Facial expression recognition method based on graph convolutional network

Abstract

The invention discloses a facial expression recognition method based on a graph convolutional network, which is used for solving the problem that because preprocessed face area image loses part of image information, expression recognition accuracy corresponding to preprocessed face area image is affected, and because image convolutional network cannot select nodes with richer identification information according to weights of different nodes, expression classification accuracy is affected The method comprises the following steps: (1) generating a training set; (2) constructing a graph convolutional network; (3) training the graph convolutional network; and (4) identifying the facial expression. On the basis of all face key points of each picture, the topological graph corresponding to the picture is generated, and a face five-sense-organ pooling module group formed by connecting seven face five-sense-organ pooling modules in parallel is built, so that the method has high facial expression classification accuracy during facial expression recognition.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a facial expression recognition method based on a graph convolutional network in the technical field of image recognition. The present invention can be applied to identify anger, disgust, fear, joy, sadness, surprise and neutral expression information corresponding to the human face from the human face image. Background technique [0002] Facial expression is one of the signals for human to communicate and transmit emotion, which intuitively expresses human's emotional feedback to external things. Facial expression recognition has attracted extensive attention for its potential applications in the fields of image processing and image recognition. Traditional convolutional neural networks usually use the entire aligned face of a 2D image as the input of the network to learn a feature representation. However, the original pixels of these images are susceptible to v...

AI Technical Summary

Problems solved by technology

It is used to solve the loss of part of the image information due to the loss of part of the image information of the preprocessed face area image, which will affect the accuracy of expression recognition corresponding to the preprocessed face area image and because the graph convolutional network cannot select identification information according to the weight of different nodes Richer nodes, resulting in the classification accuracy of affecting expressions

[0006] The idea of ​​realizing the purpose of the present invention is to generate the topological map corresponding to the picture based on all the face key points of each picture, which is used to solve the problem that the preprocessed face area image will lose part of the image information, which will affect the preprocessed image. The problem of the accuracy of expression recognition corresponding to the face area image

Build a face and facial features pooling module group composed of seven face and facial features pooling modules in parallel to solve the problem that the graph convolutional network cannot select nodes with richer identification information according to the weights of different nodes, which affects the accuracy of expression classification rate problem

Images