Glossary · Technology
Face recognition
Face recognition is the family of techniques that compare faces to decide whether they belong to the same person. It covers verification (is this the person they claim to be), identification (who in this database is this), and search (where else does this face appear). All three run on the same pipeline: detect the face, compute an embedding, compare embeddings.
Updated November 6, 2026 · 1 min read
What are the three modes?
| Mode | Question | Example |
|---|---|---|
| Verification (1:1) | Is this the claimed person? | Phone unlock, passport gate |
| Identification (1:N) | Who in the gallery is this? | Police watchlist, access control |
| Search (1:N over the web) | Where else does this face appear? | PimEyes, FaceCheck.ID, FaceSearch |
Verification compares one probe with one enrolled template. Identification and search compare one probe with many, which is one-to-many matching, and the larger the gallery the more look-alikes appear. A web search engine differs from a police system in one respect that matters legally: it returns pages, not names, and makes no claim about identity.
How does the pipeline work?
- Face detection finds the face and its landmarks.
- Alignment rotates and scales the face to a standard pose.
- A network computes a face embedding.
- The embedding is compared with others to produce a similarity score.
- A threshold or a human reviewer decides what counts as a match.
How accurate is it?
On high-quality frontal photos the best algorithms tested by NIST make very few errors. Accuracy falls with low resolution, pose, age gaps, and across some demographic groups, as the 2019 NIST demographic report documented. Independent accuracy figures exist only for vendors who submit to testing; consumer face search engines generally do not, so their accuracy claims are unverified. See accuracy, bias and testing.
How is it regulated?
As biometric processing. The GDPR and the UK GDPR treat data used to uniquely identify a person as a special category needing a specific legal basis; Illinois BIPA requires written consent before collecting a scan of face geometry; the EU AI Act bans untargeted scraping of facial images to build databases. The laws index covers each jurisdiction.
Questions
Is face recognition the same as face detection?
No. Detection finds a face in an image. Recognition compares that face with others.
Does a face search engine identify people?
It returns public pages where a similar face appears. Any name comes from the page, not from the engine.
Sources
- 1.NIST, Face Technology Evaluations (FRTE/FRVT) (checked January 15, 2026)
- 2.NISTIR 8280, FRVT Part 3: Demographic Effects (checked December 19, 2019)
Entry reviewed November 6, 2026. Found an error? Tell us.
Related entries
- Term Face detection Face detection defined: locating faces in an image with boxes and landmarks, how it differs from recognition, common detectors, and the failures that limit a search.
- Term Face embedding Face embedding defined: the numeric vector a network computes from a face, why same-person vectors sit close, how it is compared, and why it is biometric data.
- Term One-to-many matching (1:N) One-to-many (1:N) face matching defined: a probe face against a gallery of many, how it differs from 1:1 verification, and why errors grow with gallery size.
- Law General Data Protection Regulation (GDPR) and biometric data GDPR and face search: Article 9 biometric data, the household exemption, territorial reach under Article 3(2), fines up to 4%, and the Clearview and PimEyes cases.
- Concept How a face search works The face search pipeline step by step: detection, alignment, embedding, index lookup, ranking and source pages, with the limits at each stage.
Cite this entry
FaceSearch Wiki editors. "Face recognition". FaceSearch Wiki, updated November 6, 2026. https://facesearch.wiki/glossary/face-recognition/