Earlier on, AIMed reported some users who are constantly wearing face masks as a precaution practice during COVID-19 pandemic face difficulties unlocking their phones or other devices that rely on Face ID. Baskin, a San Francisco based artist took the opportunity to produce masks that are printed with faces that were generated by artificial intelligence (AI).
The faces found on these masks are not real, neither were they the faces of the users. Despite so, Baskin’s attempt still won the hearts of many, especially those residing in China as the local government employed facial recognition technology to keep track of its citizen movements. Baskin’s masks have since become objects of anti-surveillance
Unlocking mobile devices with mask on
Even if Baskin did not step in with her creativity, the fact that human is able to recognize their friends even if they have face masks on but machines can’t mean there’s a significant flaw in facial recognition technology today. Indeed, presently, there is no easy solution to overcome the challenge. As such, instead of tackling it directly, Apple’s third beta version of iOS 13.5 decided to go around it.
Typically, the front-facing TrueDepth sensor on devices with Face ID will enable rightful users to gain access. If faces are covered up, let say by face masks in this case, the lock indicator will be alerted and the phone will hence vibrate, suggesting something has gone wrong. After which, one will have to wait until Face ID is time out before they can swipe up the screen to enter passcode. According to the video shared on Twitter last week, users will no longer need to wait. Once Face ID identifies a hinder which prevents it to proceed, it will jump to the passcode enter screen.
The change is welcomed by many because users no longer have to remove their face masks and they can continue to use contactless payments via their devices. However, Apple did not reveal if this developer version of iOS 13.5 is finalized and will be released to the public soon. Nonetheless, a new contact tracing API Apple co-developed with Google is believed to be available. This new feature aims to help health authorities to find out if someone has been exposed to coronavirus in an anonymous and private manner.
A technology remains in question
As mentioned, Apple’s devices have obviously failed to learn how to read faces with masks on. The US Customs and Border Protection uses facial recognition on international flights travelers and Russian company Ntechlab said they have 150,000 cameras over Moscow. Both claimed that their technology can identify covered faces. However, experts have doubts because it’s not easy to validate such claims and there is no data nor evaluation available at the moment to do so.
Researchers from the National Institute of Standards and Technology (NIST) have been quantifying how accurate algorithms can perform on identification of individuals with face masks but their work was disrupted by the COIVD-19 pandemic. China’s SenseTime, a promising AI startup which provides Chinese government and health agencies facial recognition technology said their software is ready to make identification of semi-covered faces by focusing on eyes and upper part of one’s nose.
Regardless of which, facial recognition was mostly deployed for surveillance purpose. Even if it’s not perfect and may subject to controversy, creators still believe that the technology is invincible while individuals in these areas are constantly being watched with little to nothing they can do.