Machine Vision Based Methodologies for Monitoring Student Engagement in Online Learning: A Review
Hargobind Singh, Ravinder Singh, Kamal Deep Garg · 2023
Recent years have seen a flip of the education system from offline to online mode as during the lockdown there was no other option left. It provides flexibility in terms of time and place which provide freedom to learners to continue their learning according to their comfort. With so many positives there are a few drawbacks too such as a lot of reliability on good internet connectivity and fluctuating concentration levels. Technology is playing a crucial role in tackling the challenges that learners and teachers are facing in the online mode of education. One of the major challenges is maintaining student concentration due to the presence of learners in an uncontrolled environment. Researchers have tried multiple approaches to monitor student engagement through behavioral, collaborative, and emotional features. Data regarding the behavioral and collaborative features is mainly obtained from the user activity logs which can be fetched from the learning platform over which the e-learning session is scheduled. For obtaining data related to emotional features the video frames are processed via machine learning algorithms such as Haar Cascade, and Convolution Neural Network (CNN). Once the data is available a ML model can be trained that can be deployed to monitor student engagement in real time. This paper is a review of commonly used methods/techniques for examining student engagement and gaps in the current research related to it.