Product

One platform for every camera you already own.

Watch one reception camera go from connected to counting to alerting, in about twenty seconds. No new hardware, no code.

Talk to us

Camera ManagerOperations LeadOL

Global searchStatus · AllAdd camera group+ Add camera

HQ · Ground floor6 cameras · all healthy

LIVEAdded just now

CAM 22 · Reception

PeopleHealthy

Analytics

+Add a camera

LIVE

CAM 23 · Entrance corridor

PeopleHealthy

Analytics

LIVE

CAM 12 · Front desk

PeopleHealthy

Analytics

LIVE

CAM 31 · Meeting room 2

OccupancyHealthy

Analytics

LIVE

CAM 02 · Vehicle gate

VehiclesHealthy

Analytics

LIVE

CAM 18 · Visitor car park

VehiclesHealthy

Analytics

LIVE

Zone settingsZoneReception floorMeasurePeople in zoneAlert whenmore than 8 peopleSave zoneZone saved · counting

Reception · people in zoneLive

In zone now3

Peak today1409:40

Average stay2:10

08:0008:1508:3008:45

Playback · CAM 22 Reception · 08:41:12

12/05/202608:41:12⏮▶⏭1x4x

Events

Over capacity · 9 people08:41:12 · alert sent

Back under 808:44:50

Reviewed by Operations Lead08:45:03

08:41Tuesday 12 May

nowReception over capacity9 people in the zone · alert above 8

ViewDismiss

Sample footage (Intel sample videos, CC BY 4.0; OpenCV sample data; Mixkit). Boxes come from an object-detection model run on each clip; labels and readings are illustrative.

Core analytics

Four things it reads on every camera.

Insight from the camera network you already have. Each family covers several use cases, and they run side by side on the same feed.

Zone A · 82% fullSeating · 41%

CAM 06 · Departure hall

People counting and crowd density

Track foot traffic, watch crowd levels in real time, and alert when a threshold is passed.

Live count per zone and per camera
Crowd density over time, as a heatmap
An alert the moment a limit is passed

Use casesOccupancy and density Baggage reclaim Scheduled service arrival

Wait 03:12

Wait 04:05

Wait 05:20

Wait 06:41

Officer

CAM 11 · Screening checkpoint

Queue length and wait time

Measure how long queues are and how long people wait, to staff ahead of demand.

People in the queue, lane by lane
Wait time from joining to being served, first in first out
Several areas combined into one virtual queue

Use casesQueue and wait time Occupancy and density

Unattended · 4 min 10

CAM 17 · Departures, gate 12

Intrusion and anomaly detection

Detect unauthorised access, unsafe behaviour and unattended objects, with an instant alert.

Entries into restricted or hazardous areas
Bags and objects left unattended
People and vehicles too close for safety

Use casesRestricted zones and line crossing Unattended object detection Person–vehicle proximity PPE and compliance

ULD in · read 10:42

Tug

Stored 3 d 06 h

CAM 44 · ULD storage, lanes A–F

Vehicle and baggage tracking

Follow vehicles, licence plates, bags and containers across zones, for security and operations.

Vehicle movements, dwell and counts across a line
Licence plates read and matched
Bags and containers tracked between areas

Use casesPlate recognition and access Vehicle dwell and bay occupancy ULD identification and storage Aircraft turnaround

Add process, quality and safety detections and you have 18 use cases across four industries.

Explore solutions

How detection works

Every object found, named and placed.

Object detection identifies what is in a frame and exactly where it is. That pair, class and location, is what turns ordinary footage into numbers you can act on.

Apron camera view of an aircraft on stand with ground vehicles, each framed by a detection box

CAM 40 · Stand 24 · 09:14:06

ClassificationWhat the object is, with a confidence score.
LocalisationWhere it sits in the frame.
Bounding boxThe x, y, width and height that frame it exactly.

Detection recordSelect a box on the frame

{
  "class": "aircraft",
  "confidence": 0.98,
  "box": { "x": 261, "y": 61, "w": 936, "h": 392 },
  "camera": "CAM 40",
  "zone": "Stand 24",
  "time": "09:14:06"
}

How it works

Six screens, from first camera to report.

The Eye4.ai screens your team works in, shown as they are, one step at a time.

Screens from the Eye4.ai platform with sample data. Nothing has been added to them.

Architecture and technology

Built to run many cameras on hardware you already have.

Cloud-native, GPU-efficient and cloud-agnostic, so it fits your compliance and latency requirements.

95%Detection accuracy across people counting, queue detection, intrusion and vehicle tracking.

Works with existing CCTV, no proprietary hardware
Face blur built in, privacy by design
An easy interface, quick to roll out

Azure Kubernetes ServiceCloud-native deployment with auto-scaling and high availability.

NVIDIA DeepStreamVideo streams processed in GPU batches, so each GPU carries more cameras.

Flexible AI modelsPretrained models for common use cases, and custom training for new ones.

Operational supportSelf-healing services with proactive monitoring and automated recovery.

Deploy it your way

Choose where the processing runs.

Your site

CCTV camerasExisting, any make

VMS or NVRRecorded and live video

Eye4.ai processingModels, zones and rules

Dashboards and alertsOn your network

Video streams, encrypted

Cloud

Eye4.ai processingAuto-scaling GPUs

Dashboards, alerts and reportsFrom any browser

Not used. Nothing leaves your site.

Fastest to start. Cameras stream to a managed environment, so there is nothing to install on site.

Platform

Built to sit inside your operation.

Your cameras in, decisions out. Eye4.ai sits between the video you already record and the teams and systems that act on it.

Your sites

CCTV camerasExisting cameras, any make

VMS or NVRRecorded and live video

Operational systemsERP, MES, SCADA, bookings

Cloud · your infrastructure · hybrid

1
Ingest and healthStreams pulled from cameras or VMS; heartbeat and diagnostics on each feed
2
Detection modelsPeople, vehicles, PPE, objects and machine state
3
Zones, lines and rulesDrawn in the browser; thresholds you set, no code
4
Events and evidenceEvery event with time, location, measurement and frame

Backup processing during outagesEvery camera monitored

Your teams

Real-time alertsEmail, SMS and chat, to the person who can act

Dashboards and reportsTrends by site, shift and camera

Systems of recordAPI and webhooks back into your tools

Do we need new hardware?No. It reads the cameras and VMS you already run.

Is this another silo?No. Events are matched to ERP, MES, SCADA and operations data.

What if part of it fails?Backup processing keeps detection running during outages.

Do we need developers?No. Cameras, models, zones and rules are self-service.

Security and privacy

Privacy designed in, not bolted on.

Protected at the sourceBuilt-in face blur anonymises people at the source, so personal information is never exposed.
Video stays where you decideRun on your own infrastructure when footage must not leave your environment.
Every event traceableEach detection carries its timestamp, location and frame for investigation and audit.

ISO
27001Information securityBI3 Technologies is ISO/IEC 27001 certified.

ISO
42001AI managementBI3 Technologies is ISO/IEC 42001 certified.

Facial blurringApplied at the source where identification isn’t needed.

Your environmentOn-premises deployment keeps video under your control.

One platform for every camera you already own.

Four things it reads on every camera.

People counting and crowd density

Queue length and wait time

Intrusion and anomaly detection

Vehicle and baggage tracking

Every object found, named and placed.

Six screens, from first camera to report.

Built to run many cameras on hardware you already have.

Choose where the processing runs.

Built to sit inside your operation.

Privacy designed in, not bolted on.

See what your cameras can measure.

Talk to the Eye4.ai team.