It looks like there is a new neural network that is capable of figuring out the exact location of an image – all without the need for geotags, which is a rather impressive achievement to say the least. This particular project is headed by Google’s computer vision specialist Tobias Weyand, and it might just change the way your photo library works thanks to landmark mining.
The machine is more than capable of outperforming a human’s ability to recognize the location of an image, and even has its own algorithm to determine the location of indoor images, now how about that, considering how it is nigh impossible to obtain clues as to the place the image was shot.
Of course it is far from perfect, but the AI known as PlaNet is currently involved in a trial run where 2.3 million images were determined correctly its country of origin with a success rate of 28.4%. Apart from that, when the scope is widened to include the continent of origin, PlaNet got it right 48% of the time. Surely these scores are higher than what a human can achieve at the moment.
Basically, it can be explained this way by MIT Review, “The team created a database of geolocated images from the Web and used the location data to determine the grid square in which each image was taken. This data set is huge, consisting of 126 million images along with their accompanying Exif location data. Weyand and co used 91 million of these images to teach a powerful neural network to work out the grid location using only the image itself. Their idea is to input an image into this neural net and get as the output a particular grid location or a set of likely candidates.”