Towards Detection of Side Activities and Emotions of Anonymous TV Viewers through Body Postures

Abstract

TV viewers often perform various types of side-activities, including usage of mobile devices to browse, chat, e-mail or use social networks. These activities can be impacted by or related to TV content. In order to understand the possibility to unobtrusively monitor emotions of TV viewers, we want to get a better understanding of their body postures and side activities and how these relate to emotions.In this paper we explore the role and interconnections between body postures, side-activities and emotions of people watching TV through a questionnaire and a user study.We found that postures do not directly relate to emotions perceived from emotion eliciting video clips. Analysis of emotions and side-activities did reveal some interesting correlations between device usage and the experienced emotions, which can be useful to inform the design of applications on or in relation with the TV. A first user study showed that a subset of the relevant side-activities could be tracked based on posture data, thus enabling activity recognition and maybe even emotion estimation without the need for personal characteristics.