• A Chinese facial recognition firm, Hanwang, claims its technology can accurately identify people wearing masks 95% of the time.
  • Hanwang's chief technology officer Huang Lei told the FT that its tech previously recognized mask-wearers half the time, but said this rate improved after it created a database of photos of people wearing artificially-generated masks.
  • He added that Hanwang had received requests from the original coronavirus epicenter, Hubei province, to update its software to recognize nurses wearing masks.
  • One expert who spoke to Business Insider said Hanwang's claim of 95% accuracy was plausible.
  • Visit Business Insider's homepage for more stories.

A Chinese company claims it has improved its facial recognition tech to recognize people wearing masks 95% of the time.

Increased usage of facial recognition masks during the coronavirus outbreak has foiled facial recognition cameras and technologies such as Apple's Face ID feature. That has raised questions for governments, police forces, and companies that rely widely on facial recognition tech.

But the chief technology officer of Chinese firm Hanwang, Huang Lei, has told the Financial Times that its tech previously recognized mask-wearers half the time, but said this rate improved after it created a 6 million-strong database of photos of people wearing artificially-generated masks.

He added that Hanwang had received requests from the original coronavirus epicenter, Hubei province, to update its software to recognize nurses wearing masks – while insisting this wasn't the reason it created the vast database.

"We wouldn't wait until something explodes to act. If three or five clients ask for the same thing . . . we'll see that as important," he told the newspaper.

Beijing-based Hanwang specializes in recognition-focused technology like handwriting recognition tools and biometric fingerprint software, as well as facial recognition cameras. The FT says it has placed 2 million of these cameras at entrance gates across the world.

One expert who spoke to Business Insider anonymously said Hanwang's claim of 95% accuracy was plausible, but added that the company hadn't been sufficiently clear about whether the tech would work better on certain kinds of people.

LoadingSomething is loading.