Automated Student Engagement Monitoring and Evaluation during Learning in the Wild

With the explosive growth of edge computing and massive open online courses (MOOCs), there is an urgent need to enable pervasive learning so that students could study with high efficiency at any comfortable places at their own pace. Although there have been a number of studies and applications for student engagement monitoring and evaluation in the pervasive learning, most of the existing works are either supported by commercial eye tracking devices/software or designed for off-line studies on the basis of questionnaires, self-reports, checklists, quizzes, teacher introspective evaluations, and assignments. In this work, we investigate the feasibility of real-time student engagement monitoring and evaluation with low-cost off-the-shelf web-cameras in realistic learning scenarios. To recognizing and evaluating student engagement, a new model is developed and trained by a deep learning Convolutional Neural Network (CNN) with an open source dataset. The quantitative experimental results demonstrate that the deep learning CNN and our model work well and efficiently when monitoring student learning and detecting student engagement in real time.