Face Emotion Recognition Using Deep Neural Network (DNN)

Facial expression recognition is an intelligent human-computer interaction technology that gives a great sense of communication of the expression of our emotions, understanding, and intent between ourselves. Many researchers have attempted to categorize facial expressions, and it has been proven that there are seven universal basic emotions: happiness, sadness, anger, fear, surprise, disgust, and neutrality. Real-time emotion detection is achievable using a webcam. The paper discusses an architecture based on CNN for emotion detection. This model was trained on the FER-2013 dataset after performing some pre-processing steps such as scaling images and color mode correction. A multi-layer CNN model is designed for the classification of emotions as happy, sad, angry, fearful, or surprised, as well as disgusted and neutral. In addition to this, it focuses on real-time analysis to present solutions based on emotions detected, such as automatically playing videos to address sad or angry or fearful emotions. Our developed algorithm attained an highest accuracy of 97.8%. The developed algorithms are very robust, effective and helps in automated seizure detection.