University of Twente Student Theses


Creation and evaluation of a database for automatic video-based pain detection

Kiel, L. (2021) Creation and evaluation of a database for automatic video-based pain detection.

[img] PDF
Abstract:Pain management can considerably influence the recovery of a patient. A system for automatic pain detection can therefore not only support health practitioners in their work but also improve the patient’s well-being. Previous work on automatic painful facial expression recognition has been successful in discriminating painless from painful expressions. In this study, some of the limitations of these studies will be addressed by collecting a video database with continuous self-ratings of pain which enables to use self-ratings as ground truth at video frame level. Using machine learning algorithms, facial features that represent changes in the face due to expressions are ranked by their importance in discriminating non-painful from painful expressions. Different subsets of features are used to analyse which might be the most promising one in recognising painful facial expressions. The results show that the model performs well on new data if it was trained on data from the same participants, but on a chance level if a between-subjects design was used. Further, the model benefitted from excluding participants that indicated high levels of tiredness, which indicates that considering tiredness as a confounding variable could be important for future research about automatic painful facial expression.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:77 psychology
Programme:Psychology MSc (66604)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page