Deep Face Drawing Artificial intelligence turns drawn faces into photo-realistic images

How to Draw Realistic faces

An artificial intelligence generates photos from line drawings. This may sound creepy, but Deep Face Drawing could have some reasonable uses.

In the future, artificial intelligence could contribute to turning simple line drawings, such as Phantom images in the police environment, into photo-realistic facial images. Researchers from the University of the Chinese Academy of Sciences (UCAS) in Hong Kong have presented a technique that is supposed to be much more realistic when converting sketches into photos than was previously the case.

Hiw to Draw Realistic faces

So far, the sketches had to be very good, so that the technology was able to create an image that could only be used approximately – even for professional draftsmen this was a great challenge, it is said. The artificial intelligence that underlies this technology, on the other hand, should be able to orientate itself on the lines, but rather to understand them as a point of reference and Inspiration. The software apparently does not need details such as lip shape or eyelids, but relies on corresponding databases in terms of probability. Left and right eye, nose part, mouth part, face shape and type of hair are distinguished and considered separately. For each Zone, a variant is generated from the sketch, which is then assembled into a 512 x 512 Pixel image.

Deep Face Drawing is sponsored by Adobe 

Initially, 17,000 drawings and the corresponding photos of the drawn persons were available. So far, it seems that they are mainly Caucasian and South American faces, which occur here. This could have to do with the data source – or with the different complexity of face shapes. The researchers do not provide any further details. In the future, technology should continue to learn based on these principles. The research is supported by Adobe, among others. Artificial intelligence in image recognition is a central field that many companies in the graphic and creative field are currently exploring.

In a detailed scientific Paper, the researchers from Hongkong University explain what the technique can do, where the limits are and how the images are calculated. In a Youtube Video you can take a closer look at the results and the way there.