Facial Expression Recognition Using Xception And DenseNet Architecture

Researchers pay much attention to facial expressions recognition due to the rapid development of Artificial Intelligence. Facial expression recognition is used to help human-computer interaction. In addition, facial expression recognition is also used in psychological recognition, Human-computer interaction, assisted driving, and security station in everyday life. But most of the research focused on the machine learning approach rather than deep learning and the emotion classifications are also smaller. This facial expression recognition can be implemented using a deep learning approach. The architecture that is often used and considered to be the best in image classification is Convolutional Neural Network. Therefore, this study builds a Convolutional Neural Network Model with Xception and DenseNet architecture. The accuracy of the two models is compared, with Xception received an accuracy of 70% and DenseNet got 79%.