Is It Available To Recognize People While Wearing Mask?

 With increasing people around the world wear masks to prevent the spread of the coronavirus, they are curious about whether their face or identity can be recognized while wearing a mask. As a well-known proverb goes, there are a thousand Hamlets in a thousand people's eyes, so people also have different views on face mask detection.

 

Before we discuss the availability or accuracy of face mask detection, we need to figure out some basic concept of it. What is face mask detection? How much facial features should be covered while wearing masks? What is the technical principle of face mask detection? And what are the technical difficulties of face mask detection?

 

Today, let’s talk about it together.

 

What is face mask detection?

 

Face mask detection refer to recognize the identity of people who wear mask. Namely, who are you. Face mask detection belongs to face recognition with occlusion, which requires a good recognition rate within the acceptable range of FRR. The algorithm of face mask detection involve newly mask detection (with or without) or mask area detection, in addition to the common algorithm like face detection, key point detection and face feature capture.

How much facial feature should be covered while wearing mask?

 

There is no doubt that the occlusion of the area of mouth and nose pose a great challenge to face recognition. Generally speaking, the smaller area the mask cover, the higher recognition accuracy can be achieved.

What is the technical principle of face mask detection?

 

It is mainly combine global and local features of the face and make 3D face recognition research in virtue of 3D geometric information of the unblocked part of the face. Compared with ordinary 2D face recognition, it can effectively recognize people even though they’re wearing mask.

 

What are the technical difficulties of face mask detection?

 

1 The accuracy of face detection and critical point detection will be reduced if people wear a mask.

2  Due to mask occlusion, portrait information is reduced, and the discriminability of learning characteristics will less accordingly. Specifically, 2D texture information will be lost due to occlusion, and 3D shape information will be noisy.

3  There are many types of masks and mask covering degree is different, so how to make more use of open area information is also an influencing factor.

 

In order to effectively solve the difficulties of face mask detection, Telpo face recognition terminal adopts a more advanced face recognition algorithm. Based on the original face recognition model, it improves the face visible area ratio and adopts corresponding strategies on local feature enhancement. In virtue of introducing eye feature model, it can make comparison between eye features, which can effectively solve the face recognition difficulty with large area occlusion. 

Therefore, face recognition terminal can not only support access control, daily attendance, face recognition, identity verification, liveness detection and other functions, but also mask detection. Importantly, voice broadcast for wearing masks can be provided according to scene demand. It can better facilitate places (company, community, school, shopping mall, station, etc) to do well in epidemic prevention, identify people’s identity and improve passage efficiency.