Facial recognition is a technology-based method of identifying a human face. Such a recognition system maps facial characteristics from an image or video using biometrics. To identify a match, it compares the information gained to a database of known faces. Facial recognition may aid in the verification of a person's identification, but it also presents privacy concerns.
The facial recognition industry is predicted to expand from $4 billion in 2017 to $7.7 billion in 2022. This is due to the fact that such technology holds several business uses including monitoring and marketing.
But here's where things become difficult. If you value your privacy, you undoubtedly want some say over how your personal information (your data) is utilised. The truth is, your "faceprint" is your personal information.
How does facial recognition work?
You might be adept at identifying people's faces. You probably have no trouble recognizing the face of a family member, friend, or acquaintance. You recognize their facial characteristics — their eyes, nose, and mouth and their facial movements.
That is exactly how a face recognition system operates but on a much larger, computational scale. Recognition technology sees data where you see a face. That information may be saved and retrieved. According to Georgetown University research, half of all American adults have their photos recorded in one or more facial-recognition databases that law enforcement authorities may consult should they wish to.
So, how does facial recognition really work? Although certain technologies differ, most follow a standard procedure:
• A photograph or video of your face is obtained. Your face might be scanned alone or in a crowd. Your photo might show you gazing straight ahead or almost in a profile view.
• The geometry of your face is scanned by facial recognition software. The distance between your eyes and the distance from your forehead to your chin are important considerations. The program recognizes facial landmarks — one system even recognizes 68 of them – which are all important in differentiating your face. As a consequence, your facial signature is created.
• A database of known faces is matched to your facial signature, which is a mathematical formula. Consider the following: At least 117 million people in the United States have photos of their faces in one or more police databases. The FBI has access to 412 million of such pictures for searches, according to a May 2018 report.
• A decision is made. Your faceprint could match one in a database bringing back a positive result.
How effective is facial recognition?
Experts are concerned that face recognition might result in incorrect identifications. What if a police agency wrongly identifies someone smashing a shop window during a riot as someone who was nowhere near the incident using facial recognition technology? How probable is it that such an incident will occur?
It depends. According to the National Institute of Standards and Technology tests, the top face recognition algorithm has an error rate of under 0.08% as of April 2020. This is a significant improvement from 2014 when the best algorithm on the market had an error rate of 4.1%.
According to a 2020 report by the Centre for Strategic & International Studies (CSI), accuracy is greater when identification algorithms are used to match persons to clear, static photos, such as passport photos and mugshots. When applied in this manner, face recognition algorithms achieved up to 99.97% accuracy on the National Institute of Standards and Technology's Facial Recognition Vendor Test.
In practice, however, accuracy rates are often lower. According to the CSI report, the Facial Recognition Vendor Test discovered that the mistake rate for one algorithm increased from 0.1% when faces were matched to high-quality mugshots to 9.3% when faces were matched against images of people caught in public. When individuals were not looking straight at the camera or were partly concealed by shadows or objects, error rates increased.
Another issue is ageing. According to the Facial Recognition Vendor Test, middle-tier facial recognition algorithms exhibited mistake rates that increased by roughly a factor of ten when attempting to match photographs of participants shot 18 years earlier.
Who employs facial recognition?
Many individuals and organisations utilise face recognition in a variety of settings. Here are a few examples:
In airports, facial recognition technologies can monitor persons entering and exiting. The technology has been utilised by the Department of Homeland Security to identify persons who have overstayed their visas or are under criminal investigation.
Product manufacturers of mobile phones
Apple originally employed facial recognition to unlock the iPhone X, and since, the technology has been carried over to all subsequent models. Face ID authenticates — it ensures that you are who you say you are when you access your phone. According to Apple, the likelihood of a random face unlocking your phone is one in one million.
Websites for social networking businesses
When you post a picture to Facebook, an algorithm is used to detect faces. If you wish to tag others in your images, the social media firm will ask you. If you answer yes, a connection to their profiles is created. Facial recognition on Facebook is 98 percent accurate.
Entrance businesses and restricted zones
Some businesses have abandoned security badges in favour of facial recognition technologies.
Religious congregations at places of worship
Face recognition has been used by churches to scan their congregations to see who is there. It's a fantastic method to keep track of regulars and irregulars, as well as to adapt contribution requests.
Campaign marketers and advertisers
When targeting groups for a product or concept, marketers often consider factors such as gender, age, and ethnicity. Even during a performance, facial recognition may be used to determine such audiences.
The use of facial recognition in police enforcement
Today, facial recognition databases play an important role in law enforcement. According to an Electronic Frontier Foundation investigation, law enforcement agencies frequently collect mugshots from jailed people and compare them to local, state, and federal face recognition databases.
Law enforcement organisations may use these mugshot databases to identify persons in images collected from a number of sources, including closed-circuit television cameras, traffic cameras, social media, and photos taken by police officers themselves.
According to the Electronic Frontier Foundation, police officers may also use their mobile phones, tablets, or other devices to take images of cars or pedestrians and instantaneously match their photos to the faces in one or more facial recognition databases.
In addition, police enforcement has utilised face recognition to identify persons who may be sought in connection with crimes at huge events such as concerts, sports events, or the Olympics.
Several face recognition technologies are available to the federal authorities. Its primary database, however, is the FBI's Next Generation Identification system. This collection comprises over 30 million images.
Opponents of face recognition systems argue that although they give some protection, it is not enough to outweigh a feeling of independence and freedom. Many people believe that the usage of these technologies violates their privacy, but their worries don't stop there. They also emphasise the dangers of identity theft. Even face recognition companies recognize that as the technology becomes more widely used, the probability of identity theft or fraud increases.
As with many emerging technologies, the enormous promise of facial recognition has its downsides, but manufacturers are working to improve the usability and accuracy of their systems every day.