When it comes to facial recognition, there is good news and bad news.
The bad news is lawmakers want to ban the technology. The good news is experts believe the bans won’t stand the test of time.
California is at the center of this controversy as lawmakers consider the first state-wide ban of face recognition technology.
Assembly Bill 1215, known as the Body Camera Accountability Act, proposes banning facial recognition software in police body worn cameras (BWCs). The proposed bill is expected to set in motion a federal-scale ban of the biometric.
The law follows the City of San Francisco’s blanket ban against the biometric. The ruling bars any San Francisco government entity from using face recognition. Somerville, Mass. has followed suit, while other cities look to do the same.
Axon became the first company to ban the biometric in June. Axon’s independent ethics board determined face recognition is unreliable for BWC use. The board reports greater accuracy is needed. In other words, it must perform equally well across races, ethnicities, genders and other identity groups.
Matt Parnofiello, public safety development manager for CDW Government, is watching the face recognition debate play out. He believes the result of this legislation will be better and more accurate technology.
“There are a trio of factors — technology, the law and social acceptance — in play here, and currently they are at odds with each other,” he says. “But we have plenty of examples of once controversial technologies in use today. These laws will not stand forever,” he says.
Biometric Bans and Business
Though there are concerns surrounding biometric security and privacy, businesses can still employ biometrics in their security systems. The bans center on government use of the technology.
Businesses commonly use fingerprint, facial recognition, iris scans and palm scans in their biometric security systems.
Third-party biometric authentication technologies integrate into surveillance camera systems and access control systems. In addition, some surveillance cameras and door access control systems have biometrics built in.
Biometric security technology provides accurately identifies individuals, lowering a business’s risk of unwanted breaches. Access is granted by unique biological characteristics, such as iris scans, fingerprints or faces. These biometrics are difficult to duplicate or forge.
Facial recognition security cameras can memorize the faces of persons of interest, networks of gang members, wanted criminals and suspects in crimes. The tool alerts business owners when unwelcome individuals arrive on their property.
The advanced technology also can recognize and match an individual seen on a CCTV camera at a crime scene to someone the police encounters later.
As facial recognition technology advances, an increasing number of stores are making it a key part of their fraud and theft prevention efforts. In fact, the facial recognition market is expected to grow to $7.7 billion in 2022 from $4 billion in 2018.
Retail centers leverage facial recognition technology to reduce fraud and theft. Companies upload photos of people they want to watch, such as known shoplifters, disgruntled employees or other persons of interest, into the system. The system then watches for those individuals in the store.
Stores using biometric security report up to a 34% reduction in theft, according to FaceFirst, a firm offering retail surveillance tools.
Theft in retail also occurs at the checkout counter as cashiers fail to scan products for friends and family. Systems, such as StopLift, integrate with security cameras and leverage artificial intelligence to send alerts as this occurs.
StopLift reports clients see up to 40% reductions in inventory loss.
How Facial Recognition Works
Amid concerns over privacy and accuracy, it’s important to understand how facial recognition works.
Facial recognition relies on several technologies to work: An image capture system (camera or video surveillance), artificial intelligence and machine learning.
Facial recognition maps facial features from a photo or video and converts them into digital biometric data. It compares this digital signature to a database of known faces to find a match.
There are four basic steps to facial recognition:
- The system captures a picture of your face as you walk by. This can be a video image or a photograph.
- Facial recognition software reads the geometry of your face. It looks at things such as the distance between your eyes, forehead to chin height, and facial landmarks to develop a digital signature of your face.
- Your facial signature, a mathematical formula of ones and zeros unique to you, is then compared to a database of known faces.
- The system determines your identity.
Look for NIST Certification
There are concerns, however.
One concern is that this technology can be slow. This is a key factor if you’re using it to get groups of employees in the door at the beginning of a shift.
Selecting a biometric system certified by the National Institute for Science and Technology (NIST) can help ensure it performs consistently.
NIST certification tests biometrics for the following:
- Speed of enrollment. In other words, how long and how many tries does it take to generate a template that consistently matches a person’s live biometric?
- 1:N Match Speed. N is the number of stored templates. This test compares a biometric against a database of several thousand individual users to define how long it takes to get an accurate match. This process should take less than 2 seconds in biometric access control.
- 1:1 Match Speed: This is the time it takes for the software to match a presented template to a single stored template. This test considers how long it takes for two-factor identification, where an employee presents an identification card and a biometric to verify their identity. This match time should be a fraction of a second.
Concerns over accuracy are ongoing, but that too is improving.
Controversy grew after a Massachusetts Institute of Technology (MIT) study found face recognition technology fails on faces of women and people of color.
Other studies revealed similar issues. A 2018 ACLU study found Amazon Rekognition falsely matched 28 members of Congress to mugshot photos. The study found these failures impacted faces of color most of all.
However, NIST tests the effectiveness of facial recognition annually in its Ongoing Face Recognition Vendor Test. Its most recent report, released in April, lists the most effective facial recognition systems available. The NIST report can be found here.
NIST evaluated 127 software algorithms from 39 different developers. It found that between 2014 and 2018, facial recognition software became 20 times better at searching a database to find a matching photograph.
As lawmakers propose bans and limits on the technology, it’s important to re-emphasize: The bans and proposed legislation do not impact personal, business or federal government use of facial recognition technology.
In some areas, your business may be required to post signs indicating facial recognition use. However, the law does not prohibit its use to improve access control, heighten overall security and reduce theft and fraud.
As lead content writer of Safe and Sound Security, Ronnie Wendt, provides articles on the latest security industry news. Safe and Sound Security is the fastest growing integrator servicing the California Bay Area and Los Angeles area. We specialize in burglar alarm, access control and security camera installation.