News

Can masks and masks block face recognition?

Apple officially pushed for an iOS 13.5 upgrade, claiming that it could speed up the display of the password entry bar on the iPhone with a mask instead of letting the Face ID face system continue to recognize.In fact, the trained iPhone 11Pro Max can be directly unlocked with a face mask on.

Face recognition is also successful on an iPhone with a mask

Each time an iPhone user unlocks, he or she opens the mask, and when the screen is unlocked, he or she wears the mask again.Many iPhone users choose to turn off their Face ID when they are away, switch to a pure digital password, or even turn off password protection completely.But many people don’t do this for security reasons.However, some iPhone users wear masks and pick up the phone when they use it, they find that the lock screen interface opens automatically.Over time, even when users wear masks, they can be unlocked using a Face ID, and success rates are increasing.Unlocking a blue medical mask becomes completely unavailable only when the user replaces it with a white N95 mask, but this indirectly proves that the user’s face with a medical mask has been entered into the system.From feedback from online users, it is not an example that an iPhone can unlock a Face ID while wearing a mask. Social media has found a large number of users.

Infrared Structural Perspective Mask Mask is one of the factors to identify success

But in fact, this is actually the self-learning mechanism of Face ID at work.How does the iPhone do Face ID face recognition?It uses structured light technology to project more than 30,000 invisible infrared spots through a dot matrix projector, which penetrates the mask to map the precise depth of the face.For each subsequent recognition, the system sends the current data back to the security zone of the iPhone chip and matches the stored face data to determine whether to unlock it.It can be said that the higher security is also one of the reasons why Apple chose Face ID instead of continuing to use Touch ID fingerprint recognition.Officially, the probability of a Face ID being unlocked for a similar face is only 1 in 1 million, whereas the probability of a fingerprint being unlocked by another person is 1 in 50,000, which is a huge difference.With the depth information of structured light matching, we can’t fool the iPhone with photos or videos alone. Instead, many Android phones use only front-facing cameras for 2D face recognition, and sometimes even with masks, they still recognize successfully.The increase in pass rate is accompanied by a decrease in accuracy, but Face ID is always a more reassuring 3D recognition from the perspective of user privacy security.

Self-regulated learning is one of the factors for successful recognition of masks and masks

Another important feature of Face ID is that it learns about user face changes on its own, which many people ignore.It’s impossible for a person’s appearance to never change. Some people make up and others have a beard. In this case, the Face ID may recognize the mistake and let the user unlock it with a digital password.But Apple also made it clear in its Face ID Security Guide guide that as long as the face data identified is complete enough and users can enter the correct password immediately after the Face ID is unlocked incorrectly, the Face ID will also store the wrong face data and update its own database.

So, while we wear masks and then keep trying to unlock them, on the surface, Face ID always shows recognition failures, but as long as you immediately unlock them with your password, it will still record your face again and again.This is actually the process of autonomous learning.Over time, when the number of unlocks exceeds a certain threshold, a new Face ID data is generated, which is based on the facial features of your masked face.This may also explain why Face IDs do not recognize how many people wear masks until some time later.

There are also reasons why Apple does not publicize Face ID self-study in a high-profile way.After all, this kind of autonomous learning is out of Apple’s control and entirely dependent on the iPhone’s own intelligence.Some iPhone users spend less than a month before and after, which may also be related to the number of times they unlock using their Face ID in normal time.To speed up this learning process, some iPhones continue to actively input data to Face ID, making hundreds of identifications before completing automatic learning for some iPhones.However, it is also difficult for Apple to give a quantitative indicator of when and how long the iPhone on their hand will take to record face status under the hood.

Accurate Face ID Recognition for Faces with Masks

Face recognition with masks is also not 100% accurate.Basically only when your face is facing your iPhone can you successfully unlock it;If the angle is slightly more difficult, looking up and sideways at the screen may fail five times in 10 attempts.

The uncertainty of this result is probably one reason Apple didn’t publicize Face ID self-learning in a high-profile way.After all, if you want to achieve 100% unlock success, enter your password, or remove your mask, it’s still the best practice.