Towards Improving Online Learner Engagement Through Facial Emotion Recognition Techniques Using Macro and Micro Expressions

Abstract

Online learning has become increasingly prevalent, yet maintaining learner engagement remains a significant challenge. Traditional methods for gauging engagement, such as surveys or quizzes, provide limited real-time feedback and are often subjective. This study explores the potential of facial emotion recognition (FER) techniques to enhance online learner engagement by analysing macro and micro facial expressions. Macro expressions represent clear and voluntary emotional states, while micro expressions are brief and involuntary, providing deeper insights into cognitive and emotional responses. Using a combination of computer vision and machine learning models, this research aims to detect and classify learner emotions during online sessions. By integrating FER into online learning platforms, educators can receive real-time analytics on engagement levels and adapt their teaching strategies accordingly. The study investigates the efficacy of various FER algorithms in identifying engagement-related emotions, such as happy, sad, neutral, angry, surprise, confusion, interest, and boredom. Preliminary results demonstrate that incorporating FER significantly improves engagement metrics by enabling timely interventions and personalized feedback. This approach not only enhances the learning experience but also establishes a foundation for more empathetic and adaptive educational systems. The findings underline the potential of FER as a transformative tool in reshaping the dynamics of online education. newlineThe proposed research work with three key innovations: (1) Assessing engagement through macro facial expressions using an Enhanced Multihead Channel Attention (EMCA) mechanism integrated into Residual Network-50(ResNet-50); (2) Improving engagement levels based on micro facial expressions using an Image Feature Stabilization Algorithm to accurately capture and interpret fleeting emotional cues using DenseNet161; and (3) Scaffolding macro and micro expression features to create a unified framework for facial emotion recognit

Description

Keywords

Citation

item.page.endorsement

item.page.review

item.page.supplemented

item.page.referenced