Content Based Facial Emotion Recognition Model using Machine Learning Algorithm

Emotion recognition or sentiment analysis is identified as an important research topic in computer vision community. The challenges include identification of face, recognition of accurate emotion, appropriate database and so on. We have proposed and implemented a general Convolutional Neural network (CNN) building framework for emotion recognition. The model is formalized by developing a coincident system which fulfills the tasks of face detection and emotion classification using our proposed CNN architecture. The model is validated using the FER-2013 dataset. In the proposed work, we discuss the applicability of the proposed CNN model. This model lays a valuable analysis of the effect of adjusting the network size, pooling, and dropout. For a given model, the final accuracy on the validation data is around 63%.