Video AI analytics Vision AICentinals

Move from watching video to understanding what is happening.

  • No-code platform
  • Edge or cloud
Centinals Vision AI consoleVision AIAI-POWERED SOLUTION BY CLL SYSTEMS4 running of 408:42:15Fri 16 OctADadminSuper AdminMENUDashboardLiveReportSettingsSystem Overview4 of 4 pipelines running· online since 16/10/2026, 6:00 AMUpdated 08:42:10CAMERAS ONLINE12/ 12cameras with a running pipeline100%PIPELINES RUNNING4/ 4no errors100%DETECTORS ACTIVE6/ 6AI models in use100%ALERTS3in latest 200 eventsCAMERA MAP10 of 12 streamingStreaming 10Alert 1Idle 1© OpenStreetMap contributorsLIVE — JUNCTION 3OnlinecarcarcarcarvanA IN1718B OUT1112C IN9102026-10-16 08:42:15ALERTS LOGView all →No helmet detectedCAM 05 · Site gate46Restricted zone entryCAM 07 · Loading bay29Wrong-way vehicleCAM 03 · Junction 30PIPELINE STATUSManage →Vehicle counting4 cameras24 fpsRunningPPE detection3 cameras18 fpsRunningIntrusion zone5 cameras15 fpsRunning

Vision AICentinals applies video analytics to existing camera infrastructure. It can detect defined activities or conditions, generate alerts and help teams respond without relying only on continuous human observation.

Walkthrough

How Vision AICentinals supports a smart city

A city district with AI cameras at a junction, a police headquarters, a hospital and a smart car park
Scene 1 of 6Jalan Merdeka × Jalan Sentral

Continuous monitoring at every junction

Vision AICentinals analyzes the live feed from each junction camera, identifying every vehicle and pedestrian in real time and measuring traffic flow in each direction.

Capabilities

01

Works with your existing CCTV

Use compatible existing cameras as inputs for defined detection, monitoring and alerting use cases.

02

Edge or cloud inference

Edge AI servers run models close to the data source for lower latency and less bandwidth. A cloud inference endpoint serves scale. The architecture is chosen per site and use case.

03

No-code model building

Configure analytics for defined use cases. The setup approach and level of technical support depend on the required model and environment.

04

Smart count reporting

Object counts, dwell time and occupancy levels in real time, supporting live insight, alerting and resource allocation.

05

Alerts through the right channel

Send alerts via mobile push notifications, email, WhatsApp, Telegram and other configured channels, with short video previews and action buttons for quick response.

06

Autonomous perimeter defense

Configure detection and response workflows for defined perimeter events. Any automated response capability should be described only after the use case, operating controls and regulatory requirements are confirmed.

Inside the console

Detectors you can attach to a camera zone

Each detector is a model that runs on a zone of a camera. The tiles below show what each one returns.

Video AI detecting hard hats and safety harnesses at an oil and gas siteHard hat 0.97Harness 0.94
Oil & Gas
Video AI detecting work at height and harness use on a construction siteWorking at height 0.95Harness 0.93
Construction
Video AI detecting welding masks and hot work in a manufacturing plantWelding mask 0.95Hot work 0.91
Manufacturing
Video AI detecting ship-to-shore cranes and containers at a portSTS crane 0.93Container 0.96
Logistics & Ports
Video AI measuring crowd density and detecting people in a public spacePerson 0.97Crowd density 0.88
Public Safety
Video AI detecting vehicles at a road junction for traffic enforcementVehicle 0.94Junction 0.92
Traffic & Enforcement
Video AI counting occupied bays in a car parkOccupied 22 bays
Parking
Video AI detecting forklifts and people in a warehouseForklift 0.95Person 0.92
Warehouse
Video AI detecting smoke for disaster and environmental monitoringSmoke 0.91
Disaster & Environment

Also covering

  • Smart City
  • Traffic & Enforcement
  • Warehouse
  • Retail
  • F&B
  • Petrol Stations
  • Palm Oil
  • Airports
  • Education
  • Critical Infrastructure
  • and more…

Capture, understand, act

  1. 01 Capture

    Cameras and IoT sensors

    Your existing CCTV estate plus environmental and access sensors feed the platform live.

  2. 02 Understand

    AI inference

    Detection, tracking and analysis on every frame, with a compliance module for governance, privacy and audit-ready control.

  3. 03 Act

    Automated execution

    Middleware and dashboard orchestrate the output. Alerts, workflows and decisions trigger from live video in seconds.

One camera network can serve several functions at once. The city CCTV managing traffic also watches for floods, fire and illegal dumping.

Beyond off-the-shelf

We don't just deliver software.
We fit it to how you work.

Every organization runs differently. Our engineers work with your team to understand your processes and challenges, then tailor Vision AICentinals so your systems work together instead of in silos.

Tell us how you work
  1. 01UnderstandYour processes and challengesWe assess your sites, risks and current response procedures.
  2. 02AddModules built for your needsDetection for situations specific to your operation.
  3. 03ConnectIntegration with your existing systemsYour CCTV and NVR, access control and command center.
  4. 04AutomateWorkflows that run themselvesAlerts routed to the right person, with escalation when no one responds.

Questions

Do we need to replace our cameras?

No. The platform is designed to work with existing CCTV infrastructure. The cameras you already have become the sensors.

Edge or cloud?

Both are supported. Edge AI servers run close to the data source for lower latency and bandwidth. The cloud inference endpoint serves scale. The mix is chosen to fit the site.

Do we need AI engineers?

No. Model customization is no-code, through a visual interface, and the setup wizard is a drag-and-drop pipeline builder with ready-made components.

How do alerts reach our team?

Mobile push, email, WhatsApp, Telegram or any other communication channels, with a short video preview and action buttons. A dedicated mobile app can also be developed for your organization.

Will animals trigger false drone launches?

Response workflows are configured for the agreed use case. Any drone or automated-response capability is subject to technical, operational and regulatory review.

How fast is the perimeter response?

Response time depends on the detection model, network, system design and selected workflow. It should be confirmed for each use case.

Can one deployment cover several use cases?

Yes. The same camera network serves multiple functions simultaneously.

See Vision AICentinals running

Tell us the challenge. We'll show you what Centinals can do.