Artificial intelligence has learned to generate human faces - Hitecher
Искусственный интеллект научился генерировать лица людей

Artificial intelligence has learned to generate human faces

by Hitecher

The word combination "neural network" has become quite fashionable, but most people only know in theory what it stands for. Site thispersondoesnotexist.com, created to work with the StyleGAN neural network, is a pleasant exception.

The site’s author is a software engineer Philip Wang who created a simple system of interaction with the neural network which, in itself, is the project by Nvidia based on the AI framework invented by Jan Goodfellow. The neural network is intended to generate images based on the given parameters.

Each time you click on a link, or update a page, the site will display a random human portrait. They are not real pictures, because the people shown on the site have never existed: all faces are generated by StyleGAN.

The neural network can create not only people’s faces, but also those of animals, cartoon characters and everything that you can fantasise. The main thing is to have the initial data.

The technology has a good chance to get widespread adoption in the cinema, for example, to create crowd scenes, such as battles, for example. Learn more about it from the video below.

Share

No Comments




Your email address will not be published.