Skip to content

Embedded face recognition · ATLAS AIBOX

Recognize locally.
Respond instantly.

Turn existing IP cameras into a private face-recognition system for entry, security and customer check-in. Video and biometric templates stay inside your premises — even when the internet is unavailable.

Plan my deploymentSee the 3 top use cases
  • 100% on premises
  • Works offline
  • Uses existing cameras
A recognized visitor approaching a glass speed gate under a ceiling IP camera, with the ATLAS AIBOX appliance mounted in a rack alongside
One box. Complete local flow.Edge engine active
  • 6+ TOPSNPU inference
  • 4 TBlocal NVMe
  • 2.5GbE + 1GbEdual network

Best fit when cloud processing is too slow, restricted or unnecessary.

HospitalitySmile · pay · access

SecurityRetail · campuses · property

Check-inDental · hotels · gyms · cafés

The three deployments customers ask for most

Start with the outcome — not the technology.

Each deployment uses the same local recognition engine. What changes is who is enrolled, where the camera is placed and what should happen after a match.

AIBOX brings smile-enabled hospitality, proactive security and private check-in onto one locally managed platform.

A guest at a hotel café counter confirming an opted-in transaction with a smile, recognized by a small counter camera

Hotels · cafés · VIP lounges · entertainment

Smile detection in hospitality

Recognize an opted-in guest locally, then use a deliberate smile as the final confirmation to pay, enter a VIP area or redeem a benefit — without a card or cloud round trip.

  1. Counter or entrance camera recognizes the enrolled guest.
  2. AIBOX detects an intentional smile as confirmation.
  3. POS or access system completes the approved action.

Immediate value: faster service, memorable guest experience and card-free confirmation under local control.

A retail entrance where one person among the arriving shoppers is flagged against a watchlist on a guard's tablet

Retail · warehouses · campuses · residences

Watchlist alerts & property security

Monitor selected entrances with existing CCTV. When a person matches an authorized internal watchlist, security receives an immediate event and evidence snapshot.

  1. Live entrance video is analyzed on site.
  2. Faces are checked against the approved list.
  3. VMS, tablet or control room receives the alert.

Immediate value: proactive response for repeat threats without continuously uploading CCTV to the cloud.

A returning patient checking in at a dental clinic reception, recognized by a camera on the desk terminal

Dental · medical · hotels · gyms · cafés

Private, consent-based check-in

Recognize customers who have opted in, then open the correct appointment, membership or loyalty profile for staff — without putting biometric data in a public cloud.

  1. Customer enrolls with clear consent.
  2. AIBOX recognizes the returning customer locally.
  3. Staff sees the correct check-in or service prompt.

Immediate value: faster reception, more personal service and stronger control of sensitive customer data.

Choose the deployment in 30 seconds

The three ATLAS AIBOX face-recognition deployments compared by who is enrolled, where the camera sits, what happens after a match and the best site to start with.
Use caseWho is enrolled?Camera positionAction after matchBest first site
Hospitality smileGuests and members who explicitly opt inPOS counter, VIP door or service kioskConfirm payment, access or benefitOne café counter or VIP entrance
Watchlist securityLawfully managed persons of interestPublic-facing entrances and approach zonesAlert, bookmark video, notify operatorOne high-risk entrance
Private check-inCustomers or patients who opt inReception, counter or self-service kioskOpen appointment, profile or loyalty recordOne reception desk

From camera to action

Everything happens inside your network.

ATLAS AIBOX sits beside your cameras and business systems. It converts live video into a controlled identity event; only the action and evidence you choose need to leave the box.

  1. Connect

    Add ONVIF or RTSP camera streams. Keep the camera system you already own.

  2. Enroll

    Create local face templates for authorized staff, guests or a lawful watchlist.

  3. Recognize

    Detect, track and match faces on the embedded processor without a cloud round trip.

  4. Act

    Send a VMS event, webhook or access command to the system that should respond.

Why on-premise recognition

Fast enough to act. Private enough to trust.

Edge processing is the practical choice when a door, alert or check-in cannot wait for a cloud connection — and when face data should remain under the customer's direct control.

Deployment-ready foundation

  • Face detection and tracking
  • 1:1 and 1:N matching
  • Local watchlist management
  • On-site biometric template storage
  • ONVIF / RTSP camera input
  • VMS / API / webhook events
  • Offline, air-gapped operation
  • Local user and policy management

Local data control

Keep face templates and video evidence inside the site, under the customer's retention and access policies.

Low-latency response

Make access and alert decisions where the camera is, without waiting for an external service.

Network resilience

Continue recognition during internet outages or deploy on networks with no outbound route.

Lower video traffic

Analyze streams locally and send only selected events instead of continuously transferring raw video.

Existing infrastructure

Add intelligence to compatible IP cameras and connect it to the access or video system already in place.

Clear operational scope

Define who may be enrolled, which cameras match and exactly what action each result can trigger.

Start with one site

Bring us your entrance, camera and outcome. We'll map the deployment.

Show us one entrance photo or camera view and tell Ayonix whether you need smile confirmation, a security alert or customer check-in. We will map camera placement, the local recognition workflow and the right ATLAS AIBOX configuration.

Request a site assessmentEmail infojp@ayonix.com

Responsible deployment: Face templates are sensitive biometric data. Every installation should have a defined lawful purpose, clear enrollment or watchlist rules, appropriate notice or consent, controlled access, retention limits and human review for consequential alerts.

An installed IP camera above an office door, wired to an ATLAS AIBOX and door controller in a wall cabinet, with the recognized face shown on a panel display
IP CAMERAATLAS AIBOXACCESS ACTION

What is on-premise face recognition on the Ayonix AIBOX?

The Ayonix AIBOX runs Ayonix Face Recognition entirely on-premise or air-gapped, so images never leave the site. It is the appliance form of the ATLAS BOX edge server, suited to SDK and OEM integration where data residency and fully offline operation matter.

Evidence & sources

Related capabilities

In the same section