Remember how in the movie "blade Runner" Rick Deckard, sipping brandy from a massive glass voice commands has increased and reduced the different zones of the image on the screen of your computer in search of evidence? The picture quality did not suffer from this: there were no piles of pixels, no artifacts, nothing. Of course, this science is not yet reached, but an artificial intelligence created by the Google engineers, it can significantly improve the quality of bitmap images, thanks to clever algorithms.
A Team of specialists Google Brain on the basis of neural networks, which can significantly enrich the details in low resolution images. The first maps the low-RES image with similar images high-resolution to understand how it should or could look like in the end. The second neural network models a clearer picture based on the received first network data, while trying to achieve the most plausible results.
Of Course, the results of these improvements are still far from ideal. But in some cases the AI turns out to be very close to the original image. On a completely incomprehensible at first glance, the photo of 8 by 8 pixels, the system can recognise human faces, and even try to restore it as accurately as possible from their point of view. Such systems have great potential, because if the researchers manage to teach them to work with high accuracy, this technology may be interested in the military, intelligence agencies and the police who are constantly dealing with the videos and photos of low quality.
Over the past few years artificial intelligence has tried to teach, it seems, almost everything from complex calculations to a poker game.
what the Chinese are good at, it's utiranie nose to other States in such fields as industry, construction or even high technologies like genetics.