News register
Search section
Search area
News Type
News type
Search date
Search word OR
List of related articles
Contents of related articles
No info was found
List of related articles
Contents of related articles
No info was found
View details

01/01/2019 HYU News > General


VR Avatars Copy Your Facial Expressions

When you smile, they smile with you


Copy URL / Share SNS


Virtual reality (VR) is explosively gaining popularity, and its technology is developing day by day. Still one drawback is that the avatars are yet to reflect the current state of the players, such as their eye movement and facial expressions. For that reason, Im Chang-hwan (Division of Electrical and Biomedical Engineering) suggested a new way that allows one’s VR avatar to copy the expression of oneself. He further commented that it's “just like the movie, ‘Ready Player One.’”

As the player smiles, the avatar in virtual reality (VR) smiles also. As they frown, the avatar frowns likewise. They also smirk, look surprised, as well as scared. What seemed only possible in the Sci-Fi movies thus far is now available in real life. Finally, our VR avatars can read and replicate our facial expressions.

Im Chang-hwan (Division of Electrical and Biomedical Engineering) studies the reproduction of players' facial expressions on the VR avatar using the electromyography (EMG). 

The suggested method employs the electromyogram (EMG), the electrical activity produced when the facial muscles move. The eight sensors that are attached around the VR head-mounted display (HMD) measure the EMG as the player makes a face and reconstructs the data to guess the expression.

Although previously there had been several attempts to utilize EMG to mirror facial expressions, the crucial problem existed in that they required too many times of enrollment (from 4 up to 14 times), during which, similar to fingerprint enrollment, the player allows the machine to scan and identify each type of facial expression. “Four-times enrollment of ten facial expressions adds up to 40 times of enrollment, which is nothing near convenient,” explained Im. “Thus, our goal was to create a system that requires the minimum number of enrollment, but detects the maximum number of facial expressions, and with outperforming accuracy.”

The avatar is replicating a player's happy and surprised facial expressions.
(Photo courtesy of Donga)

The new method gained surprising results in various aspects. First of all, only one time of enrollment was sufficient to correctly identify the expression, compared to 4 to 14 times that was required previously. Also, they could detect a total of 11 facial expressions, which is the greatest number by far, and is very soon expected to be expanded to 15. Moreover, it has achieved an astonishing accuracy of 92 percent. They were also relatively affordable and extremely comfortable since the sensors can be made of rubber or cloth.

Im evaluated the technology as the possible core technology of the next-generation VR application. “It is applicable to virtually all kinds of VR applications that use player avatar.” The prime example is Facebook’s Spaces where players use avatars to play and interact, added Im. “Basically, our ultimate goal would be to actualze Sci-Fi movies like ‘Ready Player One’ – to the point where the avatars can read not only the players’ facial expressions, but their feelings as well.”

Lim Ji-woo
Photos by Park Geun-hyung 

Copy URL / Share SNS