Human face feature point locating method and apparatus
A technology of face features and positioning methods, applied in the computer field, can solve problems such as inaccurate positioning of face feature points, and achieve the effect of improving positioning accuracy and accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0050] refer to figure 1 , which shows a flow chart of Embodiment 1 of a method for locating facial feature points according to the present invention, which may specifically include the following steps:
[0051] Step 101, obtaining a face image;
[0052] Embodiments of the present invention are applicable to general-purpose or special-purpose computing system environments or configurations. Examples: Personal computers, server computers, handheld or portable devices, tablet-type devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, minicomputers, mainframe computers, including any of the above Or the distributed computing environment of the device and so on.
[0053] In the embodiment of the present invention, firstly, the face image to be positioned may be acquired. Specifically, the face image may be obtained in any manner, for example, a face image of any format and size may be obtained through network download, ...
Embodiment 2
[0078] In practical applications, there will be a problem of information loss in the process of information transmission in the convolutional layer, and as the number of layers in the deep neural network increases, the training error will also increase, and the network will be less likely to converge. In order to solve the above problems, the embodiment of the present invention adopts a deep convolutional neural network based on residual learning to solve the problem of information loss in the convolutional layer and increase the convergence speed of the network training process.
[0079] In an optional embodiment of the present invention, the deformable convolutional neural network may specifically be a deep convolutional neural network based on residual learning; the deformable convolutional neural network includes at least one convolutional layer, and A layer-hopping connection is set between the at least one convolutional layer; the acquisition of the feature map correspond...
Embodiment 3
[0113] This embodiment describes in detail the training process of the network model of the present invention on the basis of the foregoing embodiments. refer to figure 2 , showing a flow chart of an embodiment of a method for training the network model of the present invention, which may specifically include:
[0114] Step 201, collecting face sample images under various backgrounds and postures;
[0115] Specifically, face images under various backgrounds and poses can be collected from network downloads, camera photography, and the like. For example, the same face can be shot under different lighting conditions, or it can be shot from different angles to obtain face images with different backgrounds and postures, such as strong light, low light, front, side, head down, head up, wearing glasses , not wearing glasses, etc.
[0116] Step 202, marking the location information of the face feature points in the face sample image;
[0117] Specifically, the position of the fa...
PUM
Abstract
Description
Claims
Application Information
- R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com