AI-equipped cameras will help spot wildlife poachers before they can

10Jan 2019
The Guardian
AI-equipped cameras will help spot wildlife poachers before they can

ELEPHANTS are proverbially hard to miss, but even these huge beasts can be swallowed up in the vast plains of Africa. This is a big problem for park rangers whose job is to protect the animals from poachers. In Serengeti National Park in Tanzania, for example, there are just 150 rangers responsible

The innards of the TrailGuard AI camera, showing Intel’s Movidius Myriad 2 VPU chip inside.

for safeguarding an area of land roughly the size of Belgium.


A new solution to this proposed by conservation nonprofit Resolve is to use AI-equipped cameras to act as remote lookouts. Today, Resolve announced a new custom-made device called TrailGuard AI, which uses Intel-made vision chips to identify animals and humans that wander into view. The cameras will be placed on access trails used by poachers, automatically alerting park rangers who can check up on any suspicious activity.


TrailGuard AI builds on past work by Resolve to create remote cameras to aid conservation. However, early devices were bulky, had limited battery life, and were unsophisticated, sending images to rangers every time their motion sensors were tripped. This resulted in lots of false positives, as the cameras would be triggered by non-events, such as the wind shaking tree branches.


The new device, by comparison, is no thicker than a human index finger, has a battery life of a year and a half, and can reliably identify humans, animals, and vehicles. The chip used by Resolve is Intel’s Movidius Myriad 2 VPU (or vision processing unit), which is the same technology that powered Google’s automatic Clips camera.


Eric Dinerstein, director of biodiversity and wildlife solutions at Resolve, compared the cameras to an “intelligent sentinel.” Dinerstein said in a press statement that “AI-driven smart cameras will help park rangers identify poachers and stop them before they can kill.”


This is not the first time we’ve seen AI’s potential in combating poaching, though. The object recognition powers of machine learning are a natural fit for surveillance and conservation work, and researchers have even used AI to try to predict where poachers will strike by studying databases of past hunts.


Resolve notes that despite current conservation efforts, an African elephant is killed roughly every 15 minutes. At this rate, the surviving population of 100,000 animals (down from a peak of some 2 million) will be destroyed over the next few decades. Resolves hope is that artificial intelligence will help fight against this trend by offering a pair of digital eyes.


In partnership with the National Geographic Society and the Leonardo DiCaprio Foundation, Resolve hopes to deploy TrailGuard AI in 100 reserves and national parks in Africa, starting with Serengeti and Garamba.