5  Working principles

FLIR Screen-EST system uses face detection and measures the temperature at the tear duct, since that has been demonstrated to be a robust measurement point. The system uses a relative screening method, comparing subject temperatures to a dynamic baseline of measured skin temperatures.

5.1  Face detection and measurement area

FLIR Screen-EST can use either the thermal data stream or the visual data stream for face detection and definition of measurement area.
Thermal detection
The system finds the face and locates the tear duct using FLIR thermal and AI technology. The thermal camera then measures the temperature at the tear duct.
Visual detection
The system finds the face using face detection on the visual data stream. A box representing the measurement area is drawn and adjusted to cover the face. The thermal camera finds and measures the hottest spot in the measurement box, which usually would be the tear duct.
    Thermal detection has the following advantages compared to visual detection:
  • Ensures measurement at the tear duct.
  • Works in any light condition, also when it is dark.
  • Makes it harder to fool the system, e.g. by faking a face with a photo.
If the ambient temperature at the screening station is high (not recommended), the visual detection may perform better. The thermal detection can have difficulty distinguishing faces from a warm background.

5.2  Relative screening method

The system uses a relative temperature screening method, with a moving average of measured skin temperatures as reference. This allows the screening algorithm to compensate for changes in conditions over time.
The measured temperature is compared to the sum of the moving average and a configurable allowed deviation. The elevated temperature alarm will trigger if the measured temperature is above the alarm limit. The alarm limit is the sum of the moving average and the allowed deviation.

5.3  Moving average

The moving average is calculated based on a number of previously registered temperature measurement samples. The number of samples is configurable.
At the beginning of a new screening session, there are no previous samples and the system must start to build up the baseline for the moving average. This is done as part of the normal screening procedure. For the very first screened person, there will be no analysis. The first measured temperature will always be added to the average calculation (if it is within the operating range, see section 5.4 Operating range ) and the screening result will always be normal (green).
To prevent an elevated temperature from being used as the starting reference, it is recommended to make sure the first screened person has a non-elevated temperature.

5.3.1  Reset of moving average

If there has been a long time since the last screening, the moving average may not be at the correct level. For example, after a long night the moving average may still be at a high level, based on the screenings from the afternoon or evening, but persons arriving in the morning are cold. Before the moving average has adapted to the new conditions, there is a risk for a false negative; that is, a Normal screening result even though the person actually has an elevated temperature.
One way of limiting the risk of false negatives is to reset the moving average after a long period of non-screening. The system will then start to build up a new moving average.
If the operator manually resets the moving average, the normal build-up procedure applies and there will be no analysis of the first measurement. The operator should make sure that the first screened person has a non-elevated temperature.
It is possible to enable a function that automatically resets the moving average after a specified time of non-screening. Since the moving average is reset without any operator interaction, there is no control over the first screened person. For that reason, a special build-up procedure applies after an automatic reset of the moving average. In this special case, also the first measurement will be analyzed and the alarm will trigger if the measured temperature is above the sum of a configurable fixed temperature and the allowed deviation.

5.4  Operating range

If the spread of values in a data set is large, the average will be difficult to use for evaluation of individual values. For that reason, the screening algorithm will only work with temperatures that are within a certain temperature range. If the measured temperature is outside this range, the system will display a “Too cold/warm” message and there will be no further analysis.
If the system displays a “Too cold/warm” message, the visitor should wait a few minutes to allow the skin temperature to acclimatize to the temperature at the screening station and then try the screening again.
The temperature range is defined by a maximum and minimum temperature limit, which is possible to change by settings. It is recommended to use the default values. The settings should only be changed if you experience problems and understand the consequences.

5.4.1  All visitors from cold environment

In a situation where all visitors come directly from a cold outdoor environment, without possibility to adapt to the indoor temperature, the screening flow may be disturbed by many “Too cold” messages.
By changing the minimum temperature setting to a lower value, there will be fewer “Too cold” messages. Since all visitors come from the same ambient temperature, the measured skin temperatures will still be comparable.

5.4.2  Mixed flow – both cold and warm

If the visitors come both from a cold outdoor environment and from a warm indoor environment, you need to be careful about changing the minimum temperature setting.
Persons entering a building from windy and snowy weather will have a lower skin temperature, regardless of their body temperature. The cold exposure could reduce the skin temperature of someone with an elevated temperature low enough to fall under the alarm limit.
If people coming directly from a cold environment are not stopped by a “Too cold” message, they may not have time to adapt their temperature to the indoor temperature before the screening algorithm performs the evaluation. This means there is a risk that people with an elevated temperature pass the screening, since the skin temperature of their cold faces may be below the alarm limit.
If people coming directly from a cold environment are allowed to adapt their temperature to the indoor temperature before the evaluation, only persons with a normal temperature will pass the screening and there will be a screening alarm for persons with an elevated temperature.