Feedback
Tech

Where In the World is This Photo? Google's PlaNet Has a Guess

Google has made another breakthrough in creating machines that can do things more quickly than humans.

Last week, Google unveiled PlaNet, a neural network designed to locate the precise location where a photo was taken. What makes it different from other image location software is that it learns as it sees more images, according to the MIT Technology Review.

PlaNet is not perfect. The network is only able "to localize 3.6 percent of the images at street-level accuracy and 10.1 percent at city-level accuracy," project leader Tobias Weyand told MIT Technology Review.

However, it has already outpaced humans in trial runs.

Google restructures under parent company called Alphabet 0:27

Using a website called GeoGuessr, PlaNet was tested against 10 well-traveled humans, winning 28 of 50 rounds against the team of mere mortals.

"We think PlaNet has an advantage over humans because it has seen many more places than any human can ever visit and has learned subtle cues of different scenes that are even hard for a well-traveled human to distinguish," Weyand said, according to the publication.

But perhaps the most surprising fact about PlaNet is that you may someday be able to carry it in your pocket.

"Our model uses only 377 MB, which even fits into the memory of a smartphone," Weyand said.