A Coherent Way of detecting pupil’s emotions via live Camera using CNNs along with Haar -Cascasde Classifier

In terms of education, a person’s emotions influence their knowledge and ability to follow a show. Educators can use automated measurement of a person’s emotions to achieve educational goals and personalize delivery. In this study, we present video-based emotion identification technique in virtual learning programs. The main feelings we detect here are disappointment, disinterest, fear, satisfaction, commitment, sadness, and surprise. Continuity of affect, valence and stimulus states extracted from consecutive video frames, together with a new vector of latent affective and behavioral traits, were used to emotion detection. Deep learning-based temporal models and classical machine learning-based non-temporal models are trained and verified on image-level and video-level features, respectively. Images from the facial expression recognition dataset were used. The results show that the vast majority of methods can detect student learning outcomes. The model presented in this paper can provide technical support for the modeling of learners’ emotions in the teaching assistant system.