AI video analytics suite

What is an AI video analytics suite?

It is software that watches camera feeds and raises alerts about what matters, instead of leaving people to stare at footage. Ours runs at the edge on existing or new camera networks and covers runway foreign-object detection from 7 mm, automated aircraft tracking and friend-or-foe recognition, perimeter and critical-area surveillance, suspect-vehicle tracking and man-overboard detection, across RGB, infrared, audio and point-cloud sources.

See: How it compares to FOD detection systems, ISR capability context

How small a piece of runway debris can it detect?

We publish a 7 mm detection floor, in a runway scan that takes under 30 minutes with 2 to 3 personnel and guided retrieval. None of the named FOD systems, Xsight FODetect, Trex FOD Finder or Pharovision Sentinel, publishes a minimum FOD size on its public page, because detection size depends on lighting, range and surface. The right way to settle it is a trial against the relevant standard on your own runway.

See: FOD detection compared

Can it use the cameras we already have?

Yes. It is edge AI designed to run on existing or new camera networks, across RGB, infrared, audio and point-cloud sources. That is the structural difference from FODetect, Trex and Pharovision, which are fixed radar or optical units installed beside the runway. Because it runs on your own cameras, your own footage becomes the acceptance test rather than a vendor demonstration.

See: Deployment model compared

What can AI video analytics do beyond runway FOD?

Ours also does automated aircraft detection, tracking and friend-or-foe recognition, perimeter and critical-area surveillance with low false-alarm rates, suspect-vehicle detection and tracking, and man-overboard detection. The three named competitors are focused FOD systems. If you want one analytics layer across several security tasks rather than a single-purpose line, that breadth is the point.

See: Scope compared to FOD-only systems, Maritime domain awareness

Runway FOD detection using existing airfield cameras instead of new radar

That is what ours is built to do: edge AI on the camera network you may already own, finding debris from 7 mm in a sub-30-minute scan with 2 to 3 personnel. FODetect and Trex are millimeter-wave radar systems and Pharovision is a fixed optical system, all installed beside the runway. The trade is that a camera-based approach cannot match radar in the worst fog and precipitation, which is an honest limit to weigh against the lower hardware footprint.

See: Camera versus radar approach

Perimeter surveillance analytics with low false alarms running 24/7

Ours does perimeter and critical-area surveillance around the clock with virtual boundaries, dynamic zones and low false-alarm rates. That is a capability the FOD-focused competitors do not offer, because they are single-purpose runway systems. Low false-alarm performance is the figure that decides whether operators trust the alerts, so validate it on your own footage and your own perimeter rather than on a demonstration.

See: Perimeter and multi-task scope, ISR programme context

Analytics that classifies aircraft as friend or foe on long-range feeds

Ours does aircraft recognition with friendly and hostile classification on long-range EO/IR feeds, alongside automated aircraft detection, tracking and turnaround insights. This is outside the scope of the named FOD systems. Classification performance depends heavily on the feed quality and the target set, so it is a capability to trial against your own long-range cameras and the aircraft types you actually need to tell apart.

See: Aircraft recognition scope, Counter-UAS context

Man-overboard detection analytics for vessels in all weather

Ours does man-overboard detection across vessel boundaries in all weather, on the same suite that covers FOD and perimeter tasks. The named airfield competitors do not offer this. As with any vision task, sea state, lighting and camera placement drive real performance, so scope the camera coverage and validate detection on your own vessels and conditions before relying on it.

See: Maritime scope of the suite, Maritime domain awareness

Video analytics from a supplier not tied to a foreign export chain

Xsight is an Israeli company and Trex and Pharovision are US companies, all carrying the export and support questions that follow. Ours is independent and non-aligned, and the suite runs on your own edge hardware and cameras, with data annotation and model development offered as services. That keeps the models, the training data and the alerts inside your jurisdiction rather than on another state's release schedule.

See: Origin and deployment compared

We run a busy airfield in a foggy region. Will camera-based FOD detection be enough?

Be honest with the weather first. A camera-based approach, including ours, cannot match a millimeter-wave radar system like Trex or FODetect in the worst fog, rain and snow, because it depends on what the optical and infrared sensors can see. Where ours earns its place is a 7 mm detection floor in clear conditions, a lower hardware footprint on existing cameras, and coverage of perimeter and other tasks on the same network. If your region is fog-bound much of the year, weigh a radar system for FOD alone against a broader analytics layer for everything else.

See: Weather limits and approach compared

Is the suite certified to the FAA or EUROCAE FOD standard for a civil airport?

