Security · Defence
ProactiveGuard
A camera that nobody is watching records an incident. It does not prevent one.
Overview
Almost every site with a security problem already has cameras. What it does not have is anyone able to watch them all, all the time. Footage gets reviewed after something happens, which is the least useful moment to look at it.
ProactiveGuard puts an AI layer over the estate you already own — CCTV, body cameras, dashcams — and turns continuous watching into alerts. The word proactive is doing real work here: it flags an unknown face appearing at your perimeter for the third time this week, before that becomes an incident to review.
It is layered rather than monolithic. Detection, recognition, number plates and enhancement are separate models you add or remove according to what the site needs. One of those layers is mask face recognition, which very few systems have at all — and which is the layer that decides whether the rest of it is worth running at a site where people arrive covered.
Uncovered 
Covered Capabilities
What ProactiveGuard does
Recognition through a covering
Identifies people whose faces are partially covered. Most recognition systems simply fail here, and a covered face is precisely the one worth recognising.
Works with what you have
CCTV, body cameras and dashcams, across the standard video and image formats including RAW.
Monitoring without a watcher
Alerts and notifications instead of a person in front of a monitor wall. Nobody has to be watching for it to work.
Stackable AI layers
Object detection, drone detection, number plate recognition for civilian and defence vehicles, and Triveni enhancement — added or removed per site.
Tracking across zones
Follows a person between cameras and over time, so movement across a site reads as one path rather than a dozen disconnected clips.
Alerts before the incident
Someone lingering at the perimeter, the same unknown face back days later, or a person seen outside earlier now inside the premises — the reconnaissance pattern, raised as an alert with the clips attached.
What almost nobody else has
Mask face recognition
Anyone can recognise an uncovered face; that problem has been solved for years and is sold by everyone. Recognising someone who does not want to be recognised is a different problem, and very few systems attempt it at all. We built one that does.
Mask face recognition
Rare capabilityThe covered face is the one that matters
A person walking a perimeter with their face covered is not an edge case to be tolerated by the system. It is the single case the system exists to catch, and it is where conventional recognition returns nothing at all.
Trained specifically for it
This is not a general recognition model that degrades gracefully. It is a model built and evaluated for partially covered faces, against a large set of them, and it runs as its own layer.
Same identity across cameras and time
A covered face that is recognised can be tracked like any other — across camera zones and across days, which is what turns a sighting into a pattern.
Proven on your footage, not in a brochure
We do not publish accuracy percentages here. Numbers from somebody else's test set tell you nothing about your cameras, your lighting or your site. We will run it against your material and you can judge the result.
Who it is for
Built for these situations
Defence and perimeter security
Sites where continuous monitoring is required and staffing it continuously is not realistic.
Critical infrastructure and industry
Large estates with an existing camera investment and no appetite to replace it.
Investigation teams
Anyone who needs to search recorded footage for an event rather than watch it end to end.
Deployment
- On-premise, over your existing camera infrastructure.
- Handles the standard broadcast and camera formats, including RAW.
- AI layers are configured per site — you take what that site needs.
- Live feeds and archive footage are both supported.
Questions
Frequently asked
Do we need to replace our cameras?
No. It is designed to work with the security camera system already installed.
Does it need someone monitoring it?
No. It raises alerts and notifications rather than requiring a person to be watching.
Can it run without a cloud connection?
Yes. On-premise deployment is the normal configuration for this product.
Can we run only some of the layers?
Yes. The AI layers are independent and configured per site.
Does recognition still work if someone covers their face?
Yes — that is the mask face recognition layer, and it is unusual. Most systems treat a covered face as a non-detection. We do not publish accuracy figures on a marketing page; we will demonstrate it against your own footage instead.
Can it process footage we already have?
Yes. Recorded material is processed the same way as live feeds, so an incident from last month can still be investigated properly.
Interested in ProactiveGuard?
Every deployment is customised to the organisation running it. Tell us about your environment and your constraints, and we will tell you honestly whether this is the right fit.