FExR.A-DCNN: Facial Emotion Recognition with Attention Mechanism Using Deep Convolution Neural Network

Human Facial Emotions play an important role in non-verbal communication between people. Automated Facial Recognition can have various impacts on our technology, helping us to better understand human behaviour, detect mental disorders, and synthesising facial expressions. Methods based on appearance and geometry are predominantly used, but fail to achieve high accuracy with limited data-sets. In this article we proposed various techniques using deep learning concepts of CNN to identify 7 key human emotions. We achieved 98% accuracy on CK+ data set having low sample count in 100 epochs, which confirms the superiority of the model in detecting and focusing on key global features for Facial Emotion Recognition.