Protecting snow leopards from extinction starts with knowing how many exist – not an easy task given their nature. The magnificent, fluffy-tailed cats live in frigid, barren highlands; can roam hundreds of miles; and are so elusive that they’re nicknamed “ghosts of the mountain.” For the Snow Leopard Trust, a global conservation nonprofit based in Seattle, camera traps are one of the most efficient ways to study the threatened species.

Set up in the cats’ Central and South Asian habitats, the nonprofit’s cameras generate hundreds of thousands of photos. Many of the images are not of snow leopards, but of goats, camels, horses and even swaying grass – anything that triggers the cameras’ heat and motion sensors. Sorting the images into photos with and without snow leopards was traditionally a long, tedious, manual task.

But the Snow Leopard Trust now uses Microsoft AI to analyze and sort the images into a database of sightings, doing in 10 minutes what used to take 10 days.

“Analyzing all those images used to take up a considerable chunk of our time,” says Koustubh Sharma, a senior regional ecologist at the nonprofit. Based in Kyrgyzstan, Sharma has been camera-trapping and studying snow leopards for 12 years.

“By automatically analyzing the images and creating a database for us, Microsoft AI is providing our small team with the time to do more surveys and collect better data,” says Sharma, who is also the international coordinator of the Global Snow Leopard & Ecosystem Protection Program, an alliance of the governments of all 12 countries where snow leopards roam and conservation organizations.

Built by Azure Machine Learning engineers, the image classification system helps Sharma and his team save time, access critical data faster and spend more resources on protecting snow leopards from poaching, mining, climate change and other threats. The next step for the technology is automating identification of individual snow leopards, based on unique markings.

Sharma says no one really knows how many snow leopards exist and that the best available “guesstimate” is 3,000 to 7,000.

“The only way to know how many snow leopards there are is to be able to do more camera trapping-surveys in larger landscapes,” he says. As surveys expand, so will the amount of data, making AI even more crucial in protecting one of the most beautiful, elusive animals on Earth.

The nonprofit’s work is also significant to humans, because snow leopards are important “thermometers” of mountain ecologies that provide water for large communities of people.

“The presence or absence of snow leopards indicates if all is well in the ecosystem,” Sharma says. “By using cameras to study the snow leopard population, we are actually working to protect an entire ecosystem.”

Learn more about how Microsoft AI helps save snow leopards and drives innovation around the world.  

The post The Snow Leopard Trust helps protect a threatened species with Microsoft AI appeared first on Microsoft Malaysia News Center.