Realistic masks are being used to train car software to learn about humans

A small company in Japan that makes hyper-realistic face masks has found popularity within the tech industry. The masks, which cost about 300,000 yen (US $2,650 USD) to create, are made of resin and plastic by REAL-f Co. at the company’s location in Otsu.

Explain.

The company’s founder, Osamu Kitagawa, spent two years developing a method to translate facial data from high-quality photographs to 3D masks, and REAL-f Co. now receives about 100 orders a year. Each mask replicates the tiniest details of a person’s face, down to an eye’s blood vessels and fine skin wrinkles.

Of course these masks have any number of real-world applications, but I have so many questions completely unrelated to those, like which face will appear in my nightmares tonight, and why does the company’s Q&A section have a question that says, “Is it OK if I lick it?”

REAL-f Co. says it would like to see the masks used in the future for medical purposes and developing humanoid robots at a lower cost. There’s also the obvious use case of using them to test and try to thwart security measures. A Japanese car company, for example, ordered a mask of a sleeping face in order to train its tech to detect when a driver nods off. Apple also used equally creepy face masks to test false attempts at logging into the iPhone X with Face ID. (Turns out it wasn’t enough, as Vietnamese cybersecurity firm Bkav claimed it was able to bypass Face ID with a cobbled-together mask.)

“I am proud that my product is helping further development of facial recognition technology,” Kitagawa tells Reuters. “I hope that the developers would enhance face identification accuracy using these realistic masks.”

 

By Dani Deahl

This article was originally published on The Verge (www.theverge.com) and has been republished under Creative Commons