REC 00:00:00
CAM 01 – Parking Lot
Pixel-based motion detection fires on rain, headlights, shadows crossing a yard and a plastic bag. After two weeks of that nobody looks at the alerts, and the system is functionally switched off while still appearing to work.
Oshawa is an automotive town with the supply chain and the building stock that implies — large plants, tool and die shops, and a tier of suppliers whose security requirements are set by their customer rather than by themselves. Shift changes create predictable high-movement periods that defeat naive motion analytics, and contractor access during shutdowns is a recurring control problem that most systems are not configured for.
Shift changes and shutdown contractor access are the two periods where most systems fail. Both need explicit rules rather than default settings.
Any analytic needs a settling period per scene — masking a flag, excluding the road beyond the fence, setting a loiter threshold that suits how your yard actually operates. A system handed over without that period generates false alarms and gets ignored, which is the same as not having it.
Plate recognition depends on the angle to the plate, the vehicle's speed and controlled infrared illumination — not on the analytics engine. Beyond roughly 30 degrees off-axis, accuracy degrades quickly. A general-purpose camera pointed at a gate will disappoint; a dedicated LPR camera at the right angle, shutter speed and IR wavelength will read reliably day and night. Budget one camera for plates and a separate one for the overview scene.
Analytics running on the camera cost no bandwidth and no server, and scale linearly as cameras are added — but they are fixed to that camera's capability. Server-side analytics allow richer search across the whole system and retrospective re-analysis, at the cost of hardware and network. Most commercial sites are best served by edge classification for alarms and server-side search for investigations.
Specified by pixel density, not by megapixels
Read more →
Retention you can actually prove, on hardware built for it
Read more →
Credentials that cannot be copied in a parking lot
Read more →
Protection on day one, before there is power or a fence
Read more →
Someone watching, not just something recording
Read more →Under controlled conditions — a dedicated camera, a constrained lane, a reasonable angle and proper IR — it is reliable enough to drive gates and yard logs. Applied to a general camera watching a wide entrance, it is not. The install geometry matters more than the software.
No, but object classification removes the overwhelming majority of them — weather, light changes and animals. What remains is people and vehicles where they should not be, which is the alarm you wanted in the first place.
With server-side analytics, yes — by object type and attributes such as colour, across weeks. It turns a multi-hour scrub into a query, which is usually the moment a client realizes what the system is for.
Shift changes and shutdown contractor access are the two periods where most systems fail. Both need explicit rules rather than default settings. Oshawa is an automotive town with the supply chain and the building stock that implies — large plants, tool and die shops, and a tier of suppliers whose security requirements are set by their customer rather than by themselves.