Skip to main content
Login | Suomeksi | På svenska | In English

Cognitive Load Sensing with FLIR Thermal Cameras and the Effects of Calibration Correction

Show full item record

Title: Cognitive Load Sensing with FLIR Thermal Cameras and the Effects of Calibration Correction
Author(s): Malmivirta, Titti
Contributor: University of Helsinki, Faculty of Science, Tietojenkäsittelytieteen osasto
Discipline: Tietojenkäsittelytiede
Language: English
Acceptance year: 2020
Continuous thermal imaging is a way to measure psycho-physiological signals in humans. Psycho-physiological signals refer to physical signals caused by some psychological or mental situation or change. One of the technical challenges with the thermal psycho-physiological signal measurement is that the devices used to measure the temperature changes need to be sensitive and accurate enough to actually detect them. This is generally true for laboratory equipment, but the current relatively cheap and small mobile devices, including uncooled thermal cameras, could be used to make this kind of measurements cheaper, less intrusive and mobile. Currently the customer-priced mobile devices still tend to produce a lot of noise and other inaccuracies to measurements. The focus of this thesis is to evaluate the usefulness of the FLIR One thermal camera integrated in the Caterpillar Cat S60 phone for cognitive load measurement and the possibility to improve the measurement accuracy with additional calibration correction. We developed a deep learning based calibration correction method as an attempt to improve the quite noisy initial measurements of the thermal camera. Then an experiment measuring cognitive load was organised. The calibration correction method was used to reduce errors in the data from the cognitive load experiment to see if the performance of the thermal camera can be improved enough for accurate cognitive load detection. Our results show that while our calibration correction method does improve the measurement accuracy when compared to the ground truth, the fluctuations in the measurements do not decrease enough to improve the performance of the thermal camera with regards to the cognitive load sensing.
Keyword(s): Thermal Imaging Machine Learning Mobile Sensing Mobile Computing

Files in this item

Files Size Format View
Titti_Malmivirta_tutkielma_2020.pdf 1.639Mb PDF

This item appears in the following Collection(s)

Show full item record