Evaluation of Deep Learning and Machine Learning Algorithms for Facial Emotion Recognition

Facial Emotion Recognition is a crucial aspect of human-computer interaction, with applications in healthcare, security, psychology, and artificial intelligence. This research presents a comparative analysis of three models: SVM, VGG16, and a Custom CNN for classifying facial expressions into seven emotion categories. A hybrid dataset was created by combining publicly available FER datasets, including FER-2013 totaling 35,887 images. The SVM model achieved 60.2% accuracy and VGG-16 improved recognition accuracy to 73.6%. The Custom CNN outperformed both models, achieving an accuracy of 83%, precision of 84.5%, recall of 85.0%, and F1-score of 84.7%.