AI Development, Machine Learning

China Company Develops Facial Recognition with Mask Algorithm

China Company Develops Facial Recognition with Mask Algorithm


Hanwang Technology, also called Hanvon in some countries, recently announced a new, sophisticated facial recognition method. The big selling point? It can recognize people wearing masks.

This artificial intelligence technology is the first of its kind , even for a country as tech-forward as China. It’s just the right time for such a system to come out, too, as more people wear face masks to prevent the spread of COVID-19. Facial recognition software that works around masks is ideal for the Chinese government for two reasons.

First, since the coronavirus originated in China, controlling the spread of disease is especially important there. Any technology that allows people to keep wearing masks to protect themselves and others would be welcome. Secondly, facial recognition is particularly relevant in China, where the technology plays a more significant role in society.

China and Facial Recognition Technologies

Having functional facial recognition software, a computer vision technology , may not be that major an issue in all countries. But this AI technology sees a lot more use in China, in both commercial and government applications. The nation has a deep connection with facial recognition technologies.

Many Chinese office buildings and condominiums use face scans to grant people entry. Without working facial recognition, some people may not be able to get where they need to go. At the very least, it means they can’t enter buildings as quickly and effortlessly as they used to.

In more than 200 airports throughout the nation , airlines and security use facial recognition to let people on airplanes. It’s a central security measure for many day-to-day actions in China. But perhaps it’s most noteworthy application, and certainly the most controversial one, is in government surveillance.

China’s Ministry of Public Security, the department that runs the police, relies heavily on facial recognition. The Ministry uses this software to identify suspects in a crowd or find wanted criminals. In a nation of more than one billion people , these technologies make it easier to police the population.

The use of advanced surveillance tech, however morally dubious, is an essential part of Chinese government operations. The surge of mask-wearers brought about by the coronavirus outbreak has made things difficult for police. But Hanwang’s latest facial recognition system seems to provide a solution.

How It Works

Even though medical masks don’t block your entire face, they do cover a lot of it. Masks hide your chin, mouth, nose and parts of your cheeks, leaving recognition software with few reference points. As a result, most systems, like your iPhone’s FaceID, won’t work if you’re wearing one.

Hanwang’s system works around this by focusing on features a mask won’t cover. Instead of identifying you based on your entire face, it does so based on the area around your eyes. Your eyes, eyebrows and the bridge of your nose all serve as reference points for the software.

According to Hanwang, the technology can identify masked individuals with 95% accuracy. For unmasked people, this figure reaches a remarkable 99.5% accuracy. But if someone were to wear both a mask and sunglasses, the system would be ineffective.

The system also comes in two variations: a single-channel version and a multi-channel one. The single-channel variant is more simplistic and works best for things like building entry. The multi-channel version uses multiple cameras to identify up to 30 people at once.

Cloud technology, which offers near-limitless storage space , makes these systems even more useful. Officials could compare facial scans to a database of millions of people to identify people accurately in a matter of seconds.

On top of the facial scanning technology, this system can use temperature sensors. That way, people could both identify someone and see if they have a fever, one of the symptoms of COVID-19. Roughly 200 clients, including police forces, have already started using the technology.face

Got a computer vision project?

Hire us for your computer vision or machine learning projects. Learn more about computer vision and machine learning and read our computer vision case studies.

Get Started

Why Does This Matter?

Mask-proof facial recognition software has significant potential in both health and surveillance. In crowded areas like airports, taking your mask off could put you at risk of contracting the virus. But if you didn’t have to remove your mask to scan your face, this wouldn’t be an issue.

With the addition of temperature sensors, this tech could help improve public health even more. Healthcare workers could use it to identify possible carriers of COVID-19. They could then take measures like barring them from entry into some areas to prevent the virus from spreading.

Authorities could also increase security by better identifying suspects. But this is where the controversy begins. While this technology can help stop criminals, the government could also use it to fight against protesters.

During last year’s noteworthy Hong Kong riots, authorities banned citizens from wearing masks in response to the protests. Protesters had been wearing them to protect their identities from oppressive government forces. If facial recognition systems can work around masks, possible authoritarian forces could have more power over their citizens.

Some people are excited about the systems’ potential health benefits, while others worry about its surveillance implications. This technology is still new, so any consequences at this point are just theoretical. The world will have to wait and see what happens.

Author Bio:

Jenna Tsui is a technology journalist with writing experience in future & disruptive technologies, AI, medtech, and scientific development. To see more of Jenna’s work, visit  The Byte Beat  , follow her on  Twitter  or check her out on LinkedIn.

 Image credit