Notes

it is a system in which you can identify where photos were taken
you can examine small clues hidden in pixels
program called PlaNet uses the more images it sees the more smarter it will be
the geotagging abilities are already better than any human or program. It is still not done getting smart
PlaNet is able locate 1.3% of the images at street level
PlaNet still has a long ways to go before being a correct and reliable source but it already is smarter than a travelling human
PlaNet challenged about 10 travelling people on any street level picture. They played 50 rounds and PlaNet won about 28 of them
There are other computer softwares trying to do the same thing. Im2GPS has gotten high accuracy by getting an image and identifying the location
The Im2GPS can look at a forest picture scroll through the thousands of forest pictures on the internet find a match and say that these are the same.
PlaNet took a different approach and divided the earth into pixels that are geotagged
The researchers have skilled the neural network using millions of google images.
The network relies on clues to help identify the picture
The invention does have limits because it is based on the internet and it has a disadvantage because of the rarely photographed places
the invention does not have in its system artic places and oceans
Tobias the leader behind PlaNet says that attaching internet photos with satellite photos decreases the disadvantage.
PlaNet also focus on things other than landmarks and can identify non cities.
PlaNet could fit on smartphones.