eMotion: un outil pour personnaliser la reconnaissance d'émotions

Abstract

A subject's emotional expression is of course influenced by his personality, but several other factors can play a role. His position, the constraints he undergoes, or the setting of his working space can all deeply impact emotional expression through movement. When performing an automatic emotion recognition, an evaluator must be able to set up some parameters for the recognition so as to adapt to the evaluation conditions. Current recognition system do not allow for such adaptability. In this paper we present the eMotion software for movement-based emotion recognition. eMotion's software architecture allows easy integration of graphical widgets for parametering emotional features extraction to suit the task at hand.