In the thin air of Upper Mustang, a camera trap clicks. Somewhere in 770,000 square miles of mountains that spread across a number of Asian countries—an area larger than Mexico—one of the world’s 4,000 remaining snow leopards has just walked past.
Finding that single image among thousands of photos captured by the camera trap used to take park rangers hours of manual review. Now it happens automatically, with the help of AI.
Snow leopards are known as the ghost of the mountains. They slip across international boundaries as easily as rocky ridges, leaving barely a paw print in snow that could melt by morning. Tracking these apex predators is essential to inform the protections of both the species and their habitats. It also means deploying camera traps across the world’s most rugged terrain, then drowning in the data they produce.
Nepal alone operates nearly 800 camera traps. Each generates thousands of images every few months. Every hour spent reviewing photos is an hour not spent in the field preventing poaching, managing wildfires, or working with communities.
To tackle the vast camera trap data, Tencent developed “Eye of the Species”, an AI model to enhance conservation efforts. The system identifies snow leopards from thousands of images with up to 98 percent accuracy—and it works offline in the world's most remote locations. The model doesn't just find snow leopards. It recognizes 286 species and counting, with plans to reach 2,000 by 2025.
Unlike traditional single-species recognition models, this model doesn’t require extensive data to learn new species, reducing setup costs by 70 percent. It makes it a scalable tool for broader wildlife conservation efforts across different regions and species.
Originally developed for snow leopard studies in China, the model has been adapted for use in Nepal. The Tencent team recently hosted a workshop with Nepal’s National Trust for Nature Conservation (NTNC) and China’s ShanShui Conservation Center to introduce this system to conservationists from Nepal, Mongolia, Pakistan, and other regions. The next time a camera clicks in the Himalayas, the rangers will be ready.
The cross-border cooperation mirrors the snow leopards themselves: conservation that ignores human boundaries to protect animals that never recognized them in the first place.
Yao Ma, Conservation Officer, Tencent Sustainable Social Value Organization said : The reason we focus on snow leopards is twofold. Snow leopard numbers, a strong indicator of climate change, are incredibly low. Secondly, their habitat spans multiple countries across some of the most remote and difficult terrain on earth.”
CP Pokharel, Conservation Director, National Trust for Nature Conservation said that “AI could help us analyze data faster and more rigorously, picking out not just snow leopards but also prey species from massive image datasets.”
