TrailGuard: How AI can help us tackle poaching

The world's human population may be booming, but we're facing a worrisome mass extinction of other species, caused by a variety of factors, such as diseases, mass urbanization, and the biggest of all, climate change. But these factors have one common denominator: Humans.

There are about 250,000 parks in the world, and 80,000 of those are currently under threat because of poachers. To this day, animals are poached to a massive scale, and governments have utterly failed at finding a solution. Rhinos, tigers, bears, and elephants are only a few of the most poached animal species around the world, and putting a stop to it is crucial for our own survival. Ecosystems are shaped by biodiversity, made up of animals of all walks of life that affect vegetation, which in turn affects all living things on this planet, including us. It isn't a stretch to say that protecting animals is protecting our future.

When I was in South Africa, I had the privilege to experience this problem first-hand. For two weeks, I stayed at a private reserve owned by a South African couple that decided to dedicate their lives and savings to protecting rhinos from poachers. I was there in 2017, and I'm glad to say that poaching numbers for rhinos have declined steadily over recent years, although numbers are still too high. Other African species haven't been as lucky.

In the last decade, there has been a 62% decrease of African Elephant populations, and it is highly expected, if nothing is done, that this species will be extinct within the next 10 years. Poachers kill an estimate of 35,000 African Elephants every year. Why? Because a single pound of tusk ivory sells for about $2,500, and each tusk weighs at least 250 pounds. Do the math.

Thankfully, there are several active projects, led by inspiring, talented individuals that are dedicating significant work to tackle the poaching problem, and many of them are effectively using Artificial Intelligence to do so.

The Mara Elephant Project

One of these projects is the Mara Elephant Project, a protection organization that does active anti-poaching work, research and monitoring of elephants. They were recently featured on an episode of YouTube's 'The Age of A.I.', where they talk about how they worked alongside Intel and Resolve to implement TrailGuard, an AI-powered technology that efficiently catches poachers in the act and enables park rangers to catch them before damage is done.

The main problem with catching poachers has always been the lack of resources in order to be at the right place at the right time, and that's where TrailGuard comes in.

TrailGuard's Deep Learning to detect poachers

TrailGuard was designed by Resolve, and it was made to be hidden in bushes and detect motion by using a built-in camera and an offline AI algorithm which recognises humans and vehicles within the camera frame, and is able to alert park headquarters within the span of two minutes.

The way TrailGuard works is by holding together layers of functions that basically imitate the behavior of neurons in our brains. In other words, it frequently selects and sends over photos over a local network. The AI system was trained to recognize suspicious, or 'out of place' elements within the camera frame by feeding it hundreds of photos from many angles and contexts, giving it what it needed to judge and select which photos had anomalies.

By placing cameras within the known several poacher routes, TrailGuard allowed park rangers to intercept poacher gangs more efficiently than would have been possible by more conventional means.

Resolve paired up with Intel in order to implement TrailGuard into Intel's Movidius Myriad 2 vision processing unit, also known as VPU, which is the same chip used in other cameras such as Tencent's DeepGaze, or DJI's Phantom 4, or even Google's Clip. What it does is precisely speed up the machine learning algorithms to get a faster response, which is essential in tackling poachers in parks.

The implementation of TrailGuard represented a huge advance in answering the question of how A.I. could be used to save the environment. Apart from Intel, Resolve has worked with National Geographic and the Leonardo DiCaprio Foundation in order to deploy TrailGuard to over 100 reserves throughout Africa.