Educational Quality Assessment System based on Emotions using Facial Images Applying Deep Learning Techniques

Romero-Alva V.; Ramos-Cosi S.; Roman-Gonzalez A. · International Journal of Engineering Trends and Technology (2024) · DOI: 10.14445/22315381/ijett-v72i3p122

Resumen

In response to the shift towards online education prompted by the COVID-19 pandemic, this project aims to analyze students' emotional responses in a virtual classroom setting through the capture of facial expressions. By leveraging facial imagery, the initiative seeks to objectively assess levels of engagement, interest, or lack thereof, during online sessions in a discreet and non-intrusive manner. This methodology is intended to provide actionable insights to refine and enhance the effectiveness of future educational sessions, thereby optimizing the learning experience. Facial images are captured in real-time by students who have their cameras enabled during live class sessions. Subsequently, these images undergo processing to identify and evaluate the students' facial expressions. In its preliminary phase, the model successfully identifies and categorizes four primary emotions-happiness, sadness, anger, and surprise-with an impressive accuracy rate exceeding 90%, per the system's evaluation criteria. This accuracy is gauged based on the prevalence of each emotion throughout the class. The findings are then compiled and shared with educators as constructive feedback to inform and improve the planning and execution of subsequent sessions. Additionally, a Chrome browser extension has been developed to facilitate the deployment of this system within "Google Meet"platforms.

Instituciones: Universidad de Ciencias y Humanidades