Artificial intelligence makes blurry faces look 60 times more sharp
NOV. 12, 2020
Researchers have developed an AI tool that, in more detail than ever before, can turn blurry faces into eerily convincing portraits generated by computers. Previous methods can scale a picture to a resolution of eight times its original.
Researchers at Duke University have created an AI tool that can transform blurred, unrecognizable images of people's faces into eerily convincing portraits produced by computers, in more detail than ever before.
Previous methods can scale up to eight times its original resolution to an image of a face. But the Duke team has come up with a way to take a handful of pixels and create realistic-looking faces with up to 64 times the resolution, 'imagining' features like fine lines, eyelashes and stubble that wasn't there in the first place.
In a few seconds, the device can transform a face picture of 16x16-pixels to 1024 x 1024 pixels, adding more than one million pixels, equivalent to HD resolution. In the computer-generated versions, details such as pores, wrinkles and wisps of hair that are imperceptible in the low-res photos become crisp and clear.
The researchers asked 40 people to rate 1,440 images generated by PULSE and five other scaling methods on a scale of one to five, and PULSE did the best, scoring photos of actual people almost as high as those of high quality.