Help:Glossary

This page will give a short overview over the most important terms used in the Wiki.

A

Audio Recognition

• The ability of a software to identify/understand certain sounds. From a technological perspective, software processes audio relatively similarly to how video is processed: rather than feeding an image, a spectrogram is used as input for the software.

B

Behavioural Data

• Behavioural data is the data collected related to the way in which individuals uniquely behave (facial expressions, body movements, voice, etc.).

Biometric Data

• Biometric data is all data related to the body, which can be used to identify or monitor individuals or groups of individuals and is impossible or very difficult to alter (face, fingerprints, iris, etc.).

Biometric Identification

• The use of biometrical or behavioural data to identify a person.

Biometric Mass Surveillance

• Biometric Mass Surveillance is a form of monitoring, tracking, or processing of personal (biometric and behavioural) data of individuals indiscriminately and in a generalised manner without a prior criminal suspicion (FRA, 2019).
• Additionally, this surveillance occurs at a distance, in a public space and in a continuous or ongoing manner by checking them against data stored in a database.

C

Controlled Images

• Controlled images are images that are captured for the purpose of processing, aimed at optimal positions and lighting conditions. They are for example taken at a police station, or at a photographer’s studio with strict requirements, and are either contained in databases that precede the introduction of a facial recognition system (e.g., driver’s license databases) or are specifically designed to match high criteria of biometric systems (i.e., photographs for biometric passports).

Cooperative Searches

• Cooperative searches involves voluntary consent to a search. This is contrary to non-cooperative searches.

D

Datasets

• A dataset is a structured collection and storage of data. For the Wiki, it often involves the structured collection and storage of biometric data, such as fingerprints, facial images etc.

Deployments

• For the Wiki, deployments are all instances in which biometric data is being collected in public spaces.

Detection

• See Facial Detection and Person detection

E

Ear Accoustics

• The unique characteristics of the ear canal are used to identify an indivudal.
• To determine the identity the unique echo sound characteristics of each human are taken into account.
• This technology allows for the authentication of an identity by simply wearing headphone with a microphone.

Emotion recognition

• Software that categorises facial expressions into emotion categories – happiness, sadness, anger, etc. – is known to be used in billboards that are equipped with cameras, in order to analyse audience response to advertisements. For example, in airports or at train stations. While the face is claimed to be a “window into the brain” by some, the technology has been heavily criticised. Firstly, some consider it an undesirable invasion of their privacy, while other critique the technology for capturing primarily stereotypical ways of expressing oneself (van de Ven, 2017). In some places, such as at Dutch train stations, these critiques have led to disabling the cameras in billboards altogether (Het Parool, 2017).

F

Facial Recognition (Identification)

• One-to-many (1:N) searches are called identification searches. An unknown single face, picked up for example from surveillance video footage or from a passport, is run against a large dataset of known faces, in order to identify the unknown face, or to determine if it occurs on a so called “watchlist”. This can be done in the case of forensic investigations or can be deployed in remote biometric identification scenarios in the public space.

Facial Recognition (Verification/Authentication)

• One-to-one (1:1) searches are called verification or authentication searches and are used to determine whether an individual face presented to the camera matches a single face stored in the system. This is how “Face ID” works on iPhones for example. In this example, people volunteer the capture of their face, they are thus considered in a “cooperative” scenario.

Forensic Facial Recognition

• Forensic facial recognition is carried out generally in the context of judicial investigations in order to match photographs of persons of interest captured via surveillance cameras or extracted from documents to an operational database of known individuals (Al-Kawaz et al.2018). This is contrary to live facial recognition.

G

Gait Recognition

• Gait recognition consists of recognising the specific way in which a person walks (gait), but in reality it covers a broader range of criteria (body, proportions, posture, etc.) (Segal 2020,2).
• The advantages of gait recognition are that it does not require a clear access to a face, and it requires a lower image resolution (as it analyses an entire body, not only a face)
• Gait recognition, however, requires more computing power because it works on the basis of moving images (i.e., multiple frames of still images, up to 30 frames per second) rather than still images.

I

Identification

See Facial Recognition (Identification).

Institutions

For the Wiki, institutions are all entities that deploy biometric surveillance technology. For example, companies, labour unions, NGOs, governmental organisations, law enforcement, universities, etc.

L

Live Facial Recognition

• Live facial recognition uses live video feeds in order to generate snapshots of individuals and then match them against a database of known individuals – the “watchlist”. It is the most controversial deployment of facial recognition (Fussey and Murray 2019).

M

Machine Learning

The use and development of software that is able to adapt and change its behaviour automatically based on algorithms and statistical models.

N

Non-Cooperative Searches

• Non-cooperative searches are searches without the intention or consent of the individuals.
• This is the oposite of cooperative searches.

P

People Tracking and Counting

• An object detection algorithm estimates the presence and position of individuals on a camera image. These positions are stored or counted and used for further metrics.
• It is used for example to count passers-by in city centres, and for a one-and-a-half-meter social distancing monitors.
• See also Person Detection

Person detection

• Person detection denotes the ability of a software application to estimate (as in, provide a statistical probability) whether an object in the camera image is a person.
• Generally, it is able to indicate the position of the person in the image.
• Person detection systems can be used in basic analytics scenarios, where for example the presence of people is counted. Moreover, object detection algorithms can be used to track individuals between video frames, although they generally have a hard time tracking occlusions (people walking in front of others, hiding them from the camera) and specific people across multiple camera viewpoints. Person detection does not obtain any information about individuals faces.

Products

• For the Wiki, products are different software solutions or technologies that enable the analysis and input of biometric data.

Public Spaces

• Public spaces are spaces in which a general population has access to.
• Public spaces can be publicly owned (roads, streets, city squares, parking facilities, government facilities) or privately owned (shopping malls, stadiums).

R

Recognition

See Facial Recognition (Identification), Facial Recognition (Verification/Authentication) and Live Facial Recognition.

S

Software

Software is a set of instructions, data or programs used to operate computers and execute specific tasks.

Semi-Supervised Machine Learning

• Semi-supervised machine learning, is machine learning where a small part of the data is labelled.
• Also see Supervised Machine Learning and Unsupervised Machine Learning

Supervised Machine Learning

• Supervised machine learning consists of teaching the system to recognise people, cars,guns, or any other object by feeding it an annotated dataset of such objects.
• It is supervised because humans “supervise” how the computer learns, by annotating the dataset (“this is a car”, “this is a gun” etc.). The categories of the annotations (cars, guns, etc.) will thus be the only ones that the system will be able to recognise.
• Most video surveillance systems use supervised machine learning (IPVM Team 2021a, 11)
• See also Semi-Supervised Machine Learning and Unsupervised Machine Learning.

U

Uncontrolled Images

• Uncontrolled images are images that are captured outside of specific requirement, collected for example through social media scraping or video surveillance.
• This is the oposite of controlled images.

Unsupervised Machine Learning

• Unsupervised machine learning lets the system cluster objects by itself without the input of an human.
• The advantage is the open-endedness of the systems (meaning they can generate categories of objects not anticipated in the training dataset), but the disadvantage is that algorithms can potentially cluster objects along irrelevant criteria for the task (for example clustering red motorcycles, cars, and trucks in one group and green ones in another, as opposed to creating one cluster for all motorcycles, one for cars and one for trucks)
• Also see Supervised Machine Learning and Semi-Supervised Machine Learning