Communications of the ACM
'Mona Lisa' Comes to Life in Computer-Generated 'Living Portrait'
Researchers in Russia have created animated portraits of the Mona Lisa as digitally modified images or "deepfakes."
Credit: Egor Zakharov
Researchers at Samsung's AI Center in Moscow and the Skolkovo Institute of Science and Technology in Russia have created artificial intelligence (AI)-animated portraits of the Mona Lisa as digitally modified images or "deepfakes."
The researchers used a self-training convolutional neural network to analyze reference images, and then apply facial movements from a series of frames to a static image.
These living portraits' performance improves as the network uses more angles and reference images.
The researchers suggested the AI could generate "perfect realism" with only 32 reference images.
The neural network monitored three different training videos, and the three resulting living-portrait iterations of Mona Lisa based on those frames seem to have distinct personalities.
From Smithsonian.com
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA
No entries found