The role of emotion in movement segmentation

Abstract

Humans understand ongoing events by breaking reality into meaningful units through a process named "event segmentation". In psychology, theories such as the "Event Segmentation theory" have been proposed to illustrate how humans perceive the structure of ongoing behavior. Parsing discrete gestures from a continuous movement stream is also a necessary step for movement analysis. Many approaches towards automatic emotion recognition from full body movement leverage automatic segmentation methods. Nonetheless, to the best of my knowledge, no framework has applied Event Segmentation theory to the automatic segmentation of emotion conveying movements. In this paper, I propose the exploitation of a computational model of event segmentation to extend a movement-analysis framework with an event segmentation module.