Facial recognition technology has been hailed as a new crime fighting tool. But it has come under fire – many people are questioning its accuracy and its impact on privacy, while lawmakers are concerned it tampers with law enforcement efforts.
Shaun Moore, co-founder and CEO of biometrics firm Trueface, believes facial recognition is a single component of a larger advancement in technology. “As our identities continue to digitise, we need methods and solutions to protect them and authenticate them.”
He says passwords are simply too easy to hack, so biometrics have been the necessary solution for the last ten years. “I am not suggesting that face recognition as a single input is the ‘be all and end all” solution, but pairing it with social information or behavioural information certainly makes it more difficult to comprise an identity.”
According to Moore, the debate that society is having is purely about a surveillance state and begs the question of whether or not the government should have the tools to ‘monitor’ everyone in real-time.
“The fact is, this is simply not how the technology is being deployed and the government isn’t watching every street block scanning faces,” he explains. “This technology is meant to aid the already very difficult job of law enforcement in identifying someone who has already committed a crime. Instead of manually scanning through days of footage, face recognition can provide another source of information or point of reference. It is not built to ‘be the decision’ it is to be used as a tool in a toolbox.”
How it works
Moore says facial recognition isn’t as basic as taking two pictures and seeing if they match.
“Facial recognition algorithms create a mathematical representation of a human face called a face template by identifying landmarks on the face such as the nose and eyes and calculating the distance between those landmarks. It is, at its basic form, computing geometry. These equations represent the face which is then compared to other mathematical representations to find a match or a similarity score.”
According to him, every person has their own equation or mathematical representation which is unique to them but also uniquely ingested by the algorithm. “It is not a one-size-fits-all approach, as templates cannot be transferred across companies or algorithms, meaning you cannot take a face template from one company and use it to positively identify that person using a different company’s software.”
“Pairing [facial recognition] with social information or behavioural information certainly makes it more difficult to comprise an identity.”Shaun Moore, co-founder and CEO of Trueface
He says there are a variety of ways facial recognition is being used today.
Firstly, personal access control, such as using facial recognition to unlock your phone, which is referred to as ‘one to one’ matching.
Next, there’s public access control, which you’ll find in services such as customs and border control, access to sports arenas and suchlike, which is referred to as ‘one to few’.
“Finally, facial recognition is used for public monitoring, such as law enforcement forensically analysing crime footage, which is called ‘one to N’ or ‘one to many’.”
Facial recognition allows for more efficient flow of people as well as more security for the individuals inside buildings, and is being used to authorise individuals setting up bank accounts and making mobile transactions, as well as expedite the boarding process on planes, the embarking process on cruise ships, and to identify missing children.
Moore says there has been some adoption of facial recognition technology across police departments as a forensic tool, but he is unaware of any department using face recognition in real time. “In addition, we are working with companies to provide face recognition as a credential for access control to corporate offices and construction sites.”