"Thursdays are quiet" is a guess
Ask when your yard is busiest, how many trucks came in last month, or whether the counter needs a second person at 3pm, and the answer is usually an impression, shaped by the worst day anyone remembers. The footage to settle it has been recorded all along. Nobody has time to tally it off a screen.
How it works
- Detections are counted, not just shown. The analysis that labels people and vehicles in a frame also records them, so counts are a by-product of work already being done rather than new hardware on the wall.
- Counts are kept per camera. A gate, a front counter and a dock each get their own series, because a site total hides what you want to see.
- They are grouped over time. Counts roll up by hour, day and week, which is where patterns show: the 7am run of deliveries, the dead hour after lunch, the Saturday that is not quiet.
- Lines and zones sharpen it. Counting the whole frame includes passers-by. Counting crossings of a line, using zones and tripwires, tells you how many came in.
- Budgets govern how much is analysed. Per-camera and hourly limits keep processing cost predictable, which also means counts come from analysed frames, not every recorded frame.
Analysis runs on our own server; your footage stays on your recorder. AI analytics covers the rest of the platform.
What you see and get
Counts of people and vehicles by camera and period, read as a trend or compared week to week. Each count sits alongside the event timeline, so an odd spike is not a mystery: open the events behind it and watch a clip. What businesses use it for:
| Question | What the counts show |
|---|---|
| Staffing | Which hours are genuinely busy, by camera, across a month |
| Visitor patterns | Days and times people arrive, and how that shifts |
| Delivery frequency | How often vehicles enter the yard, and when |
See warehouses and logistics for how gate and dock counts work together.
Where it fits best
Yards with vehicle traffic, front counters and showrooms, trade supply floors, car yards, and sites deciding whether a shift pattern still matches the work. It is most useful as a trend over weeks and months, not a figure for one hour.
What it does not do
- It counts detections, not unique people. Somebody who walks past the counter four times is four counts. Nobody is recognised or re-identified.
- It is a sample, not a census. Analysis runs on a schedule and within budgets, so a count follows the pattern of traffic rather than being an audited total.
- It double counts across cameras. Two cameras covering the same gate will each count the same truck.
- It gets worse when it gets busy. Groups walking together, people hidden behind a pallet, poor light and awkward angles all cost counts.
- It is not a compliance or occupancy system. Do not use these numbers where a certified count or an occupancy limit is required.
Used as a trend it is genuinely useful. Used as an exact figure it will mislead you, and we would rather say so up front.