Researchers have devised a way to take headshots of mountain lions in the wild and then categorize the individual thanks to artificial intelligence (AI) facial recognition.
The remarkable study used camera traps that play a noise to pique the interest of passing puma concolor, otherwise known as cougars or mountain lions, so they turn and face the camera, allowing the researchers to get a clear picture. The team can then analyze the images using an AI online application.
As one of the most reclusive big cats in the world, mountain lions are extremely difficult to observe directly. And unlike a tiger’s unique stripes, for example, a mountain lion’s lack of distinctive body markings makes it difficult for researchers to identify and track individual animals.
However, it’s a different story when it comes to mountain lion facial markings – which is what led researcher Peter Alexander and his team to develop non-invasive camera traps enhanced with facial recognition to take close-up pictures of the secretive animals’ faces .
In an interview with Scientific Americanresearch biologist Peter Alexander, who is based in the greater Yellowstone National Park area, describes the difficulty of obtaining information on the mountain lion population with the traditional camera traps used by researchers.
AI Face Recognition Camera
Camera traps, which are about the size of a shoebox or even a coffee cup, are attached to something that is in the animal’s regular path, such as a tree that the Puma has territorially scraped. When movement is detected, the trap is triggered, resulting in a snapshot of the mountain lion as it walks. The cameras even have an infrared flash so that night shots can be taken without disturbing the animal.
Researchers around the world use this type of tool to estimate population numbers and overall species abundance. They go through the images, sometimes using machine learning algorithms, and analyze them to identify individuals.
Tigers are the “classic example” of using camera traps to identify individual animals, Alexander says, “because those stripes are like a fingerprint.”
But almost all mountain lions around the world, barring distinguishing things like markings, have light, sandy-colored fur on their sides. This lack of unique coloration on the sides of their bodies means researchers like Alexander usually can’t tell if a cougar crosses a camera trap five times, or if five separate animals pass it.
But on the other hand, if you can get a close-up image of the Puma concolor’s face, that’s much more helpful. “There’s a lot of detail in the mustache patterns and all sorts of other things. They are beautiful”, explains Alexander.
So, Alexander and his team decided to take advantage of the dramatic facial features of mountain lions. The researchers added several tweaks to their camera traps so that when motion is detected, a cougar kitten call is played. This noise certainly piqued the interest of the passing cougars, so they looked up long enough for the camera to snap a photo of the face.
Five independent investigators reviewed the photos of the cougar’s head and attempted to identify the individual animals. Compared to the traditional side-angle camera trap, the new eye-catching device was about 92% more accurate.
This study, published in the journal Ecology and evolution, earlier this year is an important step toward being able to more confidently identify and track animals in a scalable way. Taking headshots of mountain lions also opens up new possibilities for using AI techniques. This tool, which is similar to facial recognition technology used by airport security, can speed up the image analysis process for researchers.
Alexander says this new camera-trapping method could be used to track other animals that don’t have distinctive side colors but have unique features elsewhere. This includes vulnerable species such as wolverines, pine martens and even grizzly bears.
“This could be a really useful technique in the future. There have been many other face recognition studies done on animals, but never with a camera trap. So what was unique about this study was bringing these two ideas together,” Alexander says.
Image Credits: All photos by Peter Alexander.