Feature Acquisition for Facial Expression Recognition Using Deep Convolutional Neural Network

Abstract

We present a convolutional neural network for facial expression recognition based on feature acquisition. The proposed method adopts the structure of dual-channel convolution neural network, the network structure of each channel is designed according to the input sets, the extracted face and the extracted mouth are used as input to two channels simultaneously. Experiments are carried out on two different data sets include JEFFA and FER-2013 to determine the recognition accuracy, and we build a set to test our model, and we compare the generalization performance by using the confusion matrix, then we compared and analyzed the experiment results of recognition accuracy under different facial expressions. Finally, our facial expression recognition system got an accuracy of 82% and 78% respectively, and learning meta face recognition in unseen domains should be researched in the future.