If you have searched for a "3D people counter," you have probably found a mix of stereo-vision cameras, depth sensors, and a lot of vendors using the term loosely. This explains what 3D people counting actually is, how a system builds depth, where the extra dimension genuinely helps, and where it does not justify its cost. It is written to inform the decision, not to sell you one specific box.
What is 3D people counting?
3D people counting uses depth, not just a flat image, to count people. Where a 2D camera sees a top-down picture and infers heads, a 3D system measures the distance to every point in the scene, so a person is a shape with height and volume rather than a pattern of pixels. That depth makes it harder to fool with shadows, reflections, or printed images, and easier to separate two people standing close together. The two common ways to get depth are stereo vision (two lenses) and time-of-flight (timing a light pulse).
2D vs 3D counting: what changes when you add depth
A 2D counter works from a flat image and a model that decides which blobs are heads. It is cheap and it works in clean conditions. It struggles the moment the scene gets messy: a shadow stretching across the floor, sunlight through a door, a tall display that reads as a person, two shoppers whose silhouettes merge.
Adding depth changes the unit of measurement. The system now knows that the floor is two and a half metres below the sensor and a head is roughly a metre and a half below it. A shadow has no height, so it is rejected. A cardboard cutout is flat, so it is rejected. Two people produce two separate volumes even when their outlines overlap from above. The count holds up precisely in the conditions that break a flat camera.
How 3D systems get depth: stereo vision vs time-of-flight
There are two mainstream ways to measure depth, and they behave differently.
Stereo vision uses two lenses a fixed distance apart and compares the two images, the way your eyes judge distance. It produces a detailed depth map and is well proven in retail. The catch is that it is still a camera capturing images, with the privacy and compliance questions that follow.
Time-of-Flight (ToF) fires a light pulse, usually infrared, and times how long it takes to bounce back from each point in the scene. The output is a depth map of shapes, not a photograph. Because ToF provides its own light, it works in the dark, and because it records geometry rather than imagery, it captures no faces.
Both give you depth. They differ on what else they capture along the way, which is the point the privacy section returns to.
Where 3D counting is strong, and where it struggles
3D earns its premium in specific situations:
- Groups and crowds. Separating people standing close together is the headline benefit. If your doors see families, couples, or queues, depth pays for itself in accuracy.
- Difficult light. Depth methods, ToF especially, hold up where a flat camera fails.
- Anti-spoofing. A flat image of a person, a poster, or a reflection has no depth, so it is rejected.
It is not free, and it is not always the right tool:
- Cost. A 3D sensor costs more than a beam or a basic camera, and on a quiet single-person doorway the extra accuracy may not be worth it.
- Mounting and ceiling height. Depth sensors have a working range. Very high ceilings or unusual mounting can reduce the benefit.
- Occlusion across a wide space. A single overhead 3D sensor covers a defined zone. Counting movement across a large open interior is a different problem that depth at one point does not solve on its own.
3D vs camera-free counting: a privacy note
Here is the distinction worth carrying into a buying decision. "3D" tells you the system measures depth. It does not tell you whether the system captures images of the people it counts. A stereo-vision 3D counter does. A time-of-flight depth sensor does not, because it records geometry rather than a picture.
This is where Ariadne takes a different route from the stereo-camera systems most people picture when they hear "3D people counter." Ariadne does not sell a 3D people counter. It counts with Time-of-Flight depth sensing for geometry at the entrances and patented phone signal sensing for movement through the interior, fused in the Ariadne platform, with no camera involved and no identifier captured by default. If your reason for looking at 3D is accuracy on groups without putting cameras over your doors, a camera-free depth-plus-signal method reaches the same goal from a different direction. See biometric vs non-biometric counting for what each method records about a visitor.
How to evaluate a 3D counter
Whatever method you choose, the same checks apply:
- Ask for an accuracy figure measured on your own site at peak, against a manual count, not a lab number.
- Confirm what the sensor captures: images, or only depth geometry. This decides your compliance workload.
- Check the mounting range against your ceiling height and door width before you commit.
- Confirm what data leaves the building and where it is stored.
FAQ
Is a 3D people counter more accurate than a 2D one?
Usually yes in hard conditions: groups, crowds, and difficult light. On a quiet single-person doorway the difference may be small.
Does a 3D people counter use a camera?
It depends on the method. Stereo-vision 3D counters capture images. Time-of-Flight depth sensors capture geometry, not pictures, so they do not work like a camera.
Does 3D people counting work in the dark?
Time-of-Flight does, because it supplies its own infrared light. Stereo vision, like any camera, needs adequate light.
Is 3D people counting GDPR compliant?
The method does not decide that on its own. What matters is whether the system captures personal data. A depth-only or camera-free system captures none, which simplifies compliance.
What is the difference between 3D counting and time-of-flight?
Time-of-Flight is one way to produce 3D depth data. "3D" is the broader category; ToF and stereo vision are two methods inside it.
---
