Depression recognition method, device and storage medium based on facial video
A technology for identifying devices and faces, applied in the field of computer vision, can solve problems such as low accuracy of recognition results, inability to identify depression, and inability to intuitively display the facial expression status of patients with depression
Active Publication Date: 2022-05-31
CAPITAL NORMAL UNIVERSITY
View PDF5 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] Inspired by this, there have been related studies that use expression signals to complete the automatic detection and recognition of depression. However, this kind of research is mostly focused on the recognition of static facial images, and cannot be combined with dynamic facial change information for depression recognition, which leads to the recognition of depression. The result is less accurate
Moreover, the output depression recognition result is usually depression or the depression score corresponding to the static picture, but this output result cannot intuitively show the facial expression state of the depression patient, so it cannot better assist the doctor to make further judgments
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
[0100] It should be noted that the preset tailoring requirements are determined according to the target depression recognition model.
[0107] Exemplarily, when the frame number of the video training sample is 64 frames, the first preset frame number is also 64 frames.
[0118] Here, the layer parameters of the convolutional layer can specifically select the convolution kernel size (7×7×7), step size (1, 2, 2) to
[0124] In another embodiment, the spatiotemporal self-attention layer includes a temporal self-attention layer, a spatial self-attention layer
[0127] After determining the first feature and the second feature, the first feature is temporally dimensioned by a temporal average pooling layer
[0129] In another implementation manner, the temporal feature, the spatial feature determined by the fusion layer pair
[0130] Here, the determined temporal features, the spatial features, and the temporal feature weights are determined by the fusion layer.
[0138] Here, all the a...
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 present application provides a facial video-based depression recognition method, device, and storage medium, including: extracting the pre-acquired target spatio-temporal features of each facial sub-video to be recognized through a pre-trained target depression recognition model; The depression recognition model outputs the depression degree recognition score of each facial sub-video to be recognized and determines the thermal video of each facial sub-video to be recognized through the heat map generator; determines the target based on the depression degree recognition score of each facial sub-video to be recognized The first depression recognition result of the person; the second depression recognition result of the target person is determined based on the thermal video; and the final depression recognition result of the target person is determined based on the first depression recognition result and the second depression recognition result. In this way, the final depression recognition result of the target person is determined based on the depression degree recognition score obtained from the spatiotemporal characteristics of the face sub-video to be recognized and the hot video, which can effectively improve the accuracy of the depression recognition result.
Description
Depression recognition method, device and storage medium based on facial video technical field The application relates to computer vision technology field, especially relate to a kind of depression recognition method based on facial video method, device and storage medium. Background technique Depression has become a major mental illness that endangers human life and health, and this mental illness will affect all years age group, causing serious health problems to patients. Today's diagnostic methods for depression mainly focus on professional doctors During teacher-guided interviews, experienced physicians can obtain diagnostic clues from patients' abnormal expressions. Inspired by this, at present there is relevant research to utilize facial expression signal to complete the automatic detection and identification of depression, but This kind of research mostly focuses on the recognition of static facial pictures, and cannot combine dynamic facial change informa...
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 Patents(China)
IPC IPC(8): G06V20/40G06V40/16G16H50/20G16H50/70
CPCG16H50/70G16H50/20
Inventor 尚媛园潘昱辰邵珠宏刘铁丁辉
Owner CAPITAL NORMAL UNIVERSITY