Hi guys. Remember my recent post some weeks back on the AI at Facebook that were talking to themselves in languages that we humans couldn't understand. Well this time another AI is doing something less scary but still very significant.
At the recent Serigraph Conference (the premier digital photography conference) scientists found Google and MIT Computer Science and Artificial Intelligence Laboratory unveiled it. It retouches pictures in manners similar to how professional photographers would do it instantly.
This was achieved using Machine Learning and the system learning from over 5000 raw and retouched pictures. It is efficient enough to be used on phones and it can produce the finished results while the user is still talking the shot.
Lead author on the Research, Micheal Gharbi an MIT graduate student in Electrical Engineering and Computer Science had this to say, "Google heard about the work that I was doing on the Transform recipe. The key is that the bulk of the Image processing is done on a Low-Resolution image which drastically reduces time and energy consumption. But instead of trying to get an High-Res image from a Low-Res input, the system creates a formula for modifying pixels ".
The 5000 images it used were provided by the Durand group and the creators of photoshop, Adobe Systems. The AI was also trained on thousands of images produced by the application of Image Processing Algorithms like that used in High-Dynamic Range (HDR) images. The AI system would also be able to speed up those Image Processing Algorithms. We can only hope that better technologies like this would keep developing to make things nicer for us.