Transporting yourself into a video game, body in addition to all, just got easier. Artificial intelligence has been used to create 3D versions of people’s bodies for virtual reality avatars, surveillance, visualizing fashion, or movies. however of which typically requires special camera equipment to detect depth or to view someone through multiple angles. A completely new algorithm creates 3D versions using standard video footage through one angle.
The system has three stages. First, of which analyzes a video a few seconds long of someone moving—preferably turning 360° to show all sides—in addition to for each frame creates a silhouette separating the person through the background. Based on machine learning techniques—in which computers learn a task through many examples—of which roughly estimates the 3D body shape in addition to location of joints. inside second stage, of which “unposes” the virtual human created through each frame, producing them all stand with arms out in a T shape, in addition to combines information about the T-posed people into one, more accurate product. Finally, inside third stage, of which applies coloring in addition to texture to the product based on recorded hair, clothing, in addition to skin.
The researchers tested the method using a variety of body shapes, clothing, in addition to backgrounds in addition to found of which of which had an average accuracy within 5 millimeters, they will report in June at the Computer Vision in addition to Pattern Recognition conference in Salt Lake City. The system can also reproduce the folding in addition to wrinkles of fabric, however of which struggles with skirts in addition to long hair. using a product of you, the researchers can change your weight, clothing, in addition to pose—in addition to even make you perform a perfect pirouette. No practice necessary.
, http://feedproxy.google.com/~r/DrudgeReportFeed/~3/Cz6NyLFYeSQ/watch-artificial-intelligence-create-3d-product-person-just-few-seconds-video, 16 April 2018 | 3:58 am
Source : Watch artificial intelligence create a 3D product of a person—through just a few seconds of video | Science