Engagement Analysis of Learners Using Emotions: The SimEng System

There are many research and also scientific and practical attempts to detect learners’ engagement during teaching. Identifying disengaged learners and getting them engaged have always been a significant concern in education. Automatic engagement detection and effective reinforcement measures make teaching and particularly e-learning more reliable and efficient. In this paper, we evaluate complexities from the analytical viewpoint of learners’ videos in synchronous classes. These complexities are ignored, while they are undeniable issues in the learning engagement area. We note the necessity of using neural networks and deep learning in engagement detection. In this paper, we propose a model to detect learner engagement using a known CNN model to recognize emotions and to measure the engagement level as simply as possible. Primary results show engagement detection accuracy is 57%, while the expected was 60%. The expected accuracy is obtained from the accuracy of the CNN model and the used dataset.