Fever screen update: configurable thresholds and more

The team of engineers at The Cacophony Project have been busy working on the software on the thermal camera to improve its functionality.

If your camera is connected to the internet then the software will be automatically updated over the next day or so to version 0.2.1. (you can check your version by clicking on the settings cog on the top right). Here's what is included in the update.

Adjustable fever threshold

The main new functionality we are adding is the ability for you to set the temperature threshold for a fever. At any threshold you will have some false positives: healthy people marked as having a high temperature and some false negatives: febrile people (with a fever) marked as having a normal temperature. The default threshold is 37.4°. You can now change this. If you lower it, you will need to do more manual screening, if you increase it then you may let through more people that have a fever.

Image taken from this paper

To change these thresholds click on the settings cog on the top right. Enter the minimum temperature for fever  and the minimum temperature for normal and press ok.

Cold people

We have also added in a new notification to show when a person is below the normal temperature range. The reason we have added this is because our customers have found that often, if people are coming in from the cold, they were not registering as being within the normal temperature and the screen was showing empty. We think this is more clear - showing the camera as having detected the person and they are below normal temperature.

The question is what should you do when someone is marked as being cold? 

The most robust solution is to get people to acclimatise to room temperature. The research indicates that waiting 10 minutes should be enough. However this may not be practical, depending on your situation.

An alternative is to have a camera dedicated for people coming in from the cold and to adjust the fever temperature threshold to a lower level. A way of determining what this level should be is to do the following:

  • measure the temperature a person immediate when they come in from the cold (e.g. 35.3°)
  • measure the temperature of that same person after they have had a chance to acclimatise (e.g. 36.8°)
  • Drop the fever threshold by the difference in these two temperatures (e.g. 36.8-35.3 = 1.5, so set the threshold to 35.9°)
This alternative is not as robust as having people wait, but it may be a practical compromise. We are investigating whether we can automate this by measuring the average temperature of the rest of the person.

Accuracy and drift improvements

We have made further improvements to the accuracy of the temperature and to our algorithms which will mean that the temperature should not drift as much as it was. This means that you should need to calibrate less often.

Source temperature

In the last update we had included the ability to specify the calibration source as being the armpit, ear, forehead or oral. However we weren't doing anything with that information. We are now using this to adjust the calibration as follows:
 Source Normal (midpoint) Adjustment
Forehead 36.55°
Oral 37° -0.45°
Ear (tympanic) 37.45° -0.9°
Armpit 36.55° 0°

Messaging, sounds & cropping

  • We have added a message that will show when the software has updated (this message won't show until the next update after this one).
  • We have also added a warning message that the camera may be unstable that shows for 30 minutes after the camera is rebooted.
  • Sounds now work on iOS devices
  • Any cropping you have done is stored when you turn the camera off and on.


The manual and quick start guide have been updated to reflect these changes.

Leave a comment

Please note, comments must be approved before they are published