A substation or pumping compound has no staff, often a thin connection, and a boundary that matters. Sending every frame to a cloud to be analysed is the wrong shape for that: the analysis has to happen where the camera is.
By the time anyone is dispatched, the event is over. The value is in knowing within seconds whether it is worth dispatching at all.
A site on a cellular link cannot stream continuous video anywhere. Whatever runs there has to run locally and send only what matters.
Dozens of compounds each with a handful of cameras — a very different shape from one site with two hundred.
Which of these apply to you, and which cameras can carry them, is what the site survey establishes before anything is proposed.
A person enters the compound across the fence line.
The boundary camera frame, with the area lit if a floodlight is wired in.
Person or animal, decided from the frame rather than from a sensor state.
Alert with the frame to the on-call phone; voice challenge where a speaker is installed.
Movement at the gate when no visit is booked.
A timestamped frame at the gate line.
Scheduled maintenance crew, or not.
Logged with the frame; escalated where it is not on the plan.
A vehicle detected stopped at the perimeter outside working hours.
The frame showing the vehicle and its position.
A judgement a person makes in seconds with the picture in front of them.
Escalation path that you define per site.
A camera stops streaming or is obscured.
The last usable frame and the time it stopped.
Surfaced as an event rather than found during the next inspection.
A maintenance visit planned rather than a blind site discovered after an incident.
Analysis runs on the appliance at the site. Detection keeps working when the link does not, and only notifications and evidence travel.
What leaves is an alert with a frame, and a clip when you want one — not a continuous upstream per camera.
Each site processes its own cameras; events reach the people responsible, with role-based access per site.
On an unlit compound, what the camera can resolve after dark is the whole question. The survey answers it per camera before anything is bought.
Capabilities are selected per site during the survey. Nothing is on by default, and several depend on camera placement and lighting — which is what the survey establishes.
Identify people, vehicles, objects and security events as they happen.
Add identity, context, movement and event intelligence to live video.
Trigger alerts, sirens, speakers, lights, guard workflows and evidence when action matters.
Camera streams, recording, playback, AI events, evidence clips, alerts, reports, users and roles and camera health — centralised in one appliance inside your premises. Faces, clips and reports stay on site.
Camera count, retention and storage are specified per appliance configuration and confirmed during the assessment.
We do not publish improvement percentages, because none have been measured across a fleet and yours would not match them anyway. What a bounded proof of concept gives you is your own numbers, on your own footage:
For this shape of deployment, usually yes, because continuous video is not being uploaded. What travels is notifications and the clips you ask for. The survey sizes it against your actual tariff and site count.
Detection and recording continue on the appliance, because the analysis is local. Notifications queue and deliver when the link returns.
No. PARVEK is not a monitoring service and not an alarm receiving centre. If you use one, PARVEK gives them events with a picture attached, which is a better thing to act on.
Where speakers and floodlights are present or can be added, yes — both are driven from relay outputs on the appliance and wired at commissioning.
See how PARVEK supports this workflow.
See how PARVEK supports this workflow.
See how PARVEK supports this workflow.
Access and perimeter events across halls, cages and the site boundary, each with the footage that explains it.
Perimeter, restricted zones, yard movement and shift-change activity on a working plant.
Entrances, reception, restricted rooms and after-hours activity across multiple buildings.
AI surveillance and analytics are technology-assisted and do not guarantee prevention or detection of every security incident or 100% accuracy.