Artificial Intelligence is known to make things possible which cannot be thought of and a recent development by the researchers of University of California has once again proved just that. The group of boffins has posted a paper on arVix preprint server which explains how their created system can show someone doing movements (dance, in this case) which they are actually not capable of.
The paper bears the title “Everybody dance now” and consists all the information about how the AI algorithm renders someone dancing exactly like the real video source. As can be seen in the related video, a target person is shown doing skilled dance moves exactly similar to the one in the source. However, the latter is real while the former is artificially shown dancing similarly like the source.
The researchers also claim that they can showcase not just dance moves but can also “create a variety of videos, enabling untrained amateurs to spin and twirl like ballerinas, perform martial arts kicks or dance as vibrantly as pop stars.”
The algorithm which has made this fantasy possible actually involves a stick-like figure which seems to be detecting various points of the source’s body and creating a similar movement replica of the target. This is achieved from the GAN (Generative Adversarial Networks) deep learning algorithm which assists in not just creating a fictional moving character of the target but also makes it sharp and realistic.
Despite the lost sharpness and involvement of noise in the created video, the creation genuinely deserves applause as this can be considered as one of the most comprehensive uses of deep learning AI algorithms till now.
Click on Deccan Chronicle Technology and Science for the latest news and reviews. Follow us on Facebook, Twitter.
