Sometimes, when we see a picture of a beautiful place in the web, we would like to know where it was made to go one day. Thanks to state-of-the-art technology, it will soon be possible, as Google engineers are working on the PlaNet project, which is a neural network dedicated to this task.
PlaNet is one of Google’s projects designed to locate where a particular photo was taken. The American company believes that artificial intelligence works well with this task, so it is working on a new neural network that uses image recognition technology, which will briefly determine the location of the image in the image.
Technology is based on a number of different elements, analyzing the style of buildings, plants, language used, and then compares them to a database of 126 million geotagged photos, divided into 26,000 screens. The system based on the vegetation period of the plant will be able to determine that the photo was taken in Brazil, and if it sees a sign such as, in Portuguese, informs the user that the photograph could have been made in that country.
PlaNet is not an ideal system at present, because the current level of technology development does not allow it to be created. Therefore, it will be able to locate the location of the image only if it finds the appropriate data in its database. Besides, the system does not cope well in urban areas and the accuracy is only 10 percent.