Crowd and Density Detection
Live occupancy counting per zone, measured against the threshold you set, so a dangerous build up is flagged while it is still forming.
See the Demo
What it does
Counting the zone, not the camera
A single camera count tells you little when a space is covered by several. Trinocular AI counts occupancy per zone, so the number a steward sees is the number of people in the area they are responsible for, however many cameras happen to cover it.
Density, dwell and flow
Beyond a headcount, the system measures how tightly packed a zone is, how long people are staying, and which direction they are moving. Those three together are what distinguishes a busy concourse from one that is about to become dangerous, and they are what let an alert fire early enough to act on.
What a steward actually receives
When a zone crosses your configured threshold, the alert names the zone, the current occupancy and the trend, and reaches stewards and the control room at once. Historic peak load and flow patterns per entrance are retained so the next event can be planned against real numbers rather than estimates.
Privacy by design
Crowd counting identifies no one. The model counts people and measures density without matching faces or holding identity, so a zone stays measurable without creating a biometric record. That is a deliberate design decision, and it is what makes occupancy monitoring deployable in public spaces where identification would not be.
At a glance
- Measures
- Occupancy, density, dwell, flow
- Scope
- Per zone, across multiple cameras
- Identity
- None held or matched
- Output
- Live count, threshold alerts, history
- Count Live occupancy per zone, not per camera
- Measure Density, dwell time and direction of flow
- Alert Stewards and control room at your threshold
- Report Peak load and flow history per entrance
Counts people rather than identifying them, so the zone stays measurable without holding identity; the faces above are obscured for that reason. Accuracy depends on camera angle, height and how much the crowd overlaps in frame.
Questions
How accurate is the count in a dense crowd?
Accuracy falls as overlap rises. An elevated camera looking down across a zone counts far more reliably than one at head height looking along it, which is why camera angle and mounting height are assessed before deployment rather than after.
Does it identify individuals?
No. Crowd and density detection counts people without matching or storing identity. If you need identity matching that is a separate capability with its own policy requirements.
Can we set different thresholds for different times?
Yes. Zones and thresholds can be scheduled, so an area that is unremarkable at midday and critical during an event exit can be treated differently without reconfiguring anything.
Most often deployed in
See it on your own cameras
Bring your existing CCTV. We will assess what your angles can actually support and show you the detections running on your own footage.