Artificial Intelligence (AI) can be trained to be specific. But it seems that somehow the technology has excelled in one particular part. And that is creating fakery.
Engineers and researchers from Samsung’s AI Center in Moscow and Skolkovo Institute of Science and Technology have together created an AI model capable of generating realistic animated talking heads from photos or painting, without relying on traditional methods, like 3D modeling.
A paper posted by the research team suggests that while traditionally researchers have used a large number of images to create a talking head model, this technique can achieve it by few, or potentially even one image.
To achieve this, the researchers said that the model creates three neural networks during the learning process.
First of all, the AI creates an embedded network that links frames related to face landmarks with vectors.
Then using that data, the system creates a generator network which maps the face landmarks into the synthesized videos.
And finally, the discriminator network assesses the realism and pose of generated frames.
The model referred to thousands of YouTube videos of celebrities talking (from VoxCeleb2 dataset) at the meta-learning stage, to help it understand face landmarks and movements.
“Effectively, the learned model serves as a realistic avatar of a person,” explained engineer an the co-author of the paper Egor Zakharov.
While the AI can effectively generate animated head using a single image, Zakharov explained that training it through multiple images “leads to higher realism and better identity preservation.”
"Crucially, the system is able to initialize the parameters of both the generator and the discriminator in a person-specific way so that training can be based on just a few images and done quickly, despite the need to tune tens of millions of parameters," said the coauthors in a summary of the paper.
"We show that such an approach is able to learn highly realistic and personalized talking head models of new people and even portrait paintings."
The technology that focuses on synthesizing photorealistic head images and facial landmarks has practical applications for telepresence, including video conferencing and multi-player games, as well as the special effects industry.
In the examples, the AI has been made to show off animated headshots of famous figures, like Marilyn Monroe, Albert Einstein, Leonardo da Vinci’s Mona Lisa, and RZA from the Wu Tang Clan.
However, there is also a huge downside, like having it to create deep fakes easier.
The world was caught by storm when the AI was created and capable of creating convincing fake porn videos starring popular celebrities.
Read: An AI Capable In Creating Fake Porn, Is Starring Gal Gadot And More: A Terrifying Implication