Face attribute data labeling method, computer equipment and storage medium
A technology of attribute data and data, applied in the field of image processing, can solve problems such as high cost, cumbersome acquisition process, and large differences, and achieve the effects of eliminating interference, reducing costs, and improving accuracy
Active Publication Date: 2022-05-17
小视科技(江苏)股份有限公司
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
These data collection methods are expensive, the collection process is cumbersome, and the laboratory environment is quite different from the real environment. The model trained based on this data often has insufficient generalization in real scenarios.
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
[0026] This embodiment provides a face attribute data labeling method, the steps are as follows figure 1 shown, including:
[0027] 1. Face data collection and preprocessing. The face color image dataset is obtained by preprocessing the face data.
[0028] In one embodiment, the method for face data collection and preprocessing includes:
[0029] Collect a large amount of face color image data, the data contains rich lighting, expression, identity, head posture, a total of N pieces, N is an integer greater than 100;
[0030] Use the face detector to detect the face area, and crop the image in the face area;
[0031] Cut out the image in the face area, and uniformly scale it to the size of high H pixels and wide W pixels (H and W are integers greater than 32), and obtain a data set ,in It is the preprocessed image for the nth sheet (n is an integer from 0 to N-1).
[0032] 2. Label the face region mask. Detect the face area mask of the image in the established face col...
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 face attribute data labeling method, computer equipment and storage media, and a face attribute data labeling method, comprising: establishing a face color image data set; detecting a face area mask of an image in the face color image data set ;Adopt the three-dimensional deformation model, randomly initialize the parameters of the images in the face color image dataset; render the initialization parameters to obtain the rendered image; label all the image data in the face color image dataset, and obtain the marked illumination data dataset and head pose data set; input the face image into the face attribute prediction model for training; iteratively optimize the model; perform face detection on the face image to be tested, cut out the image in the face area, and input it into the trained In the face attribute prediction model, the illumination parameters and head pose of the face at this time are predicted. The invention greatly reduces the cost of training data collection and improves the accuracy of fitting.
Description
technical field [0001] The invention relates to the technical field of image processing, in particular to a method for labeling attribute data of a face image, computer equipment and a storage medium. Background technique [0002] Facial attribute analysis refers to the analysis of facial lighting, head posture, gender, age and other information based on facial images. It has a wide range of applications in security, interactive entertainment and other fields. Existing computer vision-based face attribute analysis methods rely on high-quality annotated data. For example, face lighting data requires manual observation of the face image and labeling the corresponding macro-lighting labels, such as underexposure, overexposure, reddish, etc. The labeling is subject to greater subjective influence and cannot fully describe the complex lighting in the real environment. In addition, the collection of traditional lighting data requires the deployment of a large number of light sour...
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): G06T7/11G06T15/20G06V40/16G06N3/04G06V10/82
Inventor 杨帆郝强潘鑫淼胡建国
Owner 小视科技(江苏)股份有限公司