to establish people.

Alexander says tigers are the “traditional instance” of utilizing digicam traps to establish particular person animals, “as a result of these stripes, they’re like fingerprints.”


However virtually all mountain lions all over the world, aside from a number of distinctive issues reminiscent of scars, have mild, sandy-colored fur down their sides. This lack of distinctive coloration on the perimeters of their our bodies signifies that researchers just like the Alexanders are normally unable to inform if a cougars cross the digicam entice 5 occasions, or if 5 animals have handed solo.

However then again, for those who may get a close-up of the Puma concolor face, it might be far more helpful. “There may be plenty of element within the filament patterns and every kind of issues,” Alexander explains.

Subsequently, Alexander and his group determined to benefit from the dramatic facial options of mountain lions. The researchers added a number of devices to their digicam traps in order that when movement was detected, a cougar cat’s name was triggered. This noise so reliably piqued the curiosity of passerby possums that they seemed lengthy sufficient for the digicam entice to take an image of the face.

AI tracking

5 impartial investigators reviewed puma head pictures and tried to establish particular person animals. In contrast with the standard side-angle digicam entice, the brand new attention-grabbing machine was 92% extra correct.

This research, revealed within the journal, Ecology and Evolution, earlier this yr is a vital step on the street to with the ability to establish and observe animals with higher confidence, in a scalable approach. Capturing headshots of mountain lions additionally opens up new alternatives for participating AI applied sciences. This software, which is analogous to the facial recognition know-how utilized by airport safety, may velocity up the method of analyzing photographs for researchers.


Alexander says this new digicam entice technique may very well be used to trace different animals that lack distinctive lateral colours however have distinctive options elsewhere. This contains susceptible species reminiscent of wolves, pine pricks, and even grizzly bears.

“It may very well be a extremely helpful know-how sooner or later. There have been plenty of different facial recognition research executed on animals, nevertheless it was by no means utilizing a digicam entice. That was sort of the distinctive factor about this research that was the merging of these two concepts,” Alexander says.

Picture credit: All images by Peter Alexander.