Applications / Safety

Worker Down Detection

Detects a person who has fallen, collapsed or become immobile in camera view, and raises an alert with the location so responders know where to go.

Worker Down – AI Detection for Proactive Workplace Safety — Fogsphere detection illustration

See it in action

Worker-down and immobility detection with location

What it detects

Harvested from the computer-vision use-case columns of the Q3 industry papers. Every item below is a rule you can switch on, tune per zone and report on.

Fall and collapse events

Detects a person going down in view, whether from a slip, a trip, a fall or a medical collapse.

Prolonged immobility

Flags a person who remains motionless beyond a configured threshold, which the Q3 chemical paper pairs with worker-down detection.

Abnormal posture held over time

Surfaces sustained abnormal posture as a distinct signal from a sudden fall.

Lone workers in remote zones

Covers lone-worker detection in remote or high-risk areas, and access outside normal working hours.

Out-of-hours presence

Flags people present after a shift ends, when nobody is expected to be there to notice.

SOS and man-down device correlation

Lines up SOS activations and man-down devices with the nearest camera so the alert arrives with a picture.

Confined-space and vessel entry points

Applies to tank, vessel and confined-space entries where the attendant watch and worker count are already tracked.

Slips and trips on wet or obstructed floors

Detects the events themselves, and the floor conditions the Q3 papers list alongside them.

Heat-stress indicators

Surfaces heat-stress signals where atmospheric and wearable sensor data is available to correlate.

Where it applies

The sub-sectors below are the ones the Q3 industry papers name for this detection. Each links to the industry hub.

In Oil & Gas

Person-down detection across drilling rigs, offshore platforms, refineries, gas stations and construction sites. Faster emergency response in connectivity-constrained environments.

How it works

01

Runs on the cameras you already have

Inference happens at the edge on your existing CCTV and IP streams. No new camera estate, no video leaving site, no cloud round-trip in the detection loop.

02

Thresholds and zones you configure

Draw the zone, set the rule, set the dwell, count and confidence thresholds. Rules differ by area, shift and task, because sites do.

03

Evidence, not just alerts

Every event is logged with camera, zone, timestamp and a clip, so the record is reviewable afterwards and usable in a corrective action.

04

Integrates with what you already run

Events push into your VMS, alerting and reporting stack through Fogsphere Manager, with dashboards broken down by area, shift and crew.

Risk & control objective

The variable that matters after a person goes down is elapsed time to a responder arriving at the right place. Every other factor is fixed by the time the event happens. The Q3 chemical paper tracks incident and worker-down response time as a headline operational metric for exactly this reason.

Lone working sharpens the problem: in a remote zone, out of hours, or inside a vessel, there may be nobody present to raise the alarm at all. The control objective is coverage of those situations and a location attached to the alert — so that response time is measured from the event rather than from whenever somebody happened to notice.

Related solutions

Line of Fire

View solution →

Work at Height

View solution →

Grouping & Congestion

View solution →

See Worker Down running on your own cameras

Bring a camera stream and a zone. We will show you what the model surfaces on your site, not on a demo reel.