In today’s rapidly evolving educational landscape, the analysis of student engagement through the prism of emotion ensures a transformative possibility for educators and learners. We introduce an innovative real-time emotion detection system that transcends conventional methodologies, with a specific emphasis on identifying student interest through emotion analysis in live video streams. This comprehensive approach, utilizing cutting-edge technologies such as data preprocessing, deep learning architecture, exhaustive training, real-time facial detection, emotion inference, and immersive graphical presentation, optimizes the recognition and visualization of emotions. Our methodology harnesses the powerful synergy between a pre-trained VGG19 model, renowned for its precision in feature extraction, and a custom-designed neural network finely tuned for both emotion classification and interest detection. To bolster our system’s performance and adaptability, we conducted extensive training and fine-tuning using the FER13 dataset, renowned for its comprehensiveness in emotion recognition spanning a wide range of human expressions. The custom preprocessed images are provided to a pre-trained VGG19 model with a custom neural network. The two dense layers with 256 and 128 units in the network, each accompanied by batch normalization and dropout, also ensure better system accuracy. Extensive data augmentation on the training data allows robust model training, enabling real-time emotion prediction and introducing innovative techniques for identifying student interest, exemplifying a comprehensive approach to emotion recognition and educational technology. Empowered by our meticulously refined model, our system rapidly and accurately predicts emotions, overlaying emotive descriptors onto live video feeds in real-time.