Key point detection model, detection method and device thereof and computer storage medium

A technology for detecting models and key points, which is applied in the field of face recognition, can solve problems such as difficult detection of key points of faces, and achieve the effects of saving calculation time, meeting computing power requirements, and improving success rate and accuracy

Pending Publication Date: 2021-03-26
四川云从天府人工智能科技有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, limited by the resource and computing power configuration of the mobile terminal, it is very challenging to design a CNN model for the mobile device terminal. Therefore, how to design an efficient model that can meet the needs of the mobile device terminal is a new challenge in face recognition.
[0004] In addition, phonemes such as blur, dimness, large-angle side faces, masks, and glasses in actual application scenarios will bring difficulties to the detection of key points of the face.

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
  • Key point detection model, detection method and device thereof and computer storage medium
  • Key point detection model, detection method and device thereof and computer storage medium
  • Key point detection model, detection method and device thereof and computer storage medium

Examples

Experimental program
Comparison scheme
Effect test

no. 1 example

[0032] figure 1 A schematic flowchart of the key point detection model training method according to the first embodiment of the present application is shown. As shown in the figure, the key point detection model training method of this embodiment mainly includes the following:

[0033] Step S11, according to the initial training samples, determine each coordinate parameter and each first visibility category parameter corresponding to each key point in the initial training sample.

[0034] Optionally, the initial training samples include human face images, but not limited thereto, and various other animal face images are also applicable to this application.

[0035] For example, the target human face area (ROI area) in the original picture (such as a panoramic picture) can be extracted as the initial training sample required by this application.

[0036] Optionally, image processing techniques such as cropping and resizing can be used to extract the target face area from the ...

no. 2 example

[0065] image 3 It is a schematic flowchart of the key point detection model training method according to the second embodiment of the present application. As shown in the figure, this embodiment takes the image erasing enhancement rule as an example to describe an exemplary implementation of the above step S12 in detail, which mainly includes:

[0066] Step S31, according to the picture erasing enhancement rules and the coordinate parameters corresponding to each key point, at least one key point to be erased is obtained as the target hidden point.

[0067] Specifically, the image erasing enhancement rule in this embodiment is a designated erasing rule, that is, multiple key points may be designated from the initial training samples as target hidden points to be erased.

[0068] For example, you can specify figure 2 The key points 1 to 12 involving the eyes are used as the target hidden points to be erased, so as to simulate the complex face wearing eyes; as another exampl...

no. 3 example

[0080] Figure 5 It is a schematic flowchart of the key point detection model training method according to the third embodiment of the present application. As shown in the figure, this embodiment still takes the image erasing enhancement rule as an example to describe in detail another exemplary implementation of the above step S12, which mainly includes:

[0081] Step S51 , the step of generating the target erasing area, randomly generates the position information of the target erasing area according to the image erasing enhancement rule.

[0082]Specifically, the picture erasing enhancement rule in this embodiment is a random erasing rule, that is, the target erasing area to be erased can be randomly generated from the initial training samples.

[0083] Step S52, according to each coordinate parameter corresponding to each key point and the position information of the target erasing area, the number of key points in the target erasing area is obtained.

[0084] In this emb...

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 provides a key point detection model training and detection method and device and a computer storage medium, and the method mainly comprises the steps: determining each coordinate parameter and each first visibility class parameter corresponding to each key point in an initial training sample according to the initial training sample; according to a preset data enhancement rule, executing data enhancement processing on the initial training sample to obtain an enhanced training sample and each second visibility category parameter corresponding to each key point in the enhanced training sample; and constructing a key point detection model, and training the key point detection model by taking the enhanced training sample as input and taking each coordinate parameter correspondingto each key point and each second visibility category parameter as output. According to the invention, the method achieves the optimization of the recognition of various types of complex faces, and can reduce the inference time consumption of the detection model, so as to meet the computing power demands of a mobile equipment end.

Description

technical field [0001] The embodiments of the present application relate to the technical field of face recognition, and in particular to a key point detection model, a detection method, a device, and a computer storage medium. Background technique [0002] Face key point detection is one of the key technologies in face recognition and analysis processing, and it is the basis for the realization of face recognition, facial expression analysis, and 3D face reconstruction. Face key points, also known as face alignment, help to locate face poses, restore face models, and understand facial expression attributes. In recent years, the application of face recognition technology has been relatively mature, among which, face key point detection technology has also been widely used. [0003] At present, face recognition has been applied more and more on mobile devices. However, limited by the resource and computing power configuration of the mobile terminal, it is very challenging t...

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 Applications(China)
IPC IPC(8): G06K9/00G06K9/62
CPCG06V40/172G06V40/161G06V40/168G06F18/214
Inventor 秦勤
Owner 四川云从天府人工智能科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap