A Deep Learning Model for Face Expression Detection

The field of facial expression detection is gaining a lot of attention in recent years. There have been many methods developed for recognizing facial expression. In this paper we present a Deep learning model for face expression classification. The model is designed to recognize seven different expressions: happy, sad, angry, fear, disgust, surprise and neutral. Haar cascade classifier was used for face detection. The model was trained using Facial Expression Recognition 2013 [FER2013] dataset and the results were analyzed. The proposed model achieved a training accuracy of 78% and validation accuracy of 67%. The system works for both real time video and static images. There are three modes of input, they are image from device, image capture using a webcam, and video capture using a webcam. The outputs of the model consist of the image with the detected expression, an emoticon related to the expression and a voice output stating the expression detected.