What is facial recognition technology?

• Author: European Union Agency for Fundamental Rights (EU body or agency)
• Pages: 7-8

FRA Focus

7

3. What is facial recognition technology?

Facial recognition technologies are biometric sys-

tems that allow the automatic identif‌ication and

matching of a person’s face. The technology extracts

and further processes biometric data by creating a

‘biometric template’.28 For facial images, a biome-

tric template detects and measures various facial

features.29

Facial recognition

Facial recognition is the “automatic processing

of digital images which contain the faces of indi-

viduals for identif‌ication, authentication/verif‌i-

cation or categorisation of those individuals”.

Source: Article 29 Data Protection Working Party

(2012), Opinion 02/2012 on facial recognition in

online and mobile services, 00727/12/EN, WP 192,

Brussels, 22 March 2012, p. 2

Facial recognition refers to a multitude of technol-

ogies that can perform different tasks for different

purposes. In this regard, a key distinction is whether

facial recognition is used for verif‌ication, identif‌i-

cation or categorisation. Verif‌ication and identif‌i-

cation deal with matching unique characteristics

of individuals to determine their individual iden-

tity. Categorisation deals with deducing whether

an individual belongs to a specif‌ic group based on

his or her biometric characteristics – for example,

sex, age, or race.

In the past few years, facial recognition technolo-

gies have strongly benef‌itted from increased data

availability, computing power and the development

of sophisticated machine learning algorithms.

28 Article 29 Data Protection Working Party (2012), Opinion

3/2012 on developments in biometric technologies, 00720/12/

EN, WP193, Brussels, 27 April 2012.

29 ‘Biometric template’ means a mathematical representation

obtained by feature extraction from biometric data limited to

the characteristics necessary to perform identications and

verications (see Art. 4 (12) of Regulation (EU) 2019/818 of

the European Parliament and of the Council of 20 May 2019

on establishing a framework for interoperability between

EU information systems in the eld of police and judicial

cooperation, asylum and migration and amending Regulations

(EU) 2018/1726, (EU) 2018/1862 and (EU) 2019/816, OJ L 135,

22.5.2019, pp. 85-135).

3.1.

Verif‌ication (one-to-one

comparison)

Verif‌ication or authentication is often referred to as

one-to-one matching. It enables the comparison of

two biometric templates, usually assumed to belong

to the same individual.30 Two biometric templates

are compared to determine if the person shown on

the two images is the same person. Such a procedure

is, for example, used at Automated Border Control

(ABC) gates used for border checks at airports. A

person scans his or her passport image and a live

image is taken on the spot. The facial recognition

technology compares the two facial images and if

the likelihood that the two images show the same

person is above a certain threshold, the identity

is verif‌ied. Verif‌ication does not demand that the

biometric features be deposited in a central data-

base. They may be stored, for example, on a card

or in an identity/travel document of an individual.

3.2.

Identif‌ication (one-to-

many comparison)

Identif‌ication means that the template of a person’s

facial image is compared to many other templates

stored in a database to f‌ind out if his or her image

is stored there. The facial recognition technology

returns a score for each comparison indicating the

likelihood that two images refer to the same person.

Sometimes images are checked against databases,

where it is known that the reference person is in the

database (closed-set identif‌ication), and sometimes,

where this is not known (open-set identif‌ication).

The latter operation would be applied when persons

are checked against watchlists. Using facial recog-

nition technology for identif‌ication is sometimes

referred to as Automated Facial Recognition (AFR).31

Identif‌ication can be used based on facial images

obtained from video cameras. For this purpose, the

system f‌irst needs to detect if there is a face on

the video footage. Smart phone users might know

30 See also Kindt, E. (2013), Privacy and Data Protection Issues of

Biometric Applications A comparative legal analysis (1st edn.

Springer, Governance and Technology Series 12, 2013) and

Iglezakis, I. (2013), EU Data protection legislation and case-law

with regard to biometric application, Aristotle University of

Thessaloniki, 18 June 2013.

31 For example in Davies, B., Innes, M., and Dawson, A. (2018),

An Evaluation of South Wales Police’s use of Automated Facial

Recognition, Cardi University, September 2018.