A people counter is a device or software that counts how many people enter, leave, or occupy a space, using a sensor such as an infrared beam, a thermal or 3D overhead sensor, or a camera with video analytics. The right type depends on what you need the number for: a monthly footfall report, a live occupancy limit, or a count that follows a specific rule. This guide compares the main types, their trade-offs, and when counting from the cameras you already own is enough. For Argu’s own approach, see the people counting software page.
Key takeaways
- 4 sensor types dominate: infrared beam, thermal, overhead 3D stereo, and video analytics from standard cameras.
- Beam counters are cheap but cannot tell 2 people walking side by side from 1. Overhead sensors are accurate but need 1 device mounted over every entrance.
- Video-based counting can reuse IP cameras on RTSP or ONVIF Profile S streams at 1080p to 4K and 15 to 30 fps, so no new hardware is needed.
- Live occupancy can be compared with the occupant load limit set under life-safety codes such as NFPA 101.
- Counting people is personal-data processing in many places. Decide retention and notice before you start.
What Are the Main Types of People Counter?
The main types of people counter are infrared beam, thermal, overhead 3D stereo, and video analytics. Each answers the same question, how many, with a different balance of cost, accuracy, and flexibility.
Infrared beam counters
A beam counter uses a transmitter and a receiver on either side of a doorway. Each time the beam breaks, the count rises by 1. It is the cheapest option and works for a narrow single door. The limit is obvious: two people walking together break the beam once, and a person standing in the doorway can be counted repeatedly.
Thermal sensors
A thermal sensor reads body heat from above and does not capture images. That helps where privacy rules are strict or lighting is poor. It still counts crossings of a line and still needs a device over each entrance.
Overhead 3D stereo sensors
An overhead stereo sensor uses two lenses to build a depth map and tracks people crossing a virtual line. It is the traditional answer for retail and transit where accuracy matters, and it can separate adults from children by height. The cost is hardware at every counted door, with cabling and calibration.
Video analytics on existing cameras
Video analytics counts from a standard IP camera feed using a trained model. If cameras already cover the entrance, no new sensor is needed. Quality depends on camera angle, lighting, and the model. This is where an AI agent can go beyond line crossing, for example counting a group as one visit or counting only people matching a described rule.
People Counter Types Compared
| Type | Hardware per door | Handles groups | Captures images | Best for |
|---|---|---|---|---|
| Infrared beam | 2 units | No | No | Narrow single doors |
| Thermal | 1 overhead unit | Partly | No | Privacy-sensitive sites |
| Overhead 3D stereo | 1 overhead unit | Yes | Depth only | High-accuracy footfall |
| Video analytics | 0 if cameras exist | Yes, rule-based | Video feed | Reusing installed cameras |
How Do You Choose a People Counter?
Decide what the number is for
A monthly footfall report tolerates small errors. A live occupancy limit does not. If a fire-code number is at stake, accuracy at the entrance matters more than price. For the difference between counting and reading space usage, see our guide to occupancy analytics.
Count the entrances
A store with 1 door and a warehouse with 12 gates have different economics. A sensor at every door multiplies cost and maintenance. Reusing installed cameras avoids that, if their angle and image quality are good enough.
Check the camera basics
For video counting, ask for the minimum resolution, frame rate, and mounting height the vendor supports. Streams from IP cameras and NVRs over RTSP or ONVIF Profile S, at 1080p to 4K and 15 to 30 frames per second, are the common range. Cameras that look straight along a corridor are harder to count from than cameras that look down at a door. Our overview of AI cameras explains what the model sees.
Plan for privacy
Counts without images are lighter on privacy than video. Where video is used, set a retention window, post a notice, and limit who can review footage. Individual-level queries may fall under data-access rights such as GDPR Article 15 where personal data is involved.
Where Counting Fits in the Bigger Picture
Counting is useful by itself, and more useful next to other signals. A store manager who sees footfall alongside wait time can tell whether a quiet afternoon came from fewer visitors or from visitors leaving long lines. A venue operator can pair live occupancy with entry control. For retail teams, our retail security cameras guide covers where cameras should sit.
Argu’s people counting agent works from rules written in plain language, not from fixed hardware configuration. A group of 3 can be counted as 1 visit if that is the number the business needs. A query such as how many people in a particular type of clothing entered can be answered directly from the same feed. A new counting rule can go live in about 30 seconds, according to Argu’s published agent-creation time. Counting-rule records can be retained on an operator-set 30 to 90 day window.
Installation Tips for Any People Counter
Mount for a clear view of the doorway
Overhead sensors and cameras count best when they look straight down at a door, with no hanging signs, plants, or banners in the line of sight. Check the height range the vendor supports before you drill.
Validate with a manual count
After installation, stand at the entrance for 30 minutes at a busy hour and count by hand, then compare with the system. Repeat at a quiet hour and in different lighting. Record the difference and ask the vendor to tune the system if it exceeds your tolerance.
Review the numbers regularly
Counts drift when furniture moves, a door is repainted, or a camera is nudged. A monthly check against a short manual sample keeps the numbers trustworthy.
Frequently Asked Questions
What is a people counter?
A people counter is a sensor or software that counts the people entering, leaving, or present in a space. It supports footfall reports, staffing decisions, and occupancy limits.
How accurate are people counters?
Accuracy depends on the type and installation. Beam counters struggle with groups, while overhead sensors and well-placed video analytics handle groups better. Ask any vendor to demonstrate accuracy on your own entrance.
Can I use my existing CCTV cameras as a people counter?
Often yes. If the cameras cover the entrance from a suitable angle with enough resolution, video analytics can count from the existing feed without a new sensor.
Do people counters record faces?
Not necessarily. Beam, thermal, and depth-only sensors do not capture identifying images. Video analytics works from a video feed, so retention and access rules matter.
Is people counting allowed under privacy law?
Rules vary by country and by whether the system stores identifying data. Check local requirements, post a notice, and set a retention period. Seek legal advice for your site.
What is the difference between people counting and occupancy?
Counting tracks crossings at a door. Occupancy reports how many people are in a space right now, which is the number you compare with a capacity limit.
A people counter turns entrances into numbers. Argu does it from the cameras you already own, with rules you describe in plain language. Talk to Argu about the entrances where you need a reliable count.
Last updated: October 2026
