Minimum Annotation Identification of Facial Affects for Video Advertisement

Consumer's market has witnessed a radical change. It has shifted from being a seller's market to buyer's market. Reviews and feedback play an important role deciding the marketing strategy. These reviews express opinions or sentiments of a particular individual for a particular product, which contains information that is useful for others who are interested in buying that product. The paper proposes a model to automate the product review for an advertisement by making use of affective computing. The computerized model for Facial emotion recognition using wearable biometric sensors is used analyze the consumer behavior on a video advertisement. The model detects five basic emotions of happy, sad, surprise, anger and neutral. The logistic regression classification algorithm was used to classify the various sentiments given by various users against the advertisement on the given test data. The result of the classification algorithm has been compared with training data using WEKA. The accuracy of 82% is obtained through classification algorithm.