How Does Facial Recognition Work?

posted 5/24/22 by Anna Gdalevich

Facial recognition involves identifying someone through the characteristics of their face. There are many diverse types of facial recognition systems, which may include 3D modeling, heat and vascular patterns, and skin texture analysis. The most common technology utilizes an image database and algorithms in order to create a template of a person’s face, usually by using reference points such as eye color, head shape, and the distance between various facial features. For private access for example, a person may stand in front of a facial scanner, which will then compare their face to the templates of their face stored in a database. If it is a match, the person is granted access. This type of facial recognition is most often used by companies, where employees upload their photos to the security database, for use by the company only. These databases are usually smaller, making it easier for the algorithms to match faces with people.

This technology is used on a more widespread basis, such as by police departments. This usually includes live facial recognition, which includes scanning the faces of a crowd and attempting to match them to police photos of known criminals or missing persons. This use is less common, however, as there are known issues with the quality and accuracy of identifications done on a large number of people and on the racial bias that goes into developing these databases.

Facial recognition technology has many supporters. Some benefits of facial recognition include increase in security, convenience, improved customer service, and advancement in healthcare. Facial recognition can and has been used to identify criminals quickly. It is also used to identify missing children and trafficking victims, identify documents, secure border checks, and to secure company or government buildings. Because of this, it is also more convenient for people to use facial recognition, as they do not need to remember a passcode or bring a keycard in order to access a secure area. It is also a touchless procedure, which is why it is being used to deal with the COVID-19 pandemic. For example, the company CLEAR, which helps make traveling through airport security faster, uses facial recognition to determine whether an employee is sick. Facial recognition has also been used to make important steps in healthcare, such as using it to make accurate diagnosis of disorders.

Facial recognition does have drawbacks, however, and there are many who are concerned about it. For example, there may be false matches when the sample size is too large, such as when trying to identify a criminal. Facial recognition systems often work well with a small amount of face templates, such as in a company. But, the larger the size gets, the more likely it is to match two different people who happen to look somewhat alike. Photos taken of suspects are also normally blurry, in bad lighting, or at a bad angle, making it more difficult to get a conclusive match. This can be dangerous, as the wrong person may be identified as a suspect. Furthermore, facial recognition technology is not as effective at identifying African-American and Asian faces compared to white faces. In a study of Amazon’s Rekognition, a facial recognition technology used by many US police forces, African-American and Asian faces were misidentified between ten and one hundred more than white ones. African-American women and women with darker skin were more likely to be misidentified. This is usually because facial recognition databases contain fewer examples of non-white and non-male people, which presents a major issue for the widespread use of this technology.

There are also many privacy concerns, as many people worry about putting multiple photos of themselves into a database. Data breaches happen frequently, which can lead to multiple people’s personal images being stolen, such as in 2019 when hackers managed to steal 100,000 photos of travelers from US Customs and Border Protection databases. People also worry about the technology being used to track and surveil people without reason and without their knowledge. Multiple US cities, such as San Francisco and Oakland, CA, and Somerville, MA, have banned facial recognition technology. As you can see from the map below, many parts of the US are using facial recognition technology, especially for their police departments, which are represented by the stars.

Sources

"Facial Recognition: How Does It Work in 2021." Recfaces, 27 Oct. 2020, recfaces.com/articles/how-facial-recognition-works

Gips, Michael. "How Facial Recognition Works: Purpose, Use, and Benefits."

Swiftlane, 13 July 2020, www.swiftlane.com/blog/how-facial-recognition-works/

Symanovich, Steve. "What Is Facial Recognition? How Facial Recognition Works." Norton, 20 Aug. 2021, us.norton.com/internetsecurity-iot-how-facial-recognition-software-works.html#

"Facial Recognition Fails on Race, Government Study Says." BBC, 20 Dec. 2019, www.bbc.com/news/technology-50865437

"Ban Facial Recognition." Ban Facial Recognition, www.banfacialrecognition.com/map/