Groups of researchers often experiment with video content by using neural networks. Take for example NVIDIA, which at the end of 2017, the trained neural network to change the weather and time of day on video. Another similar project was launched by researchers from the University of Carnegie-Melona, who created a neural network for blending facial expressions of one person on the face of another.
The basis of the project was the technology DeepFakes for the substitution of the persons in the video. It is based on generative-competitive form of machine learning. In the framework of the generative model is trying to deceive, discriminatory, and Vice versa, so that the system understands how content can be converted to another style.
Algorithm cycle-GAN transmission properties of another object is not ideal and allows for the presence of artifacts in the image. To improve the performance of the neural network, the researchers used his improved version of the Recycle-GAN. It takes into account not only the position of different parts of the face, but also their speed
The neural network successfully transferred the facial expressions of the television host Stephen Colbert on the face of comedian John Oliver. Moreover, she underwent the process of flowering daffodils to hibiscus.
Researchers believe that the technology could be used in film. This will speed up the process and reduce the cost of creating films. The ability of neural networks to change the weather in the video will facilitate the learning of electric cars driving in different weather conditions.
The neural network learned how to apply the facial expressions of one person on the face of another
Ramis Ganiev