A Review on Face Reenactment Techniques

Existing Face Re-enactment approaches have two major limitations, first, they require large dataset of images to create photo-realistic face models and second, they do not generalize well if the facial images are not available in training dataset. The generation of a new facial reenactment requires large image dataset and hours are required to train these models. Some progress in Deep Learning has shown quite significant results using Generative Adversarial Networks (GANs). Recent works in GAN have solved the problem of large dataset training dataset by introducing the concept of few-shot learning. This paper reviews existing approaches in Face Re-enactment with few-shot learning techniques and other approaches in Face Re-enactment.