Emotion detection based on smartphone using user memory tasks and videos

Abstract

In this paper, we present a research study on the classification of emotions, through data gathered on a smartphone. To this end, we have developed a mobile application to elicit emotions in participants using memory tasks with success – failure manipulation and also using video clips. Interactions were recorded with accelerometer and gyroscope sensors records and keystroke on the device. We trained supervised classification models, with the records, to predict the nature of emotion elicited on two dimensions (pleasure and activation) and the success or failure related tasks memory tasks. In order to evaluate the emotion induction we have proposed a self-assessment procedure. We achieved interesting results on the pleasure dimension, by proposing a protocol with natural interactions on smartphone.