Facial Recognition at the Edge

Watchlist matching and access control on the AI Edge Analytic Box with consent-first controls. Face templates stay in the tenant. Customer video is not used to train models. Professional or Enterprise for production FR deployments.

Facial Recognition

Watchlist matching with consent-first controls

Access-control and security teams need watchlist matching without sending biometrics to a shared cloud gallery. The AI Edge Analytic Box matches faces to a gallery you enroll; templates stay in your tenant. Facial recognition is restricted in some jurisdictions and is enabled only after you document a lawful basis. Production FR typically uses Professional or an Enterprise on-prem deployment.

Recommended hardware: Professional or Enterprise. Review privacy controls below before enabling FR modules.

Lobby · Entrance 1 · Watchlist match
Office lobby camera used for consent-first facial recognition
Match · J. Chen · 0.97
Gallery: staff-allow
ACCESS GRANTED
Edge FR · consent policy v3 · event logged

Enterprise-grade matching

  • 1:N watchlist match at facility gates
  • Cross-camera re-ID within your sites
  • Access-control webhook on match / no-match
  • Liveness-friendly enrollment workflow
  • Audit log of every match event

Requires an AI Edge Analytic Box Enterprise on-prem deployment for concurrent FR plus other analytics streams.

Privacy & compliance

Facial recognition involves biometric data in many jurisdictions. AtomVision ships with deployment playbooks — signage templates, retention policies, and legal review checkpoints — before FR modules are enabled on any camera.

Lawful basis & consent workflow documentation
On-premise / tenant-isolated face galleries
Configurable retention & auto-purge
No training on customer footage
GDPR-aligned DPIA templates
Regional enablement controls per site

Regional notice: Laws vary by country and municipality. Your legal counsel should approve camera placement, signage, and retention before production use. AtomVision can disable FR per site while other Video AI modules remain active.

Security & compliance documentation →

FAQ

What facial recognition capabilities does AtomVision offer?

Enterprise deployments include 1:N watchlist matching against your enrolled gallery, access-control integrations, and cross-camera re-identification within your estate. Face detection (presence only) is available on all tiers; full identification requires an Enterprise on-prem deployment and a signed biometric processing agreement.

How do you handle privacy and consent?

We require documented lawful basis before enabling FR — typically explicit consent for employees/visitors or legitimate interest with signage for access control. Face templates stay in your tenant; we do not train global models on customer video. Retention defaults to 30 days for event metadata unless you configure shorter windows. EU deployments support data residency and DPIA documentation.

Is customer video used to train your models?

No. Embedding models are trained on licensed datasets. Your gallery vectors and match events remain in your isolated environment. We provide DPAs and subprocessors lists for procurement.

Are there regions where facial recognition is restricted?

Yes. Some jurisdictions limit biometric identification in public spaces. We work with your legal team to scope camera placement, signage, and retention. FR modules can be disabled per site while other analytics continue.