SEC-11 · Security & access
AI video analytics: cameras that tell you when something needs attention
Nobody watches hundreds of cameras. Analytics classify what the camera sees and raise an alert only for the events you define, and they make hours of recording searchable in seconds.
What analytics do reliably today
- Perimeter and area protection: a person or vehicle crossing a line or entering a zone, ignoring animals, shadows and moving trees.
- Behaviour: loitering, a vehicle stopped where it should not be, an object left or removed.
- Counting: people in and out, occupancy of an area, queue length.
- Vehicles: licence plate reading, wrong direction, parking occupancy.
- Workplace safety: missing helmet or vest in a defined area, entry to a restricted zone.
- Search: find every person in a red shirt or every white pickup across all cameras within a time window.
On the camera, on the recorder or on a server
Analytics can run inside the camera (edge), in the recorder or on a dedicated server. Edge analytics need no extra hardware and keep working if the network drops; server analytics handle heavier tasks and can add intelligence to cameras already installed. We choose per site by the number of cameras, the tasks and the existing equipment, not by one approach for everything.
Honest limits
Analytics depend on the picture they are given. A camera mounted too high, facing headlights or covering too wide a scene will produce missed events and false alarms, whatever the software. We design the camera position for the analytic, tune the rules on site, and measure the false-alarm rate during commissioning before the alerts go to an operator. Face recognition involves sensitive personal data and is used only where there is a clear legal basis and policy.
Our part, start to finish
What we do on this kind of project
We select the equipment to fit your site, not to clear a stock list. We design the system, install and commission it, and keep it working.
Design & consulting
- Definition of the events worth an alert
- Camera positions and views suited to each analytic
- Choice of edge, recorder or server processing
- Alert routing: who is told, and how
Installation
- Analytics licences and servers where needed
- Rule set-up and tuning on site
- Integration with the video and alarm platform
- False-alarm measurement before handover
Maintenance
- Rule review as the site changes
- Software and model updates
- Periodic accuracy checks
- Operator feedback loop
Before you buy
Questions to settle first
These decide cost and performance more than the brand on the box. We work through them with you at survey stage.
- Which events do you need to know about immediately?
- Who receives the alert, and what do they do next?
- Are the existing cameras positioned well enough for analytics?
- How many false alarms per day are acceptable?
- Is counting or reporting needed, or only alerts?
- Are there privacy rules that limit what may be analysed?
Frequently asked questions
Can analytics be added to cameras we already have?
Often yes, through the recorder or a server, provided the picture quality and camera position suit the task. We test on your cameras before recommending.
Will analytics remove false alarms completely?
No system is perfect, but classification of people and vehicles removes most of the nuisance alarms that simple motion detection produces. We tune the rules on site and measure the result.
Do analytics need an internet connection?
No. They run on the camera, recorder or a local server. A connection is needed only if you want alerts on phones outside the site.
Works alongside
Related systems
Video surveillance (CCTV)
From standard IP systems to specialised, high-performance video.
Read moreSEC-07Control rooms
Video walls, Operator consoles, Ergonomics to ISO 11064.
Read moreCTL-03Unified monitoring platform
One map, one event list, CCTV, access, alarms, intercom, Network and device health.
Read more