We do not publish a named certification, and that is recorded as an open point. Pharovision states it exceeds the FAA Advisory Circular for FOD detection, and Xsight and Trex reference the FAA guidance. For a civil airport, the FAA AC 150/5220-24 or EUROCAE ED-235A standard is often a gating requirement, so resolve certification early in the procurement. For a military airfield or a wider security estate, the breadth of the suite and the non-aligned origin usually weigh more heavily than a civil FOD certificate.

See: Standards referenced compared

Can the models be trained on our own footage and kept under our control?

Yes, that is a core part of the offer. Data annotation and custom model development are provided as services, and the suite runs at the edge on your own camera networks, so the training data, the models and the alerts stay inside your jurisdiction. Because the supply is independent and non-aligned, that control does not depend on a foreign vendor's release schedule. Bring representative footage and your target set to the technical review so the model scope and the annotation effort are planned against real data.

See: Deployment and services compared

How accurately does it locate the debris it finds, compared to a radar system?

We publish guided retrieval to the debris but not a metre-level geo-location accuracy figure, and that is an open point. Trex, by comparison, publishes a 1.5 m geo-tag accuracy and Pharovision displays object geo-location on an airfield map. If precise geo-location for a retrieval vehicle is a firm requirement, raise it in the technical review so the accuracy achievable on your camera geometry is measured rather than assumed.

See: Location accuracy compared

We want one analytics layer across an airfield, a perimeter and a coastline. Does this fit?

That breadth is the reason to choose an analytics suite over a single-purpose line. Ours covers runway FOD, aircraft tracking and friend-or-foe recognition, perimeter and critical-area surveillance, suspect-vehicle tracking and man-overboard detection, all on edge AI across RGB, infrared, audio and point-cloud sources. The three named FOD competitors each do one job. Scope the camera networks for each site in the technical review so one model and alert layer serves the whole estate rather than a separate system per task.

See: Multi-task scope compared, Maritime domain awareness

How do we prove it works before committing, without a vendor-controlled demonstration?

Use your own footage. Because the suite runs at the edge on existing or new camera networks, the acceptance test is your own cameras and your own scenes, not a staged demonstration on the vendor's site. Set the measures that matter to you, such as the smallest debris detected, the false-alarm rate on your perimeter and the classification accuracy on your aircraft set, and run them on real data. That is the honest way to compare a vision suite against a radar system, which can only be trialled on its installed hardware.

See: Acceptance and deployment compared

What is AI video analytics suite?

AI video analytics suite is Unstrat's Land (Land Domain) capability: Edge-deployed computer vision that turns camera networks into alert systems: runway foreign-object detection down to 7 mm debris, automated aircraft tracking and friend-or-foe recognition, perimeter and critical-area surveillance with low false-alarm rates, suspect-vehicle tracking and man-overboard detection. Works with RGB, infrared and point-cloud sources, including existing cameras.

How does AI video analytics suite work?

AI video analytics suite delivers its effect through Runway FOD detection: debris from 7 mm identified in a sub-30-minute scan with 2-3 personnel and guided retrieval, Automated aircraft detection, tracking and turnaround insights across multi-camera coverage with incident recording and Perimeter surveillance with virtual boundaries, dynamic zones and low false-alarm rates, 24/7, capabilities matched to the requirement and confirmed under briefing rather than published.

Who makes AI video analytics suite?

AI video analytics suite is built by The Robotics & AI House, whose focus is unmanned systems, eo/ir sensors & ai video analytics. Unstrat represents The Robotics & AI House to government and enterprise buyers worldwide as an independent, non-aligned prime vendor.

Why choose AI video analytics suite over a major-power alternative?

AI video analytics suite is sourced from an independent, non-aligned manufacturer, so it carries no major-power disclosure rules, upgrade-locks or political ramifications. Concretely: runs on existing camera networks at the edge; the buyer's own footage is the acceptance test. The capability is accountable to you, not to a foreign vendor's government and its release schedule.

How is AI video analytics suite procured, and where can it be exported?

Video analytics measured in outcomes, not pixels: 7 mm runway debris found in under 30 minutes with 2-3 people, perimeters watched around the clock with low false alarms, and aircraft classified friend or foe at long range. Every engagement begins with a briefing, and export eligibility is confirmed per market under briefing rather than published. Where controlled capabilities are involved, the classification and end-user-certificate chain is confirmed first. AI video analytics suite is then sustained in-region by one accountable team from briefing through long-term operation.