Send the same people counting brief to five suppliers and you can get five quotations that vary by more than the cameras themselves cost. That spread is rarely a sign that one supplier is inflating prices. It is a sign that "a people counting camera" is not a single product. It is a configuration — sensors, compute, software rights, housing, power and installation — and every one of those levers moves the total.
This guide breaks down what actually drives the cost of a people counting deployment, why different camera classes sit in different cost bands, and how to compare two quotations that were not written in the same language. We deliberately publish no price figures: project pricing depends on volume, region, channel and configuration, and a single number would mislead almost every real buyer. What we can give you is the structure of the cost, which is the part that travels.
Why there is no single "people counting camera price"
Two deployments can both be described as "count people at the door" and still differ in:
- How the count is produced — 2D detection, stereo (3D) depth, thermal, or a combination;
- Where the AI runs — on the camera, on a recorder or server, or in the cloud;
- What the system must output — raw counts, direction, dwell time, zone heatmaps, queue length;
- What environment it must survive — indoor ceiling, glass facade, outdoor gate, off-grid site;
- How data must travel and how long it must be kept.
Each of these is a cost lever. The number that actually matters is therefore not "price of a camera" but three-year cost per entrance reliably covered, including licences and installation. Hold on to that framing; the rest of this guide follows from it.
The eight cost drivers
1. Sensor and optics
The imaging chain is usually the largest single component. Relevant variables include sensor size and low-light performance, lens quality and aperture, whether the unit uses one lens or a matched dual-lens stereo pair, and whether it adds a thermal detector.
A stereo (binocular) design needs a second sensor and a mechanically rigid, factory-calibrated assembly, so the optical module costs more than a single-lens camera. A thermal-equipped panoramic camera adds a microbolometer detector — a meaningful cost step again. A wide panoramic lens built for ceiling coverage costs more than a standard fixed lens with a narrow field of view. None of these is "better" in isolation; they serve different scene geometries.
2. Where the artificial intelligence runs
There are three common architectures, and they move cost between the camera and the rest of the system:
- On-device (edge) analytics. The camera carries its own compute and outputs events. Higher cost per camera, lower infrastructure cost, low bandwidth demand, and privacy advantages because raw video need not leave the device.
- Server or recorder analytics. Cheaper cameras, but you pay for capable central hardware, per-channel analytics licences, and the network to carry video streams.
- Cloud analytics. Minimal on-site hardware, but recurring per-camera subscription fees and bandwidth costs.
When you compare quotations, check which layer is doing the work. A low camera price attached to a high server and licence cost is not the cheaper system.
3. Algorithm licensing and analytics features
Counting is one feature, not a bundle. Demographics, queue and wait-time analytics, dwell time, zone heatmaps, staff exclusion, occupancy limits and POS-correlation APIs are usually licensed separately or gated by tier. Licensing models differ just as widely:
- per camera, per site, or unlimited;
- perpetual licence versus annual subscription;
- included updates versus paid major versions;
- whether a dashboard or API access is included or metered.
Two quotations for the same hardware can diverge significantly purely because one includes five years of analytics rights and the other includes one.
4. Housing, ingress protection and mounting
The enclosure is set by the environment, not by preference:
- indoor ceiling mount versus outdoor-rated housing;
- IP rating and impact resistance for dust, rain and vandalism;
- operating temperature range, plus heating or ventilation for extreme climates;
- mounting brackets, recessed kits, pole adapters, pendant mounts;
- power method (Power over Ethernet versus a separate low-voltage supply).
If a quotation does not state the IP rating, the temperature range and the mounting method, it is not yet comparable.
5. Connectivity and power
Wired Ethernet with PoE is the simplest option and usually the least expensive to run over the life of the system — provided cabling is feasible. Wi-Fi removes cabling but adds radio engineering and reliability risk. Cellular (4G/LTE) removes both, at the cost of a modem module, a SIM and an ongoing data plan. Off-grid solar adds a panel, battery, charge controller and weatherproofing: a power subsystem that can cost more than the camera it feeds.
Connectivity choice is where "cheap camera, expensive project" most often happens.
6. Installation and commissioning
Labour is frequently the largest line item in a real project and the one most often underestimated:
- mounting at the correct height and angle, which may require lifts, scaffolding or out-of-hours access;
- cable runs, conduit, trunking, power outlets and network points;
- configuration: detection lines, zones, staff-exclusion areas, threshold tuning;
- integration with POS, BI or video management platforms;
- travel and sequencing across multiple sites.
A camera that is quick to calibrate and configure reduces this cost. Ask how long commissioning takes per door, and whether any of it can be done remotely.
