An affect-based approach to detect collective sentiments of film audience: Analyzing emotions and attentions

Affective computing plays an important role in simulating human affects through multimedia stimuli. To provide appropriate responses and quantify emotions, it is essential to identify and interpret emotions accurately. Often the audience is influenced by media content and it has an impact on the audience's emotions. The authors aim to develop a machine vision based smart affective system that investigates the correlation between the basic facial expressions and corresponding ratings given by the audience. For the same, a notion of emotion vector is introduced to measure the elicited emotion. The idea of measuring attention is also suggested to find if the con- tent could keep the audience engaged. Our results reveal that such systems are useful in capturing and analyzing individual and collective sentiment.