No, this is not the script of a detective movie. Image enhancing has been around for a long time. But till now, about the only thing we have been able to do efficiently is removing a bit of grain and noise and maybe improve the lighting. But that is limited to huge RAW files from high quality sensors. But what about those teeny tiny pictures that may need some enhancing someday? Well, over at Google, it is happening. A brain research team has been working to develop “neural networks” to achieve that very feat. This obviously falls in line with numerous other Artificial Intelligence projects that the giant company has been up to recently.
The neural network I just talked about will presumably be a combination of networks where one network will be trying to calculate the technical aspects and populate the desired output region with pixels and the other one will be trying to fill in intricate details. Now, the program is currently only working with celebrity faces and so the second neural network has been fed with numerous faces of celebrities and bedrooms (celebs and bedrooms mhmm) to be able to fill those details in. This is “sort of” similar to Google’s attempts to make its A.I smarter by making it read romantic novels. So, basically, the details are going to be imagined and successively implemented like a masterful painter.
Alright, I know it is exciting. But, unfortunately this project may not really make it to the real world at all. At least not anytime soon. But, it is quite clear that it may have some huge benefits where by searching will have another dimension added to it and security services can add a new layer of protection.
Who knows! It is 2017.
The information in this article was collected from CNET.