Google’s machine learning will be able to remove obstructions from photos
With the Google Pixel being the highest rated smartphone camera, it’s no surprise that Google wants to improve upon it with some more machine learning. One of the things that Google’s Sundar Pichai mentioned at the I/O 17 keynote was image recognition improvements. More specifically, obstruction removal at a camera level on the Google Pixel.
He showed a photo of a girl at a baseball game, who was behind a fence from the view of the camera. Then, completely through machine learning, the fence was removed, resulting in just the girl playing baseball — no fence. Below, you can see the before and after of the photo.
It’s pretty impressive to see what machine learning can do in this case. If it actually works as well as they are showing, then this is certainly a feature to look forward to. Google didn’t tell us when it would be available, only their typical answer of “soon.” It is, however, exclusive to the Google Pixel and Pixel XL when it does launch.