Before the advent of camera traps, conservationists were tasked with tracking animals by using technologies like radio, GPS, and satellite collars. While this method provided relatively accurate data on the animals position, it meant conservationists would need to get dangerously close to the animal to place these collars around their neck, either by sedating the animal or by force.
Thankfully, Google are putting their Wildlife-spotting AI tool to good use and using it to help conservationists and researchers track wildlife.
Camera traps are automated cameras that takes a series of pictures every time an animal walks past it, helping conservationist estimate the the population numbers of a particular animal. One downside to camera traps is that each camera snaps thousands of pictures, meaning researchers would have to sift through a sea of unusable pictures before finding what they need.
Tech giants Google have come up with an innovative solution to this problem. Their Wildlife-spotting AI tool, which is used on their Wildlife Insights website uses machine learning to automatically sort out pictures captured by camera traps from all around the world.
“It’s very difficult to get from the raw data to a clean data set because there are thousands of images to look at and keep track of. We wanted to work in a system that would improve that using artificial intelligence, and that’s why we partnered with Google” Jorge Ahumada, executive director of Wildlife Insights told Sciencemag.com
The AI system analyses the images from a particular trap and removes all photos that don’t, or are unlikely to have any animals in them. Along with removing unwanted images, the AI also automatically labels any animals it does come across. Googles Wildlife-spotting AI not only speeds up the laborious task of sifting through thousands of images and labeling them, it also gives conservationists and researchers a chance to see how animals are immediately affected by human intervention and climate change.
As for Wildlife insights itself, the website is home to around 4.5 million images of animals in their natural habitats from all around the world. By using the detailed map on their site, users and researchers alike can filter for species, dates, and countries.
Feature image: Pixabay