Research on Emotion Recognition Based on Deep Learning for Mental Health

Abstract

This paper briefly introduced\xa0the support vector machine (SVM)\xa0based and convolutional neural network (CNN)\xa0based healthy\xa0emotion recognition method, then\xa0improved the traditional CNN by\xa0introducing Long Short Term Memory\xa0(LSTM), and finally carried out simulation experiments on three emotion recognition models, the SVM, traditional CNN, and improved CNN models, in the self-built face database. The results showed\xa0that the CNN model converged\xa0faster in training\xa0and had a smaller error\xa0when it was stable\xa0after introducing LSTM; compared with\xa0the\xa0SVM and traditional CNN\xa0models, the improved CNN had a higher recognition accuracy for facial expressions;\xa0the time\xa0consumed by the improved CNN\xa0model\xa0was the shortest in both training and testing stages.