Emotion Detection Based on De-expression Residue Learning and Its Intelligent System Design

Due to the difference of individual facial expression, it will have a significant impact on the detection of the emotion. Therefore, in order to mitigate this impact, and to solve the problems of high complexity and low accuracy of other modern emotion detection algorithms, this algorithm designed in this paper is based on an expression residue learning framework, called De-expression residue learning. Its core idea is to divide Emotion into Expression and Neutral, and then orderly carry out model training, emotion recognition and classification from the extracted Expression. It can effectively reduce the interference caused by human differences and improve the accuracy of emotion classification. It is tested on the public datasets Emotionet Dataset and Fer2013 Dataset. The experimental results show that the algorithm in this paper has high detection accuracy and effectively reduces the false detection rate. Finally, based on the algorithm in this paper, an intelligent recognition system is designed to increase the practical value of the research.