Emotion Recognition from Facial Images using Deep Learning Architectures

Abstract

Facial expressions (FE) or human countenance reflect psychological reactions or intentions stimulated within the mind in response to any social or personal event. These expressions play a significant role in conveying messages to the observer in non-verbal stealth mode. With the advancements in technology, facial expression recognition (FER) is considered crucial in understanding human behavior. We can infer those feelings and expressions are the essences of any interaction. In the same way, we need to make human machine interaction as communal as human-human interaction by making machines proficient at detecting human emotions by reading facial expressions. From recent year’s discoveries, a Set of multiple features have been recognized that provide possibly useful outcomes in the field of emotion recognition. Few preprocessing steps have been performed on the image data set before the extraction of features. In this work different prevalent methodologies, techniques and the types of features that are used by researchers in the past to predict the facial expression over time will be combined, so that a new and more efficient model can be designed. The purpose of this research is to design an automated system which can recognize seven basic emotions of human namely anger, disgust, fear, happy, sadness, Neutral and surprise for effective communication between humans and computers. The single algorithm to provide perfect recognition in all the scenarios has never been established so far; however, the research has been in progress to develop substitutes or new models to improve the recognition process. A deep learning algorithm is explained in this research work for classifying the facial expression of the human. The proffered method investigates the effectiveness of deep convolution neural network (DCNN) with the help of multiple models, and the best achieved result is 94.88% of FER2013