Tiny Intel-powered poaching camera uses AI to detect intruders on approach
Conservationists face a huge challenge in working to protect endangered species from extinction, but lately we're seeing how various technologies can be harnessed to give them a helping hand. Artificial intelligence is one example of this, and Intel's latest chips are now poised to play a part through integration with advanced new camera systems trained to detect intruders with bad intentions.
The system is called TrailGuard AI and builds on an earlier solution, simply called TrailGuard, that has been deployed previously by non-profit conservation group Resolve in Africa's protected reserves. This first-generation version was able to detect motion and then relay images to rangers for review, who could take action if they spotted suspicious behavior.
But this approach left some room for improvement, as the images needed to be analyzed by human eyes to determine if the threat was real or if the camera trap was triggered by a false positive. We have seen other parties bring artificial intelligence into the mix to sift through the images with greater efficiency, with one 2017 solution using cameras drones to gather the images from the air showing real promise.
TrailGuard AI is designed to work on solid ground instead. The system is around the size of a pencil, meaning it can be tucked away in trees or bushes to automatically snap images as it detects motion. But by using an Intel Movidius chip, a version of which it has previously stuffed into a USB stick, it allows a deep neural network to do most of the heavy lifting.
These deep learning algorithms are said to detect humans and vehicles with a "high degree of accuracy," and are hoped to avoid a large number of false-positives and save the rangers precious time. Handily, the tiny chip also powers all of the device's processing needs, allowing it to run at low power and remain in the field for up to a year and a half at a time.
"The Intel Movidius VPU allowed us to revolutionize TrailGuard AI by adding artificial intelligence to a proven end-to-end solution to stop illegal poaching around the world," said Eric Dinerstein, director of biodiversity and wildlife at Resolve. "In addition to providing the AI technology, Intel engineers worked closely with us to build, test and optimize this incredible anti-poaching solution that will make a real difference in saving animals."
TrailGuard AI will be deployed in 100 reserves across Africa in 2019, with plans to expand into Asia and South America thereafter. You can hear from those involved in the video below.