The neural network has learned to impose mimicry one person on the face of another

Groups of researchers often experiment with video content using neural networks. Take, for example, NVIDIA, which at the end of 2017 to train a neural network to change the weather and time of day the video. Another project of its kind launched by researchers from Carnegie Mellon University have created a neural network for applying facial expressions of one person on the face of another.

The neural network has learned to impose mimicry one person on the face of another

The basis of the project was DeepFakes technology to replace the individual in the video. It is based on the generative-adversarial form of machine learning. In the framework of generative model is trying to deceive discriminatory and vice versa, so that the system understands how content can be converted to a different style.

The neural network has learned to impose mimicry one person on the face of another

Algorithm cycle-GAN transmission properties of another object is not ideal and allows the presence of artifacts in the image. To improve the performance of the neural network, the researchers used an improved version of its Recycle-GAN. It takes into account not only the position of the different parts of the face, but also the speed of their movement The neural network successfully underwent facial expressions presenter Stephen Colbert's face comedian John Oliver. Moreover, she underwent the process of flowering narcissus on hibiscus.

The neural network has learned to impose mimicry one person on the face of another

The researchers believe that the technology can be used in the film. This will speed up the process and reduce the cost of making movies. The ability of neural networks to change the weather for video will simplify the training of electric driving in different weather conditions.