Detection and Classification of Human Emotion Using Deep Learning Model

Human emotions are the mental states resulting from neurophysiological changes that are diversely linked to human thoughts, feelings, and behavioral reactions. Facial expressions, eye movement and gestures are several nonverbal communication methods used in many human-machine interactions. Facial emotion recognition has been an essential field because it has many applications that convey the mental state and emotions of each individual. As there is no clear way to distinguish between emotions on a face and the complicated neural system that is connected to the face, recognizing human emotions is a difficult undertaking. In this paper, for students' facial expression recognition, we presented a model called Convolutional Neural Network (CNN). The proposed model consists of six convolutional layers, six max pooling with two fully connected layers. Using a Haar cascade detector, the system identifies the faces of students from input images. It categorizes them into seven different facial expressions, including surprised, fearful, disgusted, sadness, happiness, anger, and neutral. The FER2013 dataset yielded an accuracy rate of 77.22% for the proposed model. The comparative study is carried out with four other deep learning models. The models are Inception V3, VGG-16, VGG-19, Resnet50 and the accuracies achieved are 40%, 37%, 22.50%, 22.09%, respectively.