The Various People Counting Methods In Public Transport

This post is also available in: Deutsch (German)

This article lists the many possible people counting options that public transport operators have for counting the number of passengers in public transport.

People counting systems in general

People counting is a system of people counters and a people counting software used for measuring the number of people passing through a certain location.

There are many different purposes for people counting, with the most commonly-discussed purpose being gathering and analysing customer traffic data for businesses to gain insights for sales, marketing and budgeting decisions.

For example, people counting sensors are placed strategically in brick-and-mortar retail stores to count every shopper who enters and leaves a store as well as their movements around the store. Such data compiled in a people counting software can be used to determine how well the store is performing.

People counting is also referred to as footfall counting, visitor counting, customer counting or door counting, but for this article, we will just refer to it as people counting since we aim to discuss how the number of people is recorded to report the occupancy levels in an area.

People counting for collecting public transport occupancy data

Seeing that transport hubs like train stations and airports are among the busiest places in the world, it makes sense to use people counting technology with the highest possible accuracy in public transport.

The issue of congestion in public transport, especially with the need for social distancing, has called for a data-driven solution. Knowing that occupancy data can help passengers follow social distancing rules in public transport, this article is focusing on the use of people counting for public transport occupancy data.

Many of the methods listed in this article are Automatic Passenger Counting (APC) systems, which, of course, automatically record the number of people in public transport vehicles.

Counting can happen at different occurrences:

  • entries (entering a vehicle or location)
  • exits (exiting a vehicle or location)
  • waits (waiting for a vehicle)
  • rides (inside the vehicle)
  • transaction logs (buying tickets or topping up ride points)
  • the whole journey (any combination of the above)

In different places:

  • vehicles (either by each carriage or the total of all carriages)
  • stops/platforms
  • stations
  • station gates

By different units of measurement:

  • number of individuals
  • average weight
  • duration of entries/exits
  • speed of entries/exits

And using different technologies:

Here is an evaluation of the technologies that can possibly be used to count people in public transport.

Facial recognition

Cameras with facial recognition capabilities powered by Artificial Intelligence (AI) capture sharp visuals of the facial expressions of passengers (maybe they’re cringing about the crowded vehicle).



People flow technology

In Japan, Hitachi and JR East work together to use people flow technology, with special surveillance cameras known as Eki-Shi-Vision, to capture and visually represent occupancy data in JR East’s transport networks and stations. Eki-Shi-Vision is built to only extract people movement data and visually represent people as generic icons to protect their privacy.


  • Not creepy – Any privacy issues associated with facial recognition are not applicable in people flow tech.
  • Still effective in capturing occupancy data visually.
  • No contact required.
  • Can be used in every possible corner of the station buildings and vehicles.


  • Cannot be used for passenger sentiment analysis as facial expressions are not captured.
  • Also cannot be used for security verification and contact tracing as individual facial recognition is omitted.

Body detection sensors

These sensors detect bodies via motion, pressure or weight.

  • Motion sensors count the number of times its light beam, which is not visible to the human eye, is crossed.
  • Pressure sensors placed in the seats detect body pressure and send signals to the dashboard when the seats are occupied.
  • Weight sensors measure the total weight of the vehicle and subtract the empty vehicle’s weight from this total to calculate the rough estimate of the train’s overall occupancy.



  • Complicated installation.
  • Only work in the locations they’re placed.
  • Less accurate as it is difficult to count large groups of people.
  • Cannot get sentiment insights.
  • For pressure sensors, only the seat occupancy is considered and not the passengers who are standing.
  • The cost of pressure sensors is relatively high because each seat must be equipped with a sensor.

Heat sensors

Heat or thermal sensors, usually placed on the ceiling, detect the body heat signatures of people walking underneath the sensors using thermal imaging tech, which converts infrared radiation into visible images.


  • Not creepy due to the anonymity of body heat counting.
  • Can be used for contact tracing of people with higher body temperatures.
  • Highly accurate.
  • Easy to use.
  • Not affected by lighting conditions.
  • Easily hidden from the passengers’ sight. (hehehe)
  • Might be useful for automatically adjusting the vehicle’s air-conditioning according to occupancy, which is important for sustainability.


  • Also cannot get sentiment insights.

QR Code scans

In case you were living under a rock, a Quick Response (QR) code, either printed or digitally shown, is scanned using a mobile device’s QR scanner, using either a QR scanning app or a relevant mobile app with an in-built scanner. Each QR code is linked to a data lake that compiles all the scans of that code.

The possibilities for QR code scanning for payments and attendance tracking are endless, and thus, can be considered as an upgrade to the smart card system (to be discussed next) in public transport.


  • Quick and easy to use (as its name, Quick Response, implies).
  • No contact required.
  • Currently in use for contact tracing, so it can distinguish between one-time scans by different devices and repeated scans by the same device.
  • Accurate with the assumption that every passenger has a mobile device with a QR code scanner.


  • More likely to lack anonymity if personal details are required.
  • Requires each passenger to own a mobile device, so it might not count those without their own mobile device like children.
  • Requires a QR code scanning app or an in-built QR code scanner in each device.

