Posture annotation for studying affective interaction in multimodal corpora

Abstract

Although the relations between affects and posture were the focus of several studies in Psychology, their consideration in multimodal corpora research remains scarce. In this paper we survey related studies and propose a framework for studying affective postures by integrating knowledge coming from multiple sources such as literature, video corpora, motion capture data, artistic design, and perception studies. We believe that such a multi-source framework can be useful for integrating posture annotation in multimodal corpora research.