AI people counting cameras
An AI people counting camera is a standard network camera that does the counting itself. The counting model runs on the device, so exact in/out counts, occupancy and queue metrics come straight from the camera, and only numbers leave it. In counting mode no individuals are identified and no personal images are stored.
Counting analytics built into the camera, anonymous by default.
Trusted by
Why AI cameras
-
Analytics in the camera
Counting runs on the camera itself, so a single device covers detection, counting and delivery with no separate sensor unit to install.
-
Exact counts and queues
Directional in/out counting, live occupancy and queue metrics at entrances, checkouts and service points.
-
Anonymous counting by default
In counting mode, video is processed for analytics only. No personal images are stored and no individuals are identified.
-
One platform, every method
AI camera data lands in the same analytics platform, dashboards and API as Wi-Fi, 3D sensor, LiDAR and cellular data.
How it works
-
See
A standard network camera watches its zone, and the counting model runs directly on the device.
-
Count, privately
In counting mode the video is analysed in the moment and only the numbers leave the camera. No personal images are stored.
-
Analyse
Counts become in/out, occupancy, dwell and queue metrics for the covered point.
-
Deliver
Metrics flow into the same API and dashboards as every other Bumbee Labs method, so all methods share one source of truth.
What data you get from AI cameras
AI cameras are a precision method at defined points, with the practicality of standard camera hardware. Typical metrics:
- Exact in / out counts (directional line-crossing)
- Real-time occupancy and capacity against safe limits
- Queue length and waiting-time indicators at service points
- Dwell at the measured point
- Peaks & lows at the measured points
For whole-space paths and journeys, combine with Wi-Fi; for camera-free precision in open or dark areas, see LiDAR. The method grid shows what each delivers.
On this page
What is an AI people counting camera?
An AI people counting camera is a standard network camera with a counting model built into the device, so the camera that watches an entrance, checkout, service desk or reception also counts there. It delivers exact directional in/out counts, occupancy and queue metrics into the same analytics platform and API as every other Bumbee Labs method.
In counting mode the analysis happens in the moment and only numbers leave the camera. No personal images are stored and no individuals are identified.
Unlike conventional CCTV, which records images for people to review, a counting camera is configured to measure. It also differs from dedicated overhead 3D people counting sensors, which are purpose-built counting devices: an AI camera delivers exact in/out and occupancy on standard camera hardware.
The same privacy-first engineering that makes GDPR-compliant people counting possible runs every method we deliver. Bumbee Labs is the company behind the only footfall method in Europe approved by a data protection authority.
Can the cameras we already have do the counting?
Yes. Cameras from Hikvision, Axis and Hanwha are integrated, so a venue’s existing cameras from those makers can be used for the counting. That makes an AI people counting camera practical wherever camera infrastructure already exists or is planned, and the counts reach the same dashboards and API as data from every other method.
Where do AI cameras fit best?
AI cameras fit best where standard camera hardware is itself an advantage: sites that already run or plan camera infrastructure, checkout and service zones where queue metrics drive staffing, and organisations standardising on one hardware family across many locations. The cameras can also count vehicles. 3D sensors remain the specialist tool for classic overhead doorway precision. For camera-free measurement in dark, outdoor or wide-open spaces, LiDAR gives the same precision without any imagery at all. The LiDAR vs camera people counting guide compares the two. A hybrid setup lets you combine them on one platform. If you already run Xovis 3D sensors, we process that data too.
How does an occupancy camera work?
An occupancy camera is an AI people counting camera used to show how many people are inside a zone. Directional in/out counts at each entrance and exit give current occupancy as entries minus exits, and you can hold that figure against safe capacity limits.
The figure is anonymous and aggregate: it says how many people are inside and nothing about who they are. It goes to the same dashboards and API as every other metric. For what venues do with occupancy beyond capacity compliance, from staffing to crowd comfort, see occupancy and crowd comfort.
How do AI people counting cameras help in retail?
In retail, AI people counting cameras measure the entrance and the checkouts at the same time. At the entrance, exact directional counts anchor the in-store conversion rate: entries against transactions, store by store. At checkouts and service points, queue length and waiting-time indicators show when queues are building. Those metrics drive staffing at the entrance and the checkouts.
Because AI camera data lands in the same platform as Wi-Fi, 3D sensor and LiDAR data, a chain can add camera precision at the doors and service points that need it and keep identical metrics across the estate. Retail footfall analytics covers the rest of the retail picture, from capture rate to store-to-store benchmarking.
Curious what this looks like for your venue?
AI cameras vs 3D sensors vs LiDAR
All three are precision methods, each suited to different conditions. AI cameras use standard hardware, 3D sensors are purpose-built for counting, and LiDAR works without a camera at all.
| Capability | AI camera | 3D sensors | LiDAR |
|---|---|---|---|
| Exact in/out at a door | Yes | Yes | Yes |
| Real-time occupancy | Yes | Yes | Yes |
| Queue metrics at service points | Yes | Yes | Yes |
| Standard camera hardware | Yes | No | No |
| Works in total darkness | Partial | Partial | Yes |
| Built to be image-free | No | Partial | Yes |
| Anonymous counting output | Yes | Yes | Yes |
| Optional video security extension | Yes | No | No |
- Full
- Partial
- Not available
The same cameras for counting and video security
Counting is anonymous and aggregate, and it produces numbers, never footage. When you want more from the hardware, the same cameras can be extended with video security, under your control and your rules.
Having the large flows we have is a challenge. We constantly strive to have as efficient a station as possible. With the help of reliable data from the new measurement system, we can better plan where different service functions or stores are to be located and how we can adapt doors or passages.
Frequently asked questions
How is AI camera counting different from 3D sensor counting?
A 3D sensor is a purpose-built counting device that uses stereo vision or time-of-flight and is typically mounted overhead at a door. An AI camera is a standard network camera with a counting model built in. That makes it practical where camera infrastructure already exists or is planned, and one camera can cover wider angles than a dedicated overhead sensor. Both deliver exact in/out and occupancy into the same platform.
Does an AI people counting camera store images of people?
Not in counting mode. The camera analyses video in the moment and outputs numbers; no personal images are stored and no individuals are identified. If you separately choose a video security extension, recording is governed by your own security policy and legal basis, and it remains clearly separated from the counting analytics.
Is AI camera people counting GDPR-compliant?
It can be, when the system is built to count people without identifying them, which is how these cameras run by default. In counting mode the video is analysed in the moment and only numbers leave the camera. No personal images are stored and no individuals are identified, so what you receive is an anonymous, aggregate count. The same privacy-first engineering runs every method we deliver. Bumbee Labs is the company behind the only footfall method in Europe approved by a data protection authority.
Can AI people counting cameras do more than count?
Yes. Because they are standard cameras, the hardware can grow with you. On request, the system can be extended with video security, which you configure and which your own security policy and legal basis govern. It is kept entirely separate from the anonymous counting analytics. Counting stays anonymous either way, and you decide whether the same cameras take on both jobs.
When are AI cameras the right choice over LiDAR or 3D sensors?
Choose AI cameras where standard camera hardware is an advantage: existing camera points, service and checkout zones, or sites standardising on one hardware family. Choose LiDAR for dark, outdoor or wide-open areas where image-free measurement matters most, and 3D sensors for classic high-precision doorway counting. Many venues combine them through the same platform.