Deepfakes can be distinguished from real photos by their pupils: it is easy to check from a smartphone

Over the past couple of years, a large number of fake photos and videos with different

people who do things that never happened in real life. Computers have also learned to generate faces of people who never actually existed. 

Such photographs began to appear in fakeprofiles of social network users: such accounts pretend to be real people in order to engage in extortion and other types of deception on the Internet. 

Artificial faces are made using GAN -this is a neural network that learns from images of people, and then generates its own based on them. The generated face is then sent to another neural network, which tries to determine whether it is real or fake. Those considered fake are sent back for revision.

This process continues several times,As a result, the images become more and more realistic. But the authors of the new work note that such processing is still not ideal. The researchers found that many GANs tend to produce less round pupils. From such unusual eyes you can understand whether this is a real person or a generated picture. 

The authors point out that ordinary users canjust enlarge the image of a person's eyes to understand if there are irregularities in the pupil or not. Also, based on this feature, you can make software to filter photos on social networks.

To read Further:

An ultra-thin material was made from white graphene. It will replace servers

The Milky Way is not evolving as we thought: an unexpected property of the galaxy has been discovered

3D printed catalysts will solve the problem of overheating of hypersonic aircraft