Face Expression Recognition using Deep Neural Network

Abstract

The research topic concerned in this paper is creating a machine learning model for facial expression recognition (FER). It is a technology to which uses biometric markers to detect emotions in human faces. FER is important because of human-computer interaction, pattern recognition and image recognition. Three components are containing in our method, including dataset construction, model building and Emoji generation. First, the authors aim to build a hand-crafted Convolutional Neural Network to recognize the emotion and designed a GUI in which generates the corresponding emoji derived from the facial expression put in. This paper tested our model in the popular FER2013 dataset and our model achieves 89% accuracy on the dataset, which validates the satisfying performance of our method. Our method can both achieve satisfying accuracy on the dataset and show its strong application ability in most of the tasks. The method can be applied to the real-world scenes for expression recognition and emoji generation.