Facial expression recognition based on improved residual network

Abstract Facial expressions are an important part of human emotional signals and their recognition has become topic research in the field pattern recognition. Deep learning based methods have achieved great success facial expressions. However, with evolution convolution neural networks increased network depth, these suffer from problems such as degraded performance loss feature information. To address problems, a novel expression algorithm on improved residual is proposed. First, designed to extract deep features while retaining shallow ones. This can effectively prevent degradation performance. Moreover, when gradient Rectified Linear Units activation function used module 0, it will inactivate neurons cause this, Mish instead. The slight allowance for negative values improves flow. Next, inception introduced obtain richer information under same receptive field. Finally, by conducting verification experiments public datasets CK+ KDEF, authors manage solve insufficient extracted features, achieving accuracy rates 96.37% 93.38% two datasets, respectively.