Modern artificial intelligence techniques are revolutionizing the ways endangered snow leopards are tracked and counted, providing new hope for protecting these rare creatures known as the “ghosts of the mountains.”
Article contents:
- Introduction
- Snow leopard tracking challenges
- Traditional methods and their limitations
- Artificial intelligence solutions
- Unique pattern identification technology
- Results and accuracy
- The role of human experts
- Conclusion
Introduction
Snow leopards are known as the “ghosts of the mountains” for good reason. Imagine waiting for months in the harsh and rugged mountains of Asia, hoping for just a single glimpse of these creatures. These elusive big cats move silently across rocky slopes, where their pale coats blend so seamlessly with snow and stone that even the most experienced biologists rarely spot them in the wild.
Travel writer Peter Matthiessen spent two months in 1973 searching for them on the Tibetan Plateau and wrote a 300-page book about the effort, but never saw a single one. Four decades later, his son Alex traced his father’s footsteps—and didn’t see one either.
Snow leopard tracking challenges
Researchers have struggled to come up with a figure for the global population of these animals. In 2017, the International Union for Conservation of Nature reclassified the snow leopard from endangered to vulnerable, citing estimates ranging between 2,500 and 10,000 adults in the wild. However, the group also warned that numbers continue to decline in many areas due to habitat loss, poaching, and human-wildlife conflict.
Traditional methods and their limitations
Traditional tracking methods—searching for footprints, feces, and other markings—have limits. Rather than waiting for a lucky face-to-face encounter, conservationists from the Wildlife Conservation Society, led by experts including Stéphane Ostrowski and Soroush Pouyani Fard, began deploying automated camera traps in Afghanistan. These devices take photos whenever movement is detected, capturing thousands of images over months, all in the hopes of catching a rare glimpse of a snow leopard.
But taking photos is only half the battle. The next, and more difficult, task is distinguishing one snow leopard from another. At first, it might seem simple: every snow leopard has a unique pattern of black rosettes on its coat, like a fingerprint or a face in a crowd. In practice, however, identifying individuals through these patterns is slow, subjective, and prone to error.
Artificial intelligence solutions
Now people trying to count snow leopards are getting help from artificial intelligence systems in two ways. First, AI can quickly sift through thousands of images, flagging those containing snow leopards and discarding irrelevant ones such as those depicting blue sheep, grey and white mountain terrain, or shadows.
AI can identify individual snow leopards by analyzing unique rosette patterns, even when poses or lighting differ. Each snow leopard encounter is compared against a catalog of pre-identified images and assigned a known ID if there is a match, or entered as a new individual if not.
Unique pattern identification technology
In a recent study, several colleagues and I evaluated two AI algorithms, both separately and in coordination. The first algorithm, called HotSpotter, identifies individual snow leopards by comparing key visual features such as coat patterns, highlighting distinct “hot spots” marked with a yellow tag.
The second is a newer method called pose-invariant embeddings, which works like facial recognition technology: it recognizes layers of abstract features in the data, identifying the same animal regardless of how it is positioned in the photo or what kind of lighting is present.
Results and accuracy
On their own, each model worked about 74% of the time, correctly identifying the cat from a large image library. But when combined, the two systems together were correct 85% of the time. These algorithms were integrated into Wildbook, an open-source, web-based software platform developed by the non-profit organization Wild Me that has now been adopted by Conservation X Labs.
The role of human experts
These intelligent systems are not error-free. AI quickly narrows down candidates and flags potential matches, but expert verification ensures accuracy, especially with difficult or ambiguous photos. We conducted another study in which groups of AI-backed experts and novices pitted against each other.
Experts accurately matched about 90% of the photos and provided population estimates within about 3% of the true count. In contrast, novices identified only 73% of the cats and underestimated the total count, sometimes by 25% or more.
Conclusion
The takeaway is clear: human expertise remains essential, and combining it with AI support yields the most accurate results. We hope that using tools like Whiskerbook and the AI systems embedded within them, researchers will be able to study these elusive animals with greater speed and confidence. With AI tools like Whiskerbook unlocking the secrets of these mountain ghosts, we have another way to protect snow leopards—but success depends on a sustained commitment to safeguarding their fragile mountain homes.
