Nowadays, the use of masks has become mandatory, but there are some difficulties it brings. One of them is that face recognition systems are not working properly. So, will face recognition systems adapt to masks in the future?
In many countries of the world with the new type of coronarivus epidemic mask attach has become mandatory. As a result of this, the face recognition systems he could no longer fulfill his purpose fully.
Facial recognition, using a computer to identify human faces in images or videos, and then use a computer to identify the distance between the eyes or the relative positions of the nose, chin, and mouth. to measure certain facial characteristics includes. This information is combined to create some type of facial signature or profile and is used in security systems.
A large database is required to identify faces in a crowd:
This system smart telephone unlock When used for personal identification, such as opening, the image on the camera is compared to a saved profile. This facial “verification” process is relatively is simple. On the other hand, when face recognition systems are used to identify faces in a crowd, it is an important point that the main image can be compared. profile database must.
These profiles can be collected legally by registering a large number of users in the systems, as well as sometimes secretly.
Problem with face masks:
Anything that reduces the visibility of key characteristics (such as nose, mouth, and chin) as facial signature is based on mathematical models of relative positions of facial features inhibits facial recognition. On the other hand, although the concerns about avoiding face recognition systems have increased with the coronavirus pandemic, leaked US documents are 2018 and 2019 It shows that it existed during the years.
While the debate about the use and legality of facial recognition continues, the focus has recently shifted to the challenges of wearing masks in public. At this point US National Institute of Standards and Technology (NIST), conducted a large research project to evaluate how masks affect the performance of various facial recognition systems used around the world.
The report, published in July, suggests that some algorithms can accurately identify individuals wearing masks by up to 50%. having difficulty revealed. Some algorithms even had difficulty detecting the face when the mask was worn to cover more of the face.
Face mask has been digitally added to photos to test facial recognition algorithms:
People who are currently wearing masks to train face recognition systems and there are no sets of photos that can be used for evaluation. As can be seen above, the NIST study uses various colors, sizes and positions on face images. masks by adding addressed this issue.
While the examples do not depict a person wearing a mask in a realistic way, they are thought to be effective enough to study the effects of wearing a mask on facial recognition systems.
Many companies developing facial recognition technology, mask use will continueis already preparing for a future that will even increase. At this point, a US company is asking customers to unlock their smartphones without having to unlock them. the relevant parts of their faces sells printed masks.
Face recognition systems need to adapt to masks somehow:
Masks were used in countries like China and Japan even before the coronavirus pandemic. air pollution and viral to infection it was a common defense against. It is also known that political activists wear masks in order not to be detected in the streets.
Experts predict a future with more outbreaks, rising levels of air pollution, ongoing authoritarian regimes, and a projected increase in forest fires releasing harmful smoke. For this reason, wearing a mask in the future may become normal for many people.
Interesting ‘Demonstration’ That US President Trump Plans To Have When The Coronavirus Is Defeated
As a result, face recognition systems somehow compliance with masks is required. At this point, the fixing process; It can be made on the basis of features that remain visible from the mask, such as eyes, eyebrows, hairline and general shape of the face. Such technologies are currently under development. Several manufacturers even offer solutions that claim to provide reliable results in cases with a mask.
Source :
https://thenextweb.com/neural/2020/10/11/how-facial-recognition-systems-are-adapting-to-masks-syndication/
