Research And Implementation Of Facial Expression Recognition Based On Deep Learning

With the continuous progress of artificial intelligence,the communication between robots and humans is becoming more and more frequent,and the ability of robots to understand human emotions is becoming more and more important.Face expression recognition technology is the key to realize human-computer emotional interaction.Facial expression recognition technology has a wide application prospect in intelligent monitoring,case trial,prevention and control of depression and other fields.In recent years,Convolutional Neural Network(CNN)has achieved very good results in the task of image recognition.Therefore,this paper conducted relevant research on facial expression recognition based on deep learning.The main contents of this paper are as follows:1.Data preprocessing: make a lot of processing in Fer2013 data set,and do some data enhancement,then through the MTCNN algorithm model,do the face detection and tailoring in Jaffe and CK + data set,then add the processed image to the training data set,in order to expand the amount and accuracy of the data,improve the training effect.2.Build recognition model: According to the ideas of CNN model and VGG model,three experimental models were built.The first experimental model was built according to the traditional convolution neural network.Then using the ideas of VGG model,in the same situation of the reception field,used more small convolution kernels instead of big convolution kernels,built a second experiment model,the accuracy of the model was improved,it also proved that increasing depth can improve the accuracy of the model,finally found the second model converges faster,earlier fitting,continue to improve the model,main task was to continue to deepen the depth,the third experiment model was build,Finally,the accuracy of the model was further improved,which further proved that improving the network depth can improve the accuracy of the model.3.Design and implementation of the system: a face expression recognition system is built under Python,which realized the recognition of face expressions in pictures and cameras.The system first transmits the input information to the MTCNN module,and then transmits the cut face part to the trained face expression recognition module after the face is detected.After the expression recognition,the recognition results are displayed on the front-end interface.The system recognition results show that the trained model is feasible.