Working at the Skolkovo Institute of Science and Technology, the scientists have been able to simplify "realistic neural talking head models" which normally require a huge dataset of images to look genuine.
The researchers created life-like talking heads with just a few images of a person and even in some cases just a single image.
"Here, we present a system with such few-shot capability," write the scientists. "It performs lengthy meta-learning on a large dataset of videos, and after that is able to frame few- and one-shot learning of neural talking head models of previously unseen people as adversarial training problems with high capacity generators and discriminators. 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. We show that such an approach is able to learn highly realistic and personalized talking head models of new people and even portrait paintings."