It’s no secret that today such popular services as Google Photos or even our cell phone cameras, when they have GPS activated, can pinpoint exactly where a photo was taken. But it’s a different story when AI knows where a photo was taken just by looking at it and without any additional data.
Artificial intelligence known as Predictive Image Geolocation (PIGEON) is trained Google Street View and you can easily determine where the photos were taken.
The developers claim that their AI can correctly guess the country where a photo was taken 95 percent of the time, and usually within 25 miles of the actual location.
Undoubtedly, this creates a huge privacy problem:
“From a privacy standpoint, your location can be very sensitive information,” said Jay Stanley of the American Civil Liberties Union. NPR.
The students were inspired by the online game GeoGuessr, which places players in a random location on Google Street View and challenges them to try to guess where they are by pinning the location on a map.
To create PIGEON, they took the CLIP neural network created by ChatGPT creator OpenAI, which learns images through text, and trained it using Street View.
“We created our own dataset of about 500,000 Street View images,” he said. NPR Silas Alberti, one of the Stanford students who developed the tool. “In fact, there is not much data, [y] “We managed to pull off quite a spectacular performance.”
“The fact that this was done as a student project makes you wonder what, say, Google could do,” Stanley said. NPR.
Source: Digital Trends
I am Garth Carter and I work at Gadget Onus. I have specialized in writing for the Hot News section, focusing on topics that are trending and highly relevant to readers. My passion is to present news stories accurately, in an engaging manner that captures the attention of my audience.
