AI-generated footage (Adobe Firefly) · detections by the VISION PPE model

Green Codes VISION

AI-powered visual safety intelligence

VISION looks through inspection cameras for missing PPE, people inside exclusion zones and other potential hazards — then puts every finding in front of a competent person to confirm before anyone acts on it.

  • A person confirms every finding
  • Arabic and English, right-to-left ready
  • Feeds Green Codes Pro observations
reviewed by a person before any action
Every finding
facial recognition or worker identification
No
full Arabic and English, right to left
EN · ع
configurable evidence retention
7–1,825 days

The challenge

Sites change by the hour. Supervision cannot be everywhere.

Most unsafe moments are brief and go unrecorded. The ones that are noticed often never become an action anyone follows up.

PPE lapses are fleeting

A helmet comes off for a minute; a vest is left in the cabin. Those moments rarely make it into a report.

Exclusion zones get crossed

Under schedule pressure, people step behind barriers or into the working area of plant.

Evidence is scattered

Photos sit on personal phones and chat groups, without a time, a location or a decision attached.

Follow-up stalls

Findings that never become observations and corrective actions are rarely closed out.

Before and after

From “someone might have seen it” to a reviewed record

Without VISION

  • Hazards are caught only when someone happens to be looking
  • Photos arrive without context or a decision
  • Findings are retyped into the HSE system — or never are
  • No consistent record of what was reviewed, by whom, and why

With VISION

  • Camera frames are checked for potential hazards as they arrive
  • Each finding carries its evidence frames, time and location
  • A competent reviewer confirms or dismisses it — with a reason
  • Confirmed findings become Green Codes Pro observation drafts

Detection workflow

Capture. Analyse. Verify. Notify. Act. Review.

Six steps, one record. The AI does the watching; people make the decisions.

01 Capture

An inspection session takes frames from a USB camera on an Android device, a phone camera or a browser webcam. Inspectors can add photos, clips and notes by hand.

02 Analyse

The VISION inference service detects people, PPE, plant and smoke, follows them across frames, and applies the project’s rules and zones.

03 Verify

Each potential hazard waits for a competent reviewer, who sees the evidence and confirms it or dismisses it with a reason.

04 Notify

In-app and e-mail notifications go to reviewers and owners; confirmed findings can be spoken to the inspector through an earpiece.

05 Act

Confirmed findings become Green Codes Pro observation drafts, and corrective actions are assigned and tracked.

06 Review

Safety Walk Reports, a findings register, a risk matrix and exports give Head Office a view across sites — and show where detectors need improvement.

Product preview

Real screens, not mock-ups

These are captures of the VISION exhibition display and review inbox. The footage is AI-generated for demonstration; the detections and labels are VISION’s real output, exactly as shown.

VISION exhibition display showing a potential missing safety helmet, pending human review.
Real VISION screen — exhibition environment with AI-generated demonstration footage. Nothing on screen is staged.
VISION exhibition display with bounding boxes around people on a construction site and the helmet and vest status of each.
Real VISION screen — exhibition environment with AI-generated demonstration footage. Nothing on screen is staged.
VISION exhibition review inbox listing potential findings with Confirm and Dismiss buttons and a false-positive reason.
Real VISION review inbox — exhibition environment with demonstration data.
VISION Android app monitoring screen with camera, AI, network and battery status.

Real VISION Android app screens.

Capabilities

What VISION looks for

Each detection is raised as a potential hazard with its evidence attached; a competent person decides.

See all detection categories

Missing safety helmet

Flags a person whose head shows no hard hat, with the evidence frame attached.

Missing high-visibility vest

Flags a person without a hi-vis vest or garment in view.

Exclusion-zone entry

On a fixed camera, a person inside a zone you draw becomes a potential hazard.

Person near machinery

Highlights people close to excavators, dump trucks and bulldozers. Always review-only.

Visible smoke

Flags visible smoke outdoors. Dust and haze can trigger it, so every alert is reviewed.

Inspector-raised findings

Inspectors tag any hazard by hand with a photo, clip and note — no AI involved.

Integrations

Built into the Green Codes platform

VISION is a module of Green Codes Pro, so findings flow into the HSE records your teams already use.

  • Green Codes Pro observations

    A confirmed finding becomes an observation draft with its evidence attached.

  • Corrective actions

    Follow-up is assigned and tracked through Pro’s corrective-action workflow.

  • Inspection cameras

    USB cameras on Android devices, phone cameras and browser webcams.

  • Mobile inspector app

    Android safety walks with evidence capture and spoken alerts in Arabic or English.

  • In-app and e-mail notifications

    For high-severity hazards, escalations, assignments and completed walk reports.

  • Reports and exports

    Safety walk reports and findings registers as PDF, Excel or CSV.

See VISION with your own site in mind

Book a guided demonstration. We will walk through the exhibition scenarios, the human review workflow, and what a rollout on your projects would involve.