One platform. Every hazard that matters.
AI that runs on the cameras you already own, detecting hazards in real time, on-premise, at a sub-0.05% error rate proven offshore.
From camera to alert in under a second.
SecureSafety is an AI layer that attaches to the CCTV you already run — no rip-and-replace, no new hardware to begin. The platform ingests live streams, runs detection models on an on-site edge device, and pushes alerts to your control room, site speakers and worker wearables. Raw footage stays on your network throughout.
Ingested from your existing cameras
Video streams are pulled directly from your current CCTV infrastructure — IP cameras, DVRs and NVRs — over your local network. No new camera hardware is required and footage never leaves your site.
Analysed on-premise in real time
An edge device installed on-site runs the detection models locally, processing every frame against the active detector configuration. Latency from event to alert is typically under one second. No cloud dependency, no bandwidth cost.
Alerts surface to the right person instantly
Detections reach the control room dashboard, site PA speakers, and wearable smartwatches simultaneously. The platform filters noise to a sub-0.05% error rate, so alerts carry weight — crews respond because every alert matters.
90+ detectors. One platform.
These are safety detectors running in production today: everything from PPE and falls to red zones, suspended loads and vehicle speed. And this is only the high-frequency set: we run many more, and train new detectors continuously.
People detection & tracking
The foundation of every downstream detector. The platform tracks individuals through the scene in 2D and 3D, counting occupancy, following trajectories and handing context to higher-level detectors.
- People detection
- 3D people detection
- People detection (ROI)
- People detection (occlusion)
- Person tracking
- Pose estimation
- 3D pose estimation
- Per-camera counting
- Multi-camera counting
- Minimum people count
PPE compliance
Every worker entering a zone is checked for the right protective equipment the moment they step into frame — not during a spot check at the gate. The platform verifies item presence and colour, so contractor PPE rules are enforced alongside crew rules.
- Hard hat
- Hard hat colour
- Safety vest
- Vest colour
- Gloves
- Safety glasses
- Ear protection
- Masks
- Boots
- Harnesses / belts
Falls, fatigue & behaviour
These are the leading indicators that precede an incident — distraction at the controls, a person who has stopped moving, crowding at a pinch point. The AI surfaces the human signals while there is still time to act.
- Fallen person
- Dizziness
- Fatigue (face)
- Fever (face & body)
- Running
- Walking
- Stationary person
- Loitering
- Crowding
- Inattentive / distracted
- Walking backwards
- Looking at screen
- Specific behaviour
Zones, access & working at height
Draw an exclusion area once and the platform enforces it permanently. Red zones, line-of-fire areas and height restrictions are monitored continuously — day and night, across every camera simultaneously.
- Red-zone management
- Line crossing
- Edge / ledge proximity
- Climbing / object scaling
- Working at height (man-rider)
- Drill-floor shelter position
- Master bushing
- Escape-route obstruction
- Human-machine segregation
- Space / occupancy
Hands, tools & devices
Granular hand and tool tracking closes the gap between seeing a worker in a zone and knowing what they are actually doing there. Mobile-phone detection flags distraction at the moment it happens.
- Hand on object
- Hand on button
- Hand on handrail
- Tool in hand
- Mobile phone in hand
- Mobile phone use
- Smart-device detection
- Smart-glasses detection
Environment, load & vehicles
Vehicles, suspended loads and fire are the three hazard classes most likely to cause a fatality or a mass-casualty event. This group watches them all in parallel: speed violations raise an alert in under a second.
- Fire & smoke
- Suspended object
- People under suspended load
- Equipment in motion
- Vehicle speed
- Door / gate / floor opening
- Rain detection
- Operational exposure level
- 5S workplace organisation
- Firearm detection
…and many moreIf it’s visible on camera, we can likely detect it. And we build custom detectors for your site.Ask about a detector
Why on-premise matters for safety detection.
Every detector in the SecureSafety platform runs on an edge compute device installed on your site. No frame leaves your network in the course of normal operation. This is not a privacy feature added to a cloud system — it is the foundational architecture that the platform was built on, because the environments where we first deployed it (offshore drilling rigs, with satellite bandwidth constraints and strict data security requirements) made cloud processing impossible.
The implication for safety performance is as important as the privacy implication. Cloud video analysis introduces a network round-trip into the detection pipeline. The sub-200ms alert latency that characterises SecureSafety deployments is only achievable when the inference happens on-site.
Multi-model inference — running PPE detection, zone monitoring, vehicle tracking, fire detection and behaviour analysis simultaneously on a single camera frame — is computationally intensive. The edge hardware specification for each deployment is calculated during the Discovery phase to ensure that the GPU capacity matched to the detection load delivers consistent sub-200ms performance.
Custom detector development is available for hazards specific to your site. Custom detectors are trained on footage from your cameras, in your environment. Most custom detectors are available within four to eight weeks of commissioning.
Tell us the hazard. We’ll detect it.
Book a demo and we’ll show the detectors running live, on-premise, and scope the ones that fit your site, including anything custom you need.