few-shot-learning-of-talking-heads

Unofficial implementation of "Egor Zakharov et al. Few-Shot Adversarial Learning of Realistic Neural Talking Head Models, 2019"

Work In Progress

I created generator, embedder, and discriminator on Keras. After a few days of training, it just infers black images:( I am investigating the cause. I opened this repository to get your contribution. Feel free to open an issue and make a pull-request.