Facial Expression Recognition Based on Convolutional Denoising Autoencoder and XGBoost

In order to solve the insufficiency of feature extraction in traditional facial expression recognition process and further improve the classification accuracy, a deep learning method combining the convolutional denoising autoencoder and XGBoost model is proposed in this paper. In the early stage of feature extraction, convolutional autoencoder is used to fully learn the high-dimensional complex feature data and reduce the nonlinear dimension. On this basis, noise is introduced into the original image to enhance the robustness and generalization ability of the model. In the later classification process, XGBoost classifier is used to classify the extracted features. Experiments are performed on JAFFE and CK+ facial expression recognition datasets, the results show that this method has better recognition accuracy than other comparison methods.