Facial Emotion Recognition Using Convolutional Neural Network in a Learning Environment

Emotion is simply a state of feeling or mood of a person, it can be derived from one’s circumstance or mood and as a result of advancing technologies, there has now been development of computer-based recognition systems that can analyze facial expressions from either images or videos to give a result on the emotional state of an individual. Now in the context of a learning environment, a sad, angry, happy, or anxious student would not learn effectively in a classroom because they are distracted by their emotions at the time. This research focuses on developing a Facial Emotion Recognition System(FERS) using Convolutional Neural Networks to help recognize and interpret facial expressions of students to perceptible emotions that will assist teachers in designing and implementing the right teaching strategies to bring all students to a more convenient emotional state and ensure that attentiveness of students is at its highest efficiency during teaching. Our model consists of five stages: facial detection from the image, facial landmarking, points extraction and tracking of major facial features (eyes, eyebrows, and mouth), and then the classification of facial expression. The network model was trained on the FER 2013 database using the seven (7) major and generally accepted facial expressions. The training process was conducted over 40 epochs and achieved a learning rate of 96.47% and a test accuracy of 71.44%.