The Customized Convolutional Neural Network of Face Emotion Expression Classification

Expressions are that phenomena of human behavior which can help us to understand the human nature and sometimes feeling they are going through. Feelings relate to human behavior but it comes from your expressions. Face, migrate these expressions to the user level interactions. These interactions may use to act back for humans but not for machines. In this paper, Convolutional Neural network (CNN) are used to understand the seven different human face expressions. The seven classes are fear, angry, disgust, sad, happy, surprise and neutral. The dataset consisted is consisted upon almost 36,000 gray scale images. Our customized proposed CNN model of 4 convolutional and 2 fully connected layer is achieved 64.3 % accuracy of test data.