Emotion Detection Through Facial Expressions for Determining Students’ Concentration Level in E-Learning Platform

Students’ concentration and involvement in class lectures are critical for understanding concepts that considerably increase academic achievement. The fundamental difference between e-learning and conventional learning is that, in traditional learning, education is provided directly by teachers. Direct interaction with students helps teachers understand students’ concentration levels. However, the absence of a human supervisor poses various restrictions in the e-learning environment. Since student-to-student and teacher-to-teacher have greater distance in e-learning, face-to-face communication is impossible. The teachers have no idea about student concentration levels. Understanding students’ learning attitudes and acting accordingly to boost their learning interests is a huge challenge. In this study, the researcher came up with a new method for determining a student’s concentration level from their facial expression in an online learning environment while considering all the difficulties. Initially, we do some preprocessing on the dataset so that data can be easy for further processing, like face and eye detection. We trained the model on the FER2013 dataset and achieved around 70% accuracy using a convolutional neural network. Finally, we tested the model in a real-time environment by inputting some recording clips of 13 participants. We broke down the videos into different frames and plotted the concentration level of each participant.