Emotional Interfaces for Effective E-Reading using Machine Learning Techniques

Abstract

Emotion-aware systems are very essential for effective e-reading. The aim of the proposed work is to detect and classify cognitive states from facial expressions of the students engaged in online learning which improves the e-reading process to a greater extent. In this proposed work the emotions such as happy, irritate, sleep and yawn that are mainly used for effective E-reading are taken into consideration. The Haar cascaded classifier is used to segment the facial regions from the input images. The Zernike moment features are extracted from the selected face regions. The extracted features are fit into Random Forest and Decision Tree machine learning models. The models classify the emotions. Finally the classified emotions are interfaced with e-reading. The proposed work is found to perform better than the existing methods.