Face recognition uses now reach far beyond the phone in your pocket. Governments, banks, shops and police forces all run systems that turn a face into a set of numbers and compare it with another. Some face recognition applications are quiet conveniences and others are fiercely debated.
This guide walks through the main uses of face recognition and gives one real example of each. It also covers what regulators have ruled and how accuracy gets tested. You will see where the technology helps and where it has been restricted.
Face Recognition Uses at a Glance
Every use of facial recognition answers one of two questions: is this person who they claim to be, or who is this person? The first is a one-to-one check and the second is a one-to-many match.
The graphic below sorts the main face recognition applications by the question they answer. The table after it adds a real example of each.
| Use | Question | Real example | Scale or rule |
|---|---|---|---|
| Phone unlock | Is this you? | Apple Face ID | Apple lists random match odds of about 1 in 1,000,000 |
| Government ID | Is this you? | Aadhaar face authentication | More than 15 crore transactions in March 2025 |
| Online ID login | Is this you? | Singpass Face Verification | Compulsory security measure for all Singpass users |
| Social media | Who is this? | Facebook photo tagging | Shut down in 2021 with over a billion templates deleted |
| Policing | Who is this? | Met Police live cameras in Croydon | 173 arrests in 24 operations and one false alert |
| Retail | Who is this? | Bunnings in Australia | Found in breach of the Privacy Act in 63 stores |
| Travel | Both | TSA, CBP and the EU Entry/Exit System | Opt-out terms differ by system |
How Face Recognition Applications Work
A camera finds a face in the frame. Software measures it and turns the pattern into a template. That template is a string of numbers and not a photo.
In a one-to-one check the template is compared with a single stored one. In a one-to-many match it is compared with a whole gallery. If the score passes a set threshold the system reports a match.
Where that threshold sits is a design choice. A strict setting cuts false matches but turns away more real users. Our guide to face recognition vs fingerprint shows how the same trade-off plays out across methods.
Everyday Uses of Face Recognition: Phones and Online Accounts
Unlocking a phone is the use most people meet every day. Apple says the chance that a random person unlocks a Face ID device is about 1 in 1,000,000. It gives 1 in 50,000 for Touch ID (Apple Face ID security guide).
Apple adds that the odds differ for twins and siblings and for children under 13. It says the face data stays on the device and is not included in backups. Our list of face recognition apps covers the unlock and photo tools you can install yourself.
Social platforms have pulled back. In November 2021 Meta said it would shut down the Facebook face recognition system and delete more than a billion templates. It described the change as a move toward “narrower forms of personal authentication” (Meta).
Face Recognition Applications in Government ID and Banking
India runs one of the largest face recognition applications in public services, and UIDAI reported more than 15 crore face authentication transactions in March 2025. It said more than 100 government and private entities use the system (UIDAI release). The system won the Prime Minister’s Award for Excellence in Public Administration on 21 April 2025.
Singapore uses face checks to protect digital identities. Singpass Face Verification is a compulsory security measure for all Singpass users. It applies at registration and at login and when you reset a password or make a higher-risk transaction (GovTech Singapore).
Banks add the same kind of one-to-one check for sign-up and for cash machines. Our guide to biometric ATMs shows how that works at the counter.
Facial Recognition at Airports and Borders
Travel is the use with the most formal rules. TSA and CBP and the EU Entry/Exit System each run different systems with different opt-out terms. Our airport face recognition guide covers each one and what happens if you decline.
Face Recognition Uses in Policing and Public Safety
Police use face recognition to look for people on a watchlist. London’s Metropolitan Police ran fixed live cameras in Croydon from October 2025 to March 2026. It used static cameras at two sites on the high street and reported 173 arrests across 24 operations (Met Police release, 13 May 2026).
More than 470,000 people walked past the cameras and the force reported one false alert. It said nobody was arrested because of a false alert. Each deployment used a watchlist built no more than 24 hours earlier and deleted straight afterwards.
Critics say scanning everyone on a street to find a few people is disproportionate. Supporters point to the arrests and the low false alert count. Our guide to CCTV face recognition explains how camera networks identify people and where they fall short.