7. Platform, storage and subscription fees
Beyond the hardware:
- platform or video management licences;
- storage — on-premise disks, or cloud retention priced per camera per month;
- dashboards, user seats and reporting modules;
- API or data-export fees;
- support and maintenance agreements, and how renewals are priced after the first year.
Recurring costs are where three-year totals quietly diverge. Normalise everything to an annual figure before comparing.
8. Volume, customisation and OEM/ODM engineering
If you are buying as a distributor, integrator or private-label brand, the commercial model changes:
- minimum order quantities and volume price breaks;
- custom firmware behaviour or region-specific requirements;
- branding, packaging and documentation localisation;
- certification testing for the destination market;
- non-recurring engineering (NRE) for a custom design, and who owns the result;
- firmware maintenance commitments across the product life.
OEM/ODM pricing is a different exercise from project pricing and should be quoted as such.
How the cost logic differs by camera class
Rather than quoting numbers, think in terms of cost per entrance covered and three-year cost per counted doorway. You can see how configurations map to real store formats in our retail solutions.
- 2D AI economy cameras. Lowest entry cost. Single lens, on-device detection. Well suited to low-to-medium traffic, single doors and price-sensitive rollouts. In dense or very wide entrances they may need several units or may underperform, which changes the real cost.
- 3D stereo (binocular) cameras. A higher unit cost, justified where accuracy matters in dense traffic, wide entrances, atriums and queueing areas. Because depth handling is more robust, fewer units can sometimes cover a scene — which narrows the gap.
- Panoramic fisheye and thermal cameras. A higher unit price, but one device can cover a large area and support heatmaps and dwell analysis, reducing device count per site. Thermal adds low-light and privacy-preserving options, at a further cost step.
- 4G solar off-grid cameras. The camera may be the smaller part of the bill; the power and connectivity subsystem plus ruggedisation dominate.
- AI recorders (NVR). Shift cost from camera to central compute and per-channel licences. Attractive if you already own cameras, or if you want centralised analytics and tighter per-channel cost control.
The right question is not "which camera is cheapest" but "which configuration covers this entrance reliably at the lowest three-year cost."
Costs that rarely appear on the quotation
Budget for these explicitly, or they will surprise you:
- cabling, conduit, power points and switch capacity for PoE;
- network hardware and, at some sites, bandwidth upgrades;
- licence renewals after the included period;
- installation, travel and out-of-hours access;
- re-calibration after fit-outs or store refits;
- integration work for POS or BI systems;
- spare units and return shipping for failures;
- staff training and change management;
- storage growth and data egress charges.
How to compare quotations without being misled
Normalise before you negotiate:
- Normalise the scope. Rewrite every quotation against one list: entrances, cameras, licences, storage, installation, warranty, training.
- Separate capital from operating cost. Build a three-year total cost of ownership including renewals and support.
- Ask what "included" really means. Licences, warranty years, firmware updates, remote support, API access and installation are the usual gaps.
- Match the technology to the scene. Do not compare a 2D economy camera with a stereo camera as if they were the same product.
- Ask for the accuracy methodology, not a headline percentage. How it was measured matters more than the number.
- Check lock-in. Open protocols, ONVIF support, documented APIs and the ability to export your own data protect you later.
- Ask about reference deployments of similar scale — detailed enough to judge, anonymised if necessary.
The one comparison formula that works
Take the total three-year cost — hardware, licences, installation, support, storage, integration — and divide it by the number of entrances reliably covered. That figure, not the camera sticker price, is what to compare across suppliers and across technology classes.
FAQ
Why don't you publish a price list? Because configuration, volume, region and channel all change the result, and a published figure would be wrong for almost every real project. We quote per project against a defined scope.
Is a cheaper 2D camera always the cheaper option? Not necessarily. In low-traffic single-door sites it usually is. In dense traffic or wide entrances, extra units and rework can erase the saving. Match the device to the scene.
Are cloud subscriptions mandatory? No. On-device and on-premise architectures can run without a subscription, though you may give up centralised dashboards or remote management. Ask explicitly which features require which licence tier.
Does buying directly from an OEM/ODM manufacturer reduce cost? At volume, with the right contract, it often does, because distributor margin is removed. The trade-off is that you take on support, local stock and certification responsibilities — price those in before assuming a saving.
What should I ask for first? A scoped quotation that states the technology class, where the analytics run, licence terms, warranty, installation assumptions and a three-year total. Without those, prices cannot be compared at all.
Next step
If you want a quotation you can actually compare, tell us the entrance geometry, expected traffic, environment and whether you need project pricing or OEM/ODM volume pricing. Browse the lineup in our product center, or send your project details through the contact page and we will structure the quote so the cost drivers are visible on the page rather than buried in the fine print.