Featured Image Credit: Desmond Lee via Youtube
Dugongs— also called Sea Cows— are pretty photogenic, well in our opinion. Being a close relation to manatees, why wouldn’t they be?
But they’re also camera shy which proves to be a bit of a problem for the scientists that want to keep up with them. Of course, that doesn’t mean that they’re afraid of having their picture taken. In fact, dugongs are very social animals, but they’re also a mysterious mammals who are vulnerable to endangerment.
The best way for the scientists to track them, aside from going out and finding them in the wild, which is usually done by flying in a plane above the ocean off of Australia… is by using drones!
While it’s a great solution, it also presents a teeny problem: when you’re taking pictures with drones from high above the water’s surface… there’s bound to be a little difficulty finding even the largest animals.
And combing over photos like this one all day…?
Phew, we couldn’t! Here’s where it is:
Now, imagine doing that: but for thousands and THOUSANDS of photos.
Which is exactly why some researchers, like Dr. Amanda Hodgson at Australia’s Murdoch University and Queensland University, have decided to try to make their machines a little smarter. By using a Google-created open-source program called TensorFlow, they have created a way that the computers could learn to automatically find dugongs in photos.
By uploading photos like the ones above, the researchers marked where the dugong was located. And just like our eyes are pulled to the area in the first image once you’ve spotted the dugong, the computer can find it, and many many others!
If all goes according to plan, these and other researchers will be able to find dugongs by the hundreds, and keep a closer eye on them and how they’re affected by humans. After that, hopefully, enough so to find a solution to keep them safe!