SquintPRO: every question a site asks

Usually not. The system is designed to work with an existing CCTV estate. Some cameras will be unsuitable for analytics because of placement, angle or resolution, which is what the site survey establishes, but that is normally a subset rather than the whole estate.

More than any other single factor. A camera placed for human review is often wrong for analytics: mounted too high, angled too steeply, too far from the point where PPE needs to be visible, or looking into direct sun for part of the day. Placement, resolution at working distance and lighting set the ceiling on what any model can do, which is why the survey comes before any performance conversation.

On Saudi sites, the recurring ones are dust accumulating on camera domes, midday glare and heat haze, steep sun angles, and night-time areas lit for people rather than for cameras. None of these is disqualifying; all of them affect detection and false-positive rates, which is why a pilot should run in real conditions rather than in a corner of the site chosen to look good.

Oil and gas is one of the sectors this work runs in, and it brings constraints other sites do not have: classified hazardous areas restrict what hardware can be installed and how, work happens under permit control, and much of the workforce belongs to contractors. Those constraints shape the survey and the deployment plan, and they are covered in detail in the oil and gas guide.

The classes a site typically monitors are hard hats, high-visibility vests and the other items its own PPE matrix mandates, and detectability genuinely differs between them: large, high-contrast items like helmets and vests are the strong cases, while small items such as gloves or eye protection are harder for any vision system at CCTV distances. Ask any vendor, including this one, for the current supported class list in writing during scoping, and pilot the classes that matter on your own cameras rather than accepting a list on faith.

No. It is Interkey’s own product measurement under SquintPRO’s reference conditions, not an independent benchmark. Computer vision accuracy depends on camera placement, lighting, weather, dust and how closely your objects resemble the training data. Run a pilot on your own cameras and measure both false positives and missed detections.

Yes. Deployments can run in-Kingdom, including on-premise where footage must not leave the site at all. For many industrial and government sites this is a precondition rather than a preference.

As a design input. The deployment defines what is retained and for how long, who can view footage, where masking applies, and how employees and contractors are informed, with the site’s HSE and legal owners deciding those rules. Detection records can be kept without retaining continuous footage, and processing can stay on site entirely. What this page deliberately does not do is interpret Saudi data-protection law for you: bring your legal owner into the design conversation early.

There is no separate vendor: SquintPRO is Interkey’s own video intelligence product (product site: squintpro.ai). The product is one half; the delivery is the other. Interkey surveys the site, integrates SquintPRO with your existing VMS and alerting, deploys it in-Kingdom and tunes it against your conditions.

The alert is routed to whoever the site decides should act, in the tools the site already uses: control room, area supervisor, HSE duty phone. The event is also kept as a record with a timestamp, camera, zone and event class, so it can feed investigations and the site’s leading-indicator reporting. Designing that workflow with the HSE team is part of the deployment.

As a pilot: one site, a defined set of cameras, one or two detection cases, and success criteria written down before it starts, including an acceptable false-alert rate per shift. That gives you a real performance figure for your own conditions before any wider commitment. The pilot guide sets out the full method.