Face Recognition in Shops and Workplaces
Retail shows both the appeal and the risk. Australia’s privacy commissioner found that Bunnings collected sensitive information without consent when it ran face recognition in 63 stores in Victoria and New South Wales between November 2018 and November 2021. The ruling covered everyone who walked in (OAIC determination).
“The most intrusive option.”
OAIC finding on the Bunnings system
Bunnings paused the system during the investigation and kept the right to seek review. Workplaces use the same one-to-one check for doors and time clocks. Rules on employee consent vary from country to country.
How Accurate Are Face Recognition Applications?
NIST tests this independently. Its FRTE 1:N program has covered 713 algorithms from 224 developers since 2018 (NIST FRTE 1:N). The identification test fixes the false positive rate at 0.003 and measures how often the right person is missed.
That design matters. A headline accuracy figure can hide a trade-off. Always ask which error rate was held fixed and on what kind of photos.
NIST also reports differences in false positives between demographic groups. Apple’s note on twins and children shows that accuracy is a set of numbers and not a single score. The field also keeps moving: NIST’s data was last updated on 5 October 2026 and includes 81 algorithms submitted in 2026.
Rules That Limit the Use of Facial Recognition
The EU AI Act bans real-time remote biometric identification in public spaces for law enforcement unless it is strictly necessary. The exceptions cover victims and missing people and imminent threats and suspects of serious crimes. Each use needs prior authorization from a judicial or independent authority and urgent cases have a 24-hour window (Regulation (EU) 2024/1689, Article 5).
The Act also bans building face databases by untargeted scraping of images from the internet or CCTV footage. Australia acted through its privacy regulator in the Bunnings case. The picture changes again country by country and sometimes city by city.
Questions to Ask Before You Agree to a Face Scan
The same technology feels very different depending on the setting. Using facial recognition to unlock your own phone is not the same as being scanned at a shop door. A few questions help you judge any of these face recognition uses.
- Is it optional and is there an alternative? Singpass lets users pick an SMS code instead of a face check at login.
- Were you told before the camera captured you? The Bunnings ruling turned partly on missing notice.
- Who keeps the template and for how long? The Met says it deletes each watchlist straight after a deployment.
- What happens when the system gets it wrong? A low error rate still means some real people get stopped.
None of these questions has one right answer. Your comfort level may change from a bank app to a stadium gate.
FAQ
What are the main uses of face recognition?
The main uses are unlocking devices and confirming identity for government and bank services. Others are boarding and border checks and police watchlists and shop security. The first group asks if you are who you say you are while the second asks who you are.
Can you give examples of facial recognition?
Face ID on an iPhone is the most familiar face recognition example. Aadhaar face authentication and Singpass are national-scale examples. Live police cameras like the Croydon pilot are a more contested one.
Is the use of facial recognition legal?
It depends on where you are and who runs the system. The EU restricts some police uses and Australia’s regulator ruled against Bunnings. Check the rules in your own country before you rely on any of this.
Does face recognition work on twins?
Apple says the odds of a false match differ for twins and siblings. It suggests using a passcode if that worries you. Other systems have their own limits.
What is the difference between face detection and face recognition?
Detection finds a face in an image. Recognition decides whose face it is. Most face recognition applications run both steps.
Questions to Think About
- Would you accept a face scan at a shop door if it cut theft?
- Should police scan everyone on a street to find a few suspects?
- Who should hold your face template: you, a company or the state?
- Is a compulsory face check a safeguard or a requirement you cannot refuse?
Sources
- Apple, Face ID Security Guide
- Meta, Update on Our Use of Face Recognition (2 Nov 2021)
- UIDAI, press release on Aadhaar face authentication (28 Apr 2025)
- GovTech Singapore, Singpass
- Metropolitan Police, live facial recognition pilot results (13 May 2026)
- OAIC, Bunnings breached Australians’ privacy with facial recognition tool (19 Nov 2024)
- NIST, Face Recognition Technology Evaluation (FRTE) 1:N Identification
- EUR-Lex, Regulation (EU) 2024/1689 (AI Act)
