Expression recognition model training method and device, equipment and storage medium

Abstract

The invention relates to the field of artificial intelligence, and discloses an expression recognition model training method and device, equipment and a storage medium. The expression recognition model training method comprises the steps of: obtaining a to-be-recognized face image set and a comparison face image set corresponding to the face image set, and constructing a face image pair set based on the face image set and the comparison face image set; based on the face image pair set, performing face representation extraction training on a preset face recognition model, and outputting a training result to obtain a first network model and a face representation pair set; on the basis of the face representation pair set, performing representation separation training on the first network model, and outputting a training result to obtain a second network model, an expression representation pair set and a non-expression representation pair set; and based on an expression representation pair set and a non-expression representation pair set, performing face analysis training on the second network model to obtain an expression recognition model. According to the method, the interference of non-expression factors on the recognition result in the expression recognition process is effectively reduced, so that the accuracy of model recognition is improved.

Description

technical field [0001] The invention relates to the field of artificial intelligence, in particular to a training method, device, equipment and storage medium for an expression recognition model. Background technique [0002] Facial expression is the key information for analyzing human emotions and intentions, and facial expression recognition is a key step for applications in social robots, medical care, driver fatigue monitoring and many other human-computer interaction systems. At present, the existing face recognition methods can be divided into three categories: one is the traditional method using manual features, the other is the shallow learning method, and the other is the deep learning method. Deep learning is the current mainstream method. Generally, the representation of the face image is extracted through the convolutional neural network, and then the classification of the expression is learned and output through the representation of the face image. The overall...

AI Technical Summary

Problems solved by technology

[0004] The main purpose of the present invention is to solve the problem that the recognition accuracy of the model is insufficient because the training method of the existing expression recognition model is easily interfered by factors not related to the expression during training

Images