Student Emotion Estimation Based on Facial Application in E-Learning during COVID-19 Pandemic

Abstract

The COVID-19 pandemic occurred in late 2019, and by the beginning of 2020 the entire education system has shifted from traditional teaching methods to online learning systems around the world. COVID-19 reinforces the need to explore online learning and learning opportunities. However, the ability of teachers to recognize and see how individual students engage in online learning is more challenging. Student emotions such as self-esteem, inspiration, dedication, and others that are assumed to be determinants of student success cannot be overlooked. The main objective of this research is to evaluate the emotion of student on e-learning during COVID-19 pandemic using a facial recognition application. This application able to interpretation of facial expressions into extracted emotional states. An image processing approach has been implements in 4 types of emotion which is happy, normal, sad and surprise. Next the image will go through the identifying of emotion type from the static frontal face image. It starts with image acquisition, grayscale conversion and contrast stretching for image pre-processing, Haar Cascade or also known as Viola-Jones technique for face detection, face model technique for eye and mouth localization, skin-color segmentation technique for image segmentation, and Grey-Level Co-Occurrence Matrix (GLCM) for feature extraction. The classification for emotion type is using SVM Regression. The accuracy percentage of emotion classification is calculated. The result showed that SVM Regression has a high accuracy rate of 99.16%. A real-time application will be developed to identify human face emotion instead of static image for future work with additional of speech recognition exploration.