QR code or facial recognition: which fits better?
06.09.2026
QR code or facial recognition - this comparison has become more common as facial recognition terminals have grown cheaper. The technology looks impressive: the employee looks at the device and the record is made. In practical deployment, however, three issues emerge: sensitivity to conditions, data responsibility and the number of points. This guide compares the two approaches from the standpoint of real use.
How each works
Facial recognition: the terminal reads the employee's facial features, matches them against a stored template and, on a match, records the check-in. The recording point is the device.
QR system: a code belonging to the point is scanned with the employee's phone and the record is written by the system. The recording point is a logical concept.
Comparison
| Criterion | Facial recognition | QR code |
|---|---|---|
| Device | A terminal per point | Not required |
| Contactless | Yes | Yes |
| Lighting | Affects it | Does not affect it |
| Mask, glasses, helmet | Can cause problems | No problem |
| Personal data | High responsibility | Minimal |
| Field work | Not suitable | Suitable |
| Transfer risk | None | Managed |
| Cost of a new point | A new device | None |
The strength of facial recognition
- Contactless and fast. The employee stands still and the record is made.
- Transfer is impossible. Unlike a card or a code, it is bound to the person.
- No phone required. An advantage for a workforce without smartphones.
- Independent of hand condition. A significant difference from fingerprint.
Practical limitations
- Lighting. Direct sunlight or dim light reduces recognition accuracy. For terminals installed outdoors this is a permanent problem.
- Workwear. Helmets, masks and safety glasses are the norm in industrial settings - each affects recognition.
- Number of points. Every entrance means a separate terminal; as sites multiply the cost rises linearly.
- Data responsibility. A face template is biometric data; collecting it requires consent and protective measures.
A practical example
At a logistics centre a facial recognition terminal was installed at the main entrance and worked well. The problem started in the warehouse area: employees there used protective equipment and recognition frequently failed.
On top of that the warehouse had three separate entrances. The budget for a terminal at each was not approved, so employees walked back to the main entrance to check in - which did not reflect their real movement.
The solution was hybrid: the terminal stayed at the main entrance while QR points were created at the warehouse's three entrances. As a result the distribution of hours across areas became visible for the first time.
Privacy and regulation
Collecting face data must be accompanied by employee consent, a defined retention period and restricted access. For the company this is an ongoing obligation. In a QR system the stored data consists of the moment of the record and the point - the responsibility is far lighter.
The cost view
At one or two fixed points facial recognition can be reasonable. With more than three points, or with changing sites, the QR approach becomes considerably more economical - because the cost of an additional point is practically zero.
Which for which situation
- One fixed office entrance, stable workforce, no phones → facial recognition.
- Many points, field work, protective clothing → QR.
- Mixed conditions → a hybrid model.
Measuring recognition accuracy in real conditions
The datasheet figures for facial recognition terminals are obtained in laboratory conditions: stable lighting, an uncovered face, the correct distance. A real working environment differs.
Three scenarios should be tested during the trial: the morning shift (direct sun or dim light), work in protective clothing (helmet, mask, safety glasses), and the rush at shift start with several people in sequence.
The figure to measure is simple: out of a hundred check-in attempts, how many succeeded on the first try. If it is below ninety per cent, in practice that means dozens of repeat attempts and a queue every day.
In a code-based system the reading takes place on the employee's device and does not depend on these factors.
The legal status of the data
A face template is biometric data and collecting it creates an ongoing obligation. That obligation continues after the rollout and covers five elements.
The first is obtaining clear consent from the employee. The second is defining a retention period. The third is restricting and logging access to the data. The fourth is a deletion procedure when an employee leaves. The fifth is the availability of an alternative method for those who do not consent.
Without the last item the rollout becomes a practical problem: records for those who refuse are kept by hand and part of the record stays imprecise.
Managing these obligations requires resource and should be counted as part of the cost when deciding.
A practical route to the decision
A three-step approach works for deciding between the two options.
Step one: count the points. One or two fixed entrances - the terminal option is reasonable; more than three, or a changing structure - the code-based approach is more economical.
Step two: check the conditions. If workwear, lighting and outdoor factors are present, the real accuracy of facial recognition must be tested.
Step three: assess legal readiness. If there is no internal rule for managing biometric data, preparing one is part of the project.
The result of the three steps often leads to a hybrid model, and that is a normal outcome - conditions in real companies are never uniform.
The long-term cost comparison
Facial recognition terminals have become cheaper in recent years, which has changed the comparison. But the device price alone is not enough for a decision.
The lines that belong in a three-year calculation are: devices, installation and electrical work, adjusting the lighting, spare devices, maintenance, software updates, the time spent managing templates, and additional devices for new points.
To that must be added the administrative load of managing biometric data - consent documents, retention rules, deletion procedures.
In the code-based approach almost all of those lines are zero.
The practical conclusion: at one fixed point the difference may be small, but as the number of points grows the difference widens rapidly and becomes the deciding factor.
In summary
Facial recognition is contactless and fast, ties the record to the person and requires no phone. Its weaknesses are sensitivity to conditions, the cost per point and an ongoing obligation around biometric data.
The code-based approach removes those three weaknesses and requires a smartphone in exchange. Binding the record to a person is handled differently here: when device control, alerts on device changes and location matching work together, the practical gap narrows considerably.
The practical decision follows three steps: count the points, measure recognition accuracy in real conditions, and assess legal readiness. The result is often a hybrid structure.
The next step
Testing in real conditions before choosing is essential - particularly in areas with difficult lighting and workwear. Catalogue figures and the real environment often differ.
QRGate provides recording that does not depend on site conditions. See what QRGate can do or calculate the price.
