Face Recognition and AI Glossary
The Ayonix glossary defines the face recognition, biometrics and AI agent terms that appear in enterprise procurement documents and security standards. Each entry opens with a definition that stands on its own, then explains how the concept is measured, where it is used, and which international standard governs it where one exists.
Biometrics and face recognition terms
- Face recognitionFace recognition is a biometric technology that identifies or verifies a person by measuring the geometry of their face and comparing it against stored templates.
- Liveness detectionLiveness detection determines whether a biometric sample comes from a live person present at the sensor, rather than a photograph, printed mask, screen replay or deepfake.
- Presentation attack detectionPresentation attack detection is the set of techniques a biometric system uses to recognise a fake sample presented to its sensor: a printed photo, a screen replay, a silicone mask.
- Face verification vs identificationFace verification is a 1:1 comparison that confirms a person matches one claimed identity.
- False Accept Rate and False Reject RateFalse Accept Rate is how often a biometric system matches two different people; False Reject Rate is how often it fails to match the same person twice.
- NIST FRVT (Face Recognition Vendor Test)NIST FRVT is the Face Recognition Vendor Test, an ongoing evaluation run by the US National Institute of Standards and Technology that measures submitted face recognition algorithms on sequestered datasets.
- Watchlist matchingWatchlist matching compares every face a camera sees against a defined list of enrolled people and raises an alert only on a candidate match.
- Biometric access controlBiometric access control grants or refuses entry to a door, gate or turnstile using a physical trait such as a face or fingerprint instead of a card or PIN.
AI agent, deployment and vision terms
- Edge AI (on-device inference)Edge AI runs inference on the device where data is captured, such as a camera or an on-premise appliance, instead of sending it to a remote service.
- Air-gapped deploymentAn air-gapped deployment runs software on hardware with no connection to any external network: no model updates fetched at runtime, no telemetry sent, no cloud inference call.
- AI agentAn AI agent is a program that takes a goal, plans the steps to reach it, calls tools or APIs to carry them out, and checks its own progress before reporting back.
- Agentic AIAgentic AI describes systems that pursue a goal over multiple steps, choosing which tools to call and when to stop, rather than returning a single response to a single prompt.
- Retrieval-augmented generation (RAG)Retrieval-augmented generation answers a question by first searching a document collection, then giving the retrieved passages to a language model as the material for its reply.
- Video analyticsVideo analytics is software that interprets camera streams automatically, turning frames into events such as a person entering a zone, a vehicle stopping, or a crowd forming.
- Computer visionComputer vision is the field concerned with extracting meaning from images and video: where objects are, what they are, and how they move.
Why these definitions are worded carefully
Biometrics procurement runs on terms that sound interchangeable and are not. Verification is not identification, detection is not recognition, and a vendor describing a standard as a certification is describing something that does not exist. Each entry here states what a term means, and just as often what it does not.
No accuracy figures appear in this glossary. An error rate without its dataset, threshold and demographic breakdown is not a fact a buyer can act on, and Ayonix does not publish accuracy benchmarks.
Frequently asked questions
- What is the difference between face verification and face identification?
- Verification is a 1:1 comparison confirming that a face matches one claimed identity, such as a passport photo at a gate. Identification is a 1:N search against a gallery of enrolled people to determine who someone is. Identification is computationally harder, more tightly regulated, and restricted or unavailable on several cloud face services.
- What does FAR and FRR mean in biometrics?
- False Accept Rate is how often a system matches two different people. False Reject Rate is how often it fails to match the same person. They trade against each other through the matching threshold: tightening the threshold lowers false accepts and raises false rejects, so there is no single correct setting independent of what an error costs.
- Can a company be NIST certified for face recognition?
- No. NIST runs evaluations and publishes reports on how submitted algorithms performed; it does not endorse vendors, issue certificates or maintain an approved list. A NIST certified claim describes something that does not exist, so treat it as a reason to ask which algorithm was submitted and when.
- Is ISO/IEC 30107-3 a biometric certification?
- No. It is an international standard defining how presentation attack detection is tested and how results are reported, so vendor claims can be compared. Compliance describes the methodology used rather than a pass mark awarded, so ask which attack instruments were tested, at what level, and by whom.
- What is an air-gapped AI deployment?
- An air-gapped deployment runs entirely on hardware with no outbound network connection: no model updates fetched at runtime, no telemetry sent, no cloud inference call. It is required in defence facilities, some central banks and any environment where law or policy forbids the data leaving the premises.
- Does this glossary cover AI agent terminology as well as biometrics?
- Yes. Ayonix builds both face recognition and the ATLAS AI agent platform, and enterprise buyers increasingly evaluate them together for the same site. The glossary therefore defines agentic AI, retrieval-augmented generation and edge AI alongside the biometric terms.