Detecting depression using multimodal approach of emotion recognition

Depression is a growing problem in our society. It causes pain and suffering not only to patients but also to those who care about them. This paper presents a multimodal emotion recognition system that is capable of preventing depression. It consists of detecting persistent negative emotions for early detection of depression. Our proposal is based on an algebraic representation of emotional states using multidimensional vectors. This algebraic model provides powerful mathematical tools for the analysis and the processing of emotions and permits the fusion of complementary information such as facial expression, voice, physiological signals, etc. Experiments results show the efficiency of the proposed method in detecting negative emotions by giving high recognition rate.