Facial emotion analysis method based on convolutional neural network
A convolutional neural network and emotion analysis technology, applied in the field of facial emotion analysis based on convolutional neural network, can solve the problems of ignoring the correlation of emotional dimensions and reducing the accuracy of prediction, so as to improve the accuracy of prediction performance and good practicability Effect
Inactive Publication Date: 2019-12-24
XIAN UNIV OF TECH
View PDF1 Cites 1 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] The purpose of the present invention is to provide a facial emotion analysis method based on convolutional neural network, which solves the problem of ignoring the correlation between emotion dimensions and reducing the accuracy of prediction when the existing model performs facial emotion analysis
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0044] The present invention will be described in detail below in combination with specific embodiments.
[0045] A kind of facial emotion analysis method based on convolutional neural network of the present invention, comprises the following steps:
[0046] Step 1, select training samples and validation samples.
[0047]The present invention is carried out based on the Affect-Net database, and the Affect-Net data set provides dimension labels and discrete labels, wherein there are nine types of discrete labels corresponding to continuous emotion label values. In order to ensure the balance of the selected training samples, the present invention selects nine types of discrete samples for nine types of discrete labels, and each type of discrete samples takes 10,500 samples, 10,000 as training samples, and 500 as verification samples; therefore, select 90,000 1 as training samples and 4500 as validation samples to test the performance of the model on the test set provided by th...
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
Login to View More Abstract
The invention discloses a facial emotion analysis method based on a convolutional neural network. The method comprises the steps of selecting a training sample and a verification sample; inputting thetraining sample to obtain a normalized face image and normalized face feature points; sending the normalized face image to a convolutional neural network model to extract emotion features, sending normalized face feature point information to a full connection layer to obtain position features, and splicing the emotion features and the position features; and mapping the emotion features into a two-dimensional prediction label, defining a loss function, measuring loss, performing back propagation on the network, completing model training, and performing verification through the verification sample. According to the facial emotion analysis method, the convolutional neural network model and the multi-output root-mean-square error are adopted to extract the emotion features.Meanwhile, the facefeature points are added, so that the problem that emotion information is lost due to manual feature definition is solved, the correlation between emotion dimensions is described, and the predictionperformance accuracy of the model is improved.
Description
technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a facial emotion analysis method based on a convolutional neural network. Background technique [0002] The concept of affective computing was first proposed by the MIT Media Lab in the 1990s. Scientists tried to turn "subjective emotions" into information that computers can recognize and calculate. Affective computing can realize barrier-free communication between humans and computers, making Computers tend to be intelligent. Emotional computing can be roughly divided into two processes: recognition and communication. Recognition is to allow machines to accurately analyze human emotional states through learning, and expression is to allow machines to accurately convey emotions on a suitable carrier. The current research focus is on How to make the machine accurately recognize and analyze facial emotions. [0003] Traditional facial emotion recognition base...
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
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04
CPCG06V40/174G06V40/172G06N3/045G06F18/24323
Inventor 孙强刘磊张龙涛
Owner XIAN UNIV OF TECH