Automated Pain Detection from Video Monitoring of Facial Expressions
The SHORE system of Fraunhofer IIS is a highly performant system for detection of faces and analysis of facial expression with successful applications in consumer research. In this pilot project, it has been investigated how SHORE can be applied to the identification of facial expressions of pain in a clinical context. The Physiological Psychology group provided their expertise in psychological experimental research of determinants in decoding and encoding facial expressions of pain. The Cognitive Systems group provided their expertise in standard machine learning as well as in high-level symbolical learning on structural data (e.g. ILP, grammar learning) to learn classifiers for sequences of Action Units (AUs).
Partners
- Jens Garbas, Fraunhofer IIS, Intelligent Systems, Erlangen
- Miriam Miriam Kunz, Medical Psychology, University of Groningen, NL
- Stefan Lautenbacher, Physiological Psychology, University of Bamberg
- Ute Schmid, Cognitive Systems, University of Bamberg
Funding
Fraunhofer IIS, 2015-2016
Publication
Siebers, M., Schmid, U., Seuß, D., Kunz, M., Lautenbacher, S. (2016). Characterizing Facial Expressions By Grammars of Action Unit Sequences -- A First Investigation Using ABL. Information Sciences. Vol 329, 866-875 (doi:10.1016/j.ins.2015.10.007)