This Picasso painting had never been seen before. Until a neural network painted it.

Background: A woman’s face has always been faintly visible beneath the paint of one of the most famous paintings from Picasso’s Blue Period, The Guitarist. X-rays and infrared analysis in the 90s showed a seated woman holding out her left arm. But we had no idea what the original really looked like. So researchers have used a machine vision technique called neural style transfer to "retrieve" it.

The technique: Neural style transfer uses a neural network that is trained in a way that allows it to convert any image into the style of a particular artist. In the past, it has been used to produce artworks, comics, or even movies in the style of any chosen artist.

The big reveal: The team took a version of the X-ray images of the hidden image and passed it through a neural style transfer network trained on art from Picasso’s Blue Period. The result is a full-color version of the painting in the style that Picasso was exploring at the time. Of course, there's no way of knowing if this is exactly what the original looked like (probably not), but it still gives art historians a fascinating insight into a long-lost classic. Read more here.