Do you have any image of which you want to determine the location? If yes, then it is possible due to Google’s new AI. It has the ability to determine the location of any image by using the image pixels.
The person behind this new technology inside Google’s Campus is Tobias Weyland, who is a computer vision scientist at Google. He has helped Google to make a neural network supported by over 91 Million Geo-Tagged Images captured across the world, hence making it possible to determine the location of almost any image. Moreover, it can also determine flora and fauna, typical objects and even different landscape in the images.
PlaNet examines the pixels in Images and after that it cross reference’s images with the millions of images present in its database and then check its resemblances. Neural Network utilizes about 377MB space, which is not a problem for Google.
Initially it was tested with over 2.3 Million Images and this new AI was effective in determining the country of the image within accuracy of 28.4 percent. Amazingly, it can determine the Continent of image with accuracy of 48 percent.
PlaNet is also effective in determining the location of nearly 3.6 percent of images at Street Level Accuracy while as it has 10.1 percent accuracy in determining city level images. PlaNet isn’t fully accurate yet, but Google is working to make it better day by day.
PlaNet also carried several tests and it has successfully overcome humans.
MIT Technology Review stated that “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,”.
It is really a nice move by Google, which could help us in determining the location of almost any image. If you liked this article, feel free to share it.
Post A Comment:
0 comments:
Post a Comment