A time series feature of variability to detect two types of boredom from motion capture of the head and shoulders.

Authors: Witchel, H., Westling, C., Tee, J., Needham, R., Healy, A. and Chockalignam, N.

Start date: 1 September 2014

Journal: ACM Transactions on Multimedia Computing, Communications & Applications

Publisher: Association for Computing Machinery (ACM)

ISBN: 978-1-4503-2874-6

ISSN: 1551-6857

DOI: 10.1145/2637248.2743000

Boredom and disengagement metrics are key to accurately timed adaptive interventions in interactive systems. Psychological research suggests that boredom is a composite state incorporating cycles of lethargy and restlessness. Here we present innovative metrics of the components of boredom, based on motion capture and video analysis of head and shoulder movement. Healthy seated volunteers interacted with discrete, screen-presented stimuli ranging from engaging to boring, using a handheld trackball rather than a mouse, to allow for uninhibited noninstrumental shoulder movements. Our results include a feature (standard deviation of windowed ranges) potentially suitable for implementation in computer vision algorithms for early detection of disengagement.

The data on this page was last updated at 05:32 on April 17, 2021.