Smart card taps

A smart card is a card that contains a microchip which connects to a card reader either by tapping or using contactless RFID (radio frequency identification) tech. The microchip allows smart cards to store large amounts of data, run on-card functions like encryption and interact with the card reader.

The smart card data, that is the data about the entry and exit of each passenger captured at the station gates, can be referred to as the passenger counter data.


  • Not so creepy – Protects privacy and security due to its features including authentication, secure data storage, encryption, strong device security, secure communications, and biometrics.
  • Great memory thanks to its microchip.
  • Contactless smart card doesn’t require contact.
  • Straightforward people counting method, if every passenger is required to possess a smart card and the option to buy tickets is unavailable.
  • More convenient than buying tickets at the counter or ticket machine daily.


  • Contact required if the smart card has to be tapped.
  • Requires passengers to buy the smart cards.
  • Requires smart card readers at the station gates.
  • Might require passengers to top up the smart card balances at the counter or ticket machine if the online top up option is unavailable.
  • Might have to be cross-checked with ticket sales records if both smart card and ticket purchase options are available.
  • No sentiment analysis.

Ticket sales records

It is exactly what the heading says. Public transport operators can choose to check the data of the total number of tickets sold at their stations to determine the number of passengers in a certain network in a certain time period.


  • Less likely to be creepy as the ticket cashier might not record who purchased the tickets.


  • Requires passengers to purchase tickets at the counter or ticket machine every time they travel.
  • Contact is unavoidable when buying tickets.
  • Might have to be cross-checked with smart card data.
  • Might be tedious/messy/inconvenient to look at the ticket sales records.
  • Subject to human errors and biases.
  • No sentiment analysis if the cashier doesn’t care to remember the facial expressions or body language of the passenger at the counter.

Passenger surveys

A passenger survey directly asks the passengers to record their observations and experiences, either in a mobile app or an online form.


  • Accurate and straightforward if the passengers are honest.
  • Passengers’ feedback can be used for sentiment analysis.
  • Not creepy since the passengers give consent or the data can be made anonymous.
  • No contact required.
  • Can be used for contact required.


  • Less accurate if the passengers are dishonest.
  • Requires passengers to own a mobile device.
  • Requires mobile devices to have Internet connection.

Wi-Fi and Bluetooth connections

Wi-Fi and Bluetooth beacons count people by sending probe requests to mobile devices they detect nearby. A probe request is like, “Hello! Whoever here has switched on their Wi-Fi or Bluetooth toggles, will you connect with me?”. It’s basically a connection request, to which the device user can decide to accept or reject. The number of probe requests will be considered as the number of passengers. Another way to count using Wi-Fi is to count the devices that are already connected to the Wi-Fi.



  • It may not work on buses unless the buses are equipped with Wi-Fi or Bluetooth beacons.
  • Requires each passenger to own a mobile device, so it might not count those without their own mobile device like children.
  • Assumes that each person carries one mobile device.
  • Requires passengers with mobile devices to switch on their Wi-Fi or Bluetooth toggles.
  • Counting connected devices may be less accurate than counting probe requests because the chances of coming across already-connected devices is low compared to devices that are not connected.
  • If a mobile device gets the latest operating system update, its hardware ID will be changed, so it appears as a new device which then receives a new probe request. Although this might protect the user’s privacy, it will lower the accuracy of the tracking.

Manual counting by a counting professional/driver

This isn’t a technology but it’s still a method used by some public transport operators, whereby a human being like a counting professional or the vehicle driver manually counts the number of passengers entering the vehicle with a tally counter.


  • No privacy concerns as the person counting might not care to remember who the passengers are (unless a passenger who enters the vehicle is someone the counter knows).
  • No contact required as long as the counter keeps a distance from the passengers.
  • Might be able to read the passengers’ facial expressions or body language if the counter is able to do so while counting.


  • Less accurate.
  • Costly.
  • Might get bored, tired, distracted, annoyed or offended while counting.
  • Likely to make judgment calls or be biased.
  • Requires breaks.


A quick Google search for people counting systems shows that there are so many types of people counting systems out there. It also shows that this topic has been widely-studied for many years, even back in 1998. As we can see in the public transport context, the people counting methods can be categorised in two ways: Ticket-related and non-ticket-related.

Should public transport operators count based on ticket sales (tickets, QR code scans or smart card taps) or others (cameras, sensors, signals or employees)? Choosing which method works best for a public transport operator depends on what the operator is taking into account.

One of the things the operators need to consider is that there seems to be a trade-off between insights-driven decisions and privacy concerns. For example, the ability to analyse passengers’ sentiments by identifying who they are and how they behave is seen as creepy while prioritising anonymity means missing out on sentiment insights. At the end of the day, these decisions boil down to what the operator is prioritising.

[Note: Counting the number of people who search routes in an app or a website is not a good way to count passengers because not everyone who searches routes will use those routes, but instead, might just do it for informational purposes. That’s why this method wasn’t included in this article.]

Are you looking for ways to get the best out of your data?

If yes, then let us help you use your data.