Minimum Vehicle Presence Detection: Fleet Availability Alerts
Most monitoring rules fire when something appears. This one fires when something is missing, and that inversion turns out to be useful in more places than you would expect. Eight vans are supposed to be on the apron at 05:30; there are five. Three trucks should be standing at the loading dock during the shift; the dock is empty. The yard should hold at least two shunters overnight; by 02:00 there are none. Nobody is watching for absence, because absence produces no motion, no noise and no event — which is exactly why it goes unnoticed until a customer calls.
Minimum vehicle presence detection closes that gap. You define a region, a required count and a time window, and the system raises the alarm when the count inside the region falls below the threshold for longer than you allow.
What the Rule Actually Measures
Three parameters define every deployment. The region is a polygon drawn on a camera view — an apron, a dock face, a muster area, a covered bay. The minimum count is the number of vehicles that should be there. The window and tolerance say when the rule is armed and how long the shortfall must persist before it becomes an event. A count of two when three are expected for ninety seconds during a shift handover is noise. The same shortfall for twenty minutes is a dispatch problem worth waking someone for.
Underneath, the pipeline is the same one that powers vehicle detection across the rest of the platform: frames are decoded on the appliance, a detector proposes candidates, a classifier confirms the class, and each confirmed vehicle carries a track ID. The rule counts distinct active tracks inside the region rather than raw detections, which is what stops one truck seen by two overlapping cameras from being counted twice.
Handling Occlusion, Turnover and Night
Counting is easy in an empty yard and hard in a working one, so three things matter in configuration. Overlap: where two cameras see the same ground, exclude the shared strip from one of them, or let the platform reconcile duplicate tracks across channels. Turnover: if vehicles legitimately come and go inside the window, the tolerance must exceed a normal round trip or you will fire on every departure. Occlusion: a truck behind another truck is two vehicles, not one, so use a high oblique view that keeps them visually separated.
Night works with infrared-capable cameras and modest apron lighting, with reliability matching the daytime count provided the region sits within the camera’s usable envelope. The compute runs on the appliance rather than a data centre — the architectural choice described under edge computing — and the counting method is a direct application of object detection.
Where It Earns Its Keep
Fleet and transport operators use it as a curfew check: are the vehicles that should be home actually home, and when did the last one arrive? Logistics sites use it at the dock to catch a shift where the planned transport never turned up, rather than discovering it from a missed departure. Plant and transportation depots apply it to stand-by vehicles — the fire pump, the recovery truck, the spare shunter — where appearing on a register and being physically present are two different facts.
It also pairs with the rules around it: expecting a delivery that never arrived is the mirror image of parking occupancy rules, and the two often sit on the same camera with opposite thresholds. Where the shortfall is a missing driver rather than a missing vehicle, workstation abandonment detection covers the equivalent absence indoors. Sites already running perimeter security rules usually add this one at no extra hardware cost.
Configuration and Reporting
Alerts go where the duty rota says — an operator dashboard, an SMS to the shift supervisor, a line in the handover log. More valuable than the alert is the trend: a count history per region per shift, which tells you whether the shortfall is an incident or a pattern. Sites that export this into dispatch reporting usually find the recurring shortfall is structural — a rota assuming four vans when only three exist.
We walk through the reasoning in smart parking system cost.
Frequently Asked Questions
Is this the opposite of a maximum vehicle count rule?
Yes, and the two are often configured on the same region. A maximum count catches over-occupancy and unauthorised standing; a minimum count catches absence, non-arrival and early departure. Same detection layer, opposite threshold.
How do you avoid double counting across two cameras?
Either exclude the overlapping strip from the second camera’s region, or let the platform reconcile duplicate tracks so a vehicle seen by both is counted once. We confirm which is appropriate during configuration, from your actual camera geometry.
What if a vehicle is partly hidden behind another one?
Partial occlusion is handled by the tracker, but severe occlusion is a siting problem rather than a software one. A high oblique view that keeps vehicles visually separated is the reliable fix, and it is usually achievable with a camera you already have mounted at a different angle.
Do we need to buy new cameras or a cloud subscription?
No. The rule runs on your existing streams on an appliance inside your network, from 2 to 128 channels and 1 to 256 TOPS, with no per-camera licence and no video leaving your site. If a region has no coverage today, one ordinary IP camera is typically all that is required.
Tell Us Your Scenario
Every site draws its zones differently, and the difference between a rule that works and a rule that gets switched off is usually thirty minutes of configuration. Send us your camera layout and the behaviour you need to catch, and we will tell you which of the 198+ algorithms fit and what accuracy to expect on your footage. Tell Us Your Scenario — or start with a Starter Kit preloaded with tested configurations, running on your own cameras in a single afternoon.