CNN-Driven Novel Facial Expression Detection and Analysis

Emotion is defined as gesture that humans use to express their thoughts and feelings. Different ways of expressing emotions includes facial expressions, body language and voice tone. Human facial expressions have similar patterns, its tedious to recognize the facial impressions with naked eye. Therefore, Convolutional Neural Network, a Deep Learning technique is used in this research. There are 4 impressions of face which are to be identified here and they are laughter, crying, shame, and frustration. The dataset was created which consists total of 1760 images out of which 1600 were for training purposes and 160 images for testing purposes. In this study, CNN method was used to recognize these basic four emotions. After comparing the preprocessing methods in the CNN performance, 96.88% of recognition accuracy was achieved.