Modeling human affective postures:an information theoretic characterization of posture features

Abstract

One of the challenging issues in affective computing is to give a machine the ability torecognize the mood of a person. Efforts in that direction have mainly focused on facial andoral cues. Gestures have been recently considered as well, but with less success. Our aim is tofill this gap by identifying and measuring the saliency of posture features that play a role inaffective expression. As a case study, we collected affective gestures from human subjectsusing a motion capture system. We first described these gestures with spatial features, assuggested in studies on dance. Through standard statistical techniques, we verified that therewas a statistically significant correlation between the emotion intended by the actingsubjects, and the emotion perceived by the observers. We used Discriminant Analysis tobuild affective posture predictive models and to measure the saliency of the proposed set ofposture features in discriminating between 4 basic emotional states: angry, fear, happy, andsad. An information theoretic characterization of the models shows that the set of featuresdiscriminates well between emotions, and also that the models built over-perform thehuman observers. One of the challenging issues in affective computing is to give a machine the ability torecognize the mood of a person. Efforts in that direction have mainly focused on facial andoral cues. Gestures have been recently considered as well, but with less success. Our aim is tofill this gap by identifying and measuring the saliency of posture features that play a role inaffective expression. As a case study, we collected affective gestures from human subjectsusing a motion capture system. We first described these gestures with spatial features, assuggested in studies on dance. Through standard statistical techniques, we verified that therewas a statistically significant correlation between the emotion intended by the actingsubjects, and the emotion perceived by the observers. We used Discriminant Analysis tobuild affective posture predictive models and to measure the saliency of the proposed set ofposture features in discriminating between 4 basic emotional states: angry, fear, happy, andsad. An information theoretic characterization of the models shows that the set of featuresdiscriminates well between emotions, and also that the models built over-perform thehuman observers.