Unlock instant, AI-driven research and patent intelligence for your innovation.

A Facial Expression Recognition Method Based on Cost-sensitive Convolutional Neural Network

A convolutional neural network, facial expression recognition technology, applied in the field of intelligent human-computer interaction, can solve the problems of reducing the practicability of facial expression recognition models, different costs, and large differences, achieving strong facial emotion recognition capabilities, reducing Loss, the effect of improving the accuracy of expression recognition

Active Publication Date: 2021-08-24
BEIJING INSTITUTE OF TECHNOLOGYGY
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In fact, the actual application costs of different expression misclassification results are different or even very different. Ignoring the cost of expression misclassification will inevitably reduce the practicability of the facial expression recognition model

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Facial Expression Recognition Method Based on Cost-sensitive Convolutional Neural Network
  • A Facial Expression Recognition Method Based on Cost-sensitive Convolutional Neural Network
  • A Facial Expression Recognition Method Based on Cost-sensitive Convolutional Neural Network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0085] The present invention uses the CK+ data set, wherein 80% of the data is used as a training data set for model training, and 20% of the data is used as a test set for model verification. Table 1 shows some experimental data, and the output layer contains 6 types of expressions.

[0086] Table 1 Experimental data

[0087]

[0088] like Figure 5 The experimental results shown provide a comparison of the experimental results of the method proposed in the present invention, the existing machine learning method SVM+LBP and the convolutional neural network method. Experimental results show that: using the method proposed in the present invention for facial expression recognition, the overall cost is lower than that of SVM+LBP and CNN methods.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention proposes a facial expression recognition method based on a cost-sensitive convolutional neural network. On the basis of considering the loss caused by misclassification of facial expression categories, the cost caused by misclassification is integrated into the whole process of model training and parameter optimization. In the process of model training, the different effects caused by misclassification of different expression categories are considered. And according to the size of the loss caused by the misclassification, the model parameters are adaptively adjusted to minimize the loss caused by the misclassification of the expression type. This method solves the problem that in the existing facial expression recognition model based on convolutional neural network, it assumes or defaults that the misclassification costs of different expression categories have equal influence, and improves the accuracy of expression recognition while reducing the loss caused by misclassification of expressions. Strong facial emotion recognition ability.

Description

technical field [0001] The invention belongs to the technical field of intelligent human-computer interaction, and relates to a facial expression recognition method, specifically, a method for training a facial expression recognition model using a cost-sensitive convolutional neural network and performing facial expression recognition . Background technique [0002] Expression is the expression of human emotion on the face, which contains a lot of human emotion and psychological activity information. Expression recognition aims to mine the hidden emotional characteristics of human faces and perform emotion classification. At present, expression recognition has become a hot spot in the Internet and related industries, especially in the fields of traffic safety, marketing strategy, intelligent human-computer interaction, emotional robot, and smart home. [0003] Facial expression recognition mainly includes three parts: face detection, expression feature extraction, and emoti...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00G06N3/04G06N3/08
CPCG06N3/084G06V40/174G06N3/045
Inventor 李慧芳石峰娟袁艳
Owner BEIJING INSTITUTE OF TECHNOLOGYGY