{"slug":"facial-recognition","name":"Facial Recognition","summary":"Facial recognition matches a photo of a face to an identity.","order":1,"published":true,"references":[{"label":"NIST Face Recognition Vendor Test (FRVT)","url":"https://www.nist.gov/programs-projects/face-recognition-vendor-test-frvt"},{"label":"ISO/IEC 19795-1: Biometric performance testing and reporting","url":"https://www.iso.org/standard/41446.html"},{"label":"ICAO Doc 9303: Machine Readable Travel Documents (Face images)","url":"https://www.icao.int/publications/pages/publication.aspx?docnum=9303"}],"createdAt":"2025-05-07T00:00:00.000Z","updatedAt":"2025-11-10T00:00:00.000Z","updates":[],"faq":[{"question":"How does facial recognition work?","answer":"Facial recognition uses algorithms to analyze facial features—such as the distance between eyes or the shape of cheekbones—to create a unique digital signature for identification."},{"question":"How accurate is facial recognition technology?","answer":"Modern facial recognition systems can achieve very low false match rates (below 0.1%) under controlled conditions, but accuracy may vary with lighting, pose, and image quality."},{"question":"What are common concerns around facial recognition?","answer":"Concerns include privacy violations, potential bias against certain demographic groups, and misuse by surveillance systems without consent."},{"question":"What is Presentation Attack Detection (PAD)?","answer":"PAD aims to detect spoofing attempts (e.g., photos, masks, deepfakes) and is evaluated separately from matching accuracy using standards-driven tests."},{"question":"What are 1:1 vs 1:N operations?","answer":"1:1 verifies a claimed identity (authentication), while 1:N searches across a gallery (identification). They have different accuracy and scalability considerations."}],"title":"Facial Recognition","category":"Technology","license":"CC-BY-4.0","_meta":{"dataset":"technologies","schemaVersion":"1.0.0","license":"CC-BY-4.0","sourceType":"markdown","sourcePath":"content/technologies/face-recognition.md","canonicalPath":"/technologies/facial-recognition","apiPath":"/api/technologies/facial-recognition"},"description":"Facial recognition is a biometric modality that analyses the geometry and texture of a person’s face to establish identity. Modern systems rely on CNN embeddings and achieve <0.1% FNMR on NIST FRVT benchmarks.\n\n### Latest Updates\n* **2025-07-01:** Tinder, Deliveroo, Uber Eats and Just Eat expanded selfie or facial-verification checks to fight bots, deepfakes and illegal-account rentals in CA and UK gig-economy platforms."}