People counting with re-identification across entrances
Simple door counters double-count people who come back or use several entrances. Nara counts people crossing lines at each entrance and uses person re-identification (ReID) to recognise the same visitor across cameras and devices, giving unique visitor counts and entries and exits by hour.
[ 01 ]The problem
Simple door counters double-count people who return or use several entrances.
- 01inputEntrance cameras
- 02processPeople detection, tracking and line crossing; ReID links the same person across cameras and devices
- 03actionUnique visitor counts and entries and exits by hour, on dashboards and over MQTT
How it works
Nara detects and tracks people at each entrance and counts them as they cross the line in or out. Person re-identification recognises the same visitor across cameras, and multi-device sync shares that identity between Nara devices, so a returning visitor is not counted twice. Results feed dashboards and are published over MQTT to other systems.
Works with Nara’s built-in models. You set up the zones, rules and outputs.
[ 02 ]Variations
The same building blocks, applied a few ways.
- 01
Malls and stores
Unique visitor counts, entries and exits by hour, across every entrance.
- 02
Events
Attendance per event space, combined with zone traffic and demographics for organisers and sponsors.
- 03
Pets and pet carts
Pets are detected and counted at entrances for pet-friendly programmes. Pet carts can be trained in Reva.
What you need
- IP cameras at each entrance
- One or more Nara devices; several can be paired into one site
- Optional: an MQTT broker for other systems
Nara building blocks
- Multi-camera input
- People detection
- Tracker with line counting
- Person re-identification (ReID)
- Multi-device sync
- Dashboard
- MQTT publish
- AI model: People detection
- AI model: Object tracking and analytics
- AI model: Person re-identification (ReID)
What it improves
- Accurate footfall
- Peak-hour staffing
- Tenant, marketing and event decisions
[ 03 ]Where it runs
Proof from real deployments.
FAQ
Questions, answered.
Does re-identification use faces?
No. Person re-identification recognises the same person across cameras from their overall appearance, without faces.
Can it cover a whole building?
Yes. Several Nara devices can be paired into one site and share ReID data, so a visitor keeps the same ID across devices. Our shopping mall deployment runs 120+ cameras on 4 edge devices in sync.
Related
Related use cases
Heatmaps and dwell time
Operators can count visitors, but cannot see where they go, how long they stay or what catches their attention.
cameranara seesVisitor demographics
Operators know how many people visit, but not who they are, which makes tenant, marketing and event planning guesswork.
Queue monitoring
Long queues lose customers, and in factories work-in-progress builds up unseen until it stops the line.
Talk to us about people counting.
Tell us about your site, cameras and machines, and we’ll show you how it runs on your own hardware, or download Nara and try it yourself.