Expression recognition method based on multistage deep neural network

Abstract

The invention relates to the technical field of expression recognition, and particularly discloses an expression recognition method based on a multistage deep neural network. The expression recognition method comprises the following steps: uniformly modifying labels of data of mood, nausea, fear and light skin expression labels with higher recognition complexity into other labels; and then sending all data into a first expression recognition network model for training, so that four types of categories including happiness, surprise, sadness and the like can be successfully recognized; then, sending the data sets with the labels of the angel, the nausea, the fear and the light skin into the feature extraction network model, and outputting feature vectors of the data; then, processing the feature vectors by a standardized flow model; and finally, sending the processed feature vector into a multi-layer perceptron (MLP). Through continuous learning and training of the MLP, finally, the MLP can successfully recognize four expressions of anger, nausea, fear and light bamboo. Therefore, the multi-stage neural network model after training can be utilized to perform high-precision identification on seven basic expressions.

Description

technical field [0001] The invention relates to the technical field of expression recognition, in particular to an expression recognition method based on a multi-level deep neural network. Background technique [0002] Expression is a very important kind of information in our daily communication. In actual communication, expressions often play a role in enhancing the effect of mutual communication. Psychologist A. Mehrabia mentioned in his book An Approach to Environment Psychology that in people’s daily communication, the information conveyed through language only accounts for 7% of the total information, while the information conveyed through facial expressions accounts for 7% of the total information. 55%. At the same time, with the development of machine learning technology in recent years, face recognition technology has also received extensive attention. In particular, facial expression recognition has received more attention in the fields of security, robot manufac...

AI Technical Summary

Problems solved by technology

However, the accuracy of the current expression recognition is not too high, especially for some expressions with overlapping features, the recognition effect is not excellent

Images