Smart E-Learning: Emotion Recognition Using VGG-16 and BiLSTM for Adaptive Learning

In modern education, E-Learning has contributed a lot in providing very convenient and accessible learning experiences. However, this technology lack in assessing student engagement and comprehension. This paper presents an implementation of enhancing machine learning based e-learning platform which incorporates real-time emotion recognition from facial expression and sentiment analysis from textual data. The proposed system uses a VGG-16 model for detecting emotions from facial expression and Bidirectional Long Short Term Memory (BiLSTM) network for text data emotion classification. The dataset used is FER-2013 for training the model for facial recognition, while a labelled emotion dataset is used for textual analysis. Our system enables teachers to improve and assess student engagement levels and adapt teaching in the way students would have wanted. The results demonstrates 86% accuracy for facial emotion recognition and 89.4% accuracy for text-based sentiment analysis. The findings highlight the potential of integrating AI-driven emotion recognition into e-learning platforms to enhance student engagement and learning outcomes.