[Reading Science] Do You Have to Wear a Mask to Open the Door?

AI can verify identity even when wearing a mask. Due to the spread of COVID-19, if you do not wear a mask, a message prompting you to "Please wear a mask" will appear.
[Photo by LG CNS promotional video screenshot]

[Asia Economy Reporter Kim Jong-hwa] What products gained attention due to the spread of the novel coronavirus infection (COVID-19)? Probably masks, other quarantine supplies, and Korean-made COVID-19 diagnostic kits. Excluding these well-known products, there is a technology that has gained technical attention out of necessity.

It is the 'facial recognition' technology. Facial recognition technology is a method that extracts features of a person’s face and verifies identity by comparing them with an existing database (DB).

The process consists of three main stages: 'person capture,' which uses CCTV cameras to check various aspects of a person's face; 'data extraction,' which analyzes each part of the face such as eyes, nose, mouth, and chin to extract unique facial features; and 'identity verification,' which compares the extracted data with the existing DB to decide access permission.

However, the COVID-19 situation introduced a complication called the 'mask.' Since masks completely cover the mouth and nose except for the eyes and forehead, questions arose about whether facial recognition would be impossible.

The technology that instantly dispelled these doubts is L company's AI-based 'facial recognition access control system.' This system installs terminals at access gates to recognize faces and analyzes facial information using AI technology to authenticate identity and control access. When an employee shows their face to the terminal, the system completes facial recognition, information retrieval, identity verification, and gate opening decision all within 0.3 seconds.

Moreover, this system displays a message saying "Please wear a mask" to employees who come to work without a mask or who wear the mask only on the chin or mouth without covering the nose, and it does not open the gate for them. Even if the person wears glasses, heavy makeup, or has different face angles, the system can read the face with an accuracy exceeding 99%.

In a test, even when passing through the access gate while wearing a mask and looking at a smartphone, the system identifies the person and opens the gate. The spread of COVID-19 has ironically become an opportunity to demonstrate the excellence of this technology. AI can distinguish people by deep learning countless facial data, relying only on the appearance around the eyes and nose.

The system also includes a thermal detection function that does not open the gate if the body temperature is 37.3 degrees Celsius or higher. The access gate opens only when the mask is worn properly and the body temperature is normal.

Besides AI directly analyzing the face, there are methods using infrared cameras where thermal cameras read heat points generated by blood flow on the face, and 3D measuring devices that analyze facial features such as the angles and distances of the eyes, mouth, nostrils, chin, and the degree of bone protrusion.

Compared to card tagging and biometric technologies like fingerprint, iris, or vein recognition, facial recognition technology has clearer advantages. It does not require actions like taking out a card or bringing a finger or eye close to a terminal. The face is recognized simply by approaching within 2 meters of the terminal, allowing users to pass through gates naturally without stopping.

Especially with the advancement of CCTV performance, facial recognition is possible from left, right, up, down, and tilted angles, and subjects can be identified from distances of about 15 meters.

Facial recognition technology has long established itself as an essential tool in the fields of public safety and physical security. Major institutions worldwide actively use facial recognition to track criminals or block terrorists. New York City in the United States has installed cameras that recognize drivers’ faces on all bridges and tunnels, and the Chinese government has applied facial recognition technology to CCTV surveillance cameras at train stations.

The postponed '2020 Tokyo Olympics' also plans to apply facial recognition technology for security. When athletes and media personnel enter the stadium, their faces will be checked against the DB for verification.

China has installed facial recognition systems at subway turnstiles to track all movement paths. Fares are also retrieved from the facial database and charged to the individual's account.
[Photo by YouTube screen capture]

Outside the security sector, the financial industry is accelerating the use of this technology. Plans are underway to enable services such as payments and remittances through facial recognition without bankbooks or cards. Major commercial banks in China have already applied facial recognition technology to ATM machines and mobile financial services, and facial recognition systems are installed at subway ticket gates to monitor residents’ movements.

Facial recognition technology is also applied to unmanned delivery systems. It allows easy verification of the orderer’s identity, enabling fast and accurate product delivery without direct contact between the delivery company and the customer. Additionally, it can be used for identity verification of hotel reservation holders and guests, examinee verification in large-scale tests, library borrower identification, finding missing children, and locating elderly dementia patients.

According to the Korea Institute of Science and Technology Information (KISTI), the global facial recognition technology market size is expected to reach $2.8362 billion (about 3.2 trillion KRW) this year. It has been growing at an average annual rate of 13.3% over the past five years.

However, institutional measures to regulate misuse such as 'deepfake' technology must be established first. More thorough management is needed to prevent personal information leaks and privacy violations related to employee DBs. There is an urgent need for institutional supplements that strongly punish violations.

In the Big Brother era, is personal surveillance and control inevitable? China’s case is representative. When will technology be developed that raises human morality alongside scientific advancement?