Facial expression recognition method fusing parallel network features

The invention discloses a facial expression recognition method fusing parallel network features, and the method comprises the steps: firstly improving a conventional VGG13 network, and replacing the last common convolution group in the conventional VGG13 network with a cavity convolution group; and secondly, a weighting processing layer is added between a fourth convolutional layer and a fifth convolutional layer of the Alexnet network, two networks of the improved Alexnet network and the improved VGG13 network are used to carry out feature extraction on the data set at the same time, the extracted features are fused, and then classification and identification are carried out. The purpose of this innovation is to obtain a larger receptive field by using cavity convolution and convert single feature information into richer feature information through a feature fusion mode under the condition that time and parameters are almost not increased. The system can be used in an online classroom, improves the control o