Coming soon: automatic calibration and face tracking
This is a quick update to let you know what we're working on. This functionality isn't available at the moment but is on its way.
Right now, to achieve the accuracy we need our cameras are calibrated by taking the temperature of a person in front of the camera and typing that into the calibration screen. The calibration can drift particularly when the room temperature changes resulting in the need for frequent calibration. Although this is fairly simple to do it means you need to have someone nearby who is trained to re-calibrate the camera. The calibration is also reliant on the accuracy of the thermometer used to take the temperature of the person.
The solution to this is to have a constant heat source that is in field of view of the camera and used to calibrate the camera constantly. The University of Canterbury has developed a constant heat source on our behalf. We've confirmed that we can use this and are working to integrate it into the camera. This will significantly improve the usability and accuracy of the camera.
This will be made available at no additional cost to those that have already purchased cameras.
We have implemented face tracking - to automatically find a face in the image. This will be used for a number of purposes:
- To help ensure we are measuring the forehead. At the moment the camera locks onto the hottest spot it can identify. Sometimes this is the forehead, neck or arm.
- Help with our reporting - where we will provide reports on the number of people scanned.
- Help us collect data for researching approaches for identifying people that have come in from the cold.
- Automatically ignore other hot spots like coffee cups or lights, reducing the need to crop the scanning area.
Note: we're not doing facial recognition but instead are just finding where a face is in the image.
Some of our customers are reporting that over 80% of people being screened are showing as having a colder than normal temperature. This frequently happens when people are coming in from being exposed to cooler temperatures outside. This is a problem with any thermal screening and the standard recommendation is to give people enough time to acclimatise (10 minutes). This isn't practical when you're screening customers or staff coming into a facility..
We are researching whether it is possible to detect when someone isn't acclimatised and have our camera adjust its estimate of their core body temperature. It's uncertain whether we'll find a satisfactory solution here, or how long this will take. The image below shows how a cold person's extremities and clothes show colder in a thermal image.
The development team is planning on settling into a rhythm of releasing software updates every two weeks. We will communicate via this blog and our newsletter about what has been released.