Google unveils super-resolution 'zoom-and-enhance' tech

Exploiting the power of neural networks and an unrivaled database of trillions of photos, the Mountain View corporation showed that its possible to render a detailed, higher resolution image from a tiny, pixelated source, some only 8×8 pixels. Kudos to law enforcement! The first is called a "conditioning network", and it basically maps out the pixels of the 8 x 8-pixel image into a similar looking but higher resolution image.

Two artificial intelligence systems built by Google are able to transform a heavily pixellated, low quality, image into a clear photo or a person or object.

On television, this image enhancer is called "zoom in, enhance", an age-old trope that usually shows our hero unrealistically taking a blurry image, zooming in on it, and enhancing the details in order to catch the "bad guy".

Google Brain's work with neural networks has led to a plethora of interesting projects.

The prior network ingests a large number of high-res images when the source image is upscaled, it tries to add new pixels that match what it knows about that class of image. And also offer an idea as to the aspect of a pixelated image. In the case of an 8×8 block, a brown pixel on the far right and far left would correspond to eyebrows.

At the end of each neural network's process, the Google researchers combined the results to create a final image.

Ars Technica offered a simplified explanation of the process: "For example, if there's a brown pixel towards the top of the image, the prior network might identify that as an eyebrow: so, when the image is scaled up, it might fill in the gaps with an eyebrow-shaped collection of brown pixels".

The computer-generated images were accurate enough to fool a test audience. Google Brain did okay in testing, proving capable of fooling human observers 10 percent of the time with its computed images of celebrities.

In the image above, the images on the left show the 64 square pixel source images, while the ones in the middle show what Google Brain's new system can construct from them. Bicubic scaling, a method that interpolates data points on a two-dimensional regular grid, didn't manage to fool anyone.

Now, the latest Google Brain software may help bring an image revival. This is important to consider, else we might fall for it just like on TV.

The two networks working in harmony effectively redraw their best guess of what the original facial image would be. Some hunches are spot on, though, as evident in these press photos. But it looks like a bunch of Google researchers have figured out a way to do that - or at least get pretty darn close.