Training and detecting methods and systems for key human facial feature point detection model

A technology of key features and training methods, applied in character and pattern recognition, instruments, computer components, etc., can solve the problem of poor detection accuracy of non-significant key feature points, low confidence discrimination accuracy of key feature points, and insufficient detection accuracy and other issues to achieve the effect of improving training and detection accuracy, enhancing stability and accuracy, and enhancing error tolerance

Active Publication Date: 2016-03-16
CHONGQING INST OF GREEN & INTELLIGENT TECH CHINESE ACADEMY OF SCI +1
View PDF3 Cites 38 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The existing methods for detecting key feature points of faces can be summarized as follows: firstly, according to the feature information of the initial key feature points of the training set and the real key feature points, the corresponding regression model is trained; secondly, through feature extraction and combined with the regression model, find The best position of each key feature point; although the detection accuracy of the key feature points of the face is improved under multi-angle and illumination, there are sti

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
  • Training and detecting methods and systems for key human facial feature point detection model
  • Training and detecting methods and systems for key human facial feature point detection model
  • Training and detecting methods and systems for key human facial feature point detection model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0046] like figure 1 As shown, it is a flow chart of a human face key feature point detection model training method in the embodiment of the present invention, which is described in detail as follows:

[0047] Step S101, using a face detection algorithm to obtain the face position of the input image;

[0048] Wherein, the input picture is one of bmp, jpg, tiff, gif, pcx, tga, exif, fpx, svg, psd, cdr, pcd, dxf, ufo, eps, ai, raw in any of the following formats, and is none Compressed pictures.

[0049] Before step 1, the pictures containing faces are collected, and the face position area and key feature points of faces in the pictures are calibrated according to preset rules to generate a training set. Specifically, for the pictures containing human faces collected by the user through various channels, the face position area and key feature points of the face in the picture are marked according to the preset rules of the training set, and the positions of the marked face pos...

Embodiment 2

[0070] like figure 2 Shown, for the embodiment of the present invention figure 1 The dynamic initialization regression model training flow chart in the middle is detailed as follows:

[0071] In step S201, the position of the real key feature points is mapped to a preset 3D (yaw / pitch / roll) face model, and the three-dimensional rotation angle of the face is calculated according to the POSIT algorithm;

[0072] In step S202, the face of the 3D face model is mapped to the 2D space according to the three-dimensional rotation angle and similar transformation is performed to obtain the updated initial position of the key feature point;

[0073] In step S203, the initial position of the key feature point before the update and the initial position of the key feature point after the update are subjected to histogram specification processing;

[0074] In step S204, the difference between the initial positions of the key feature points before and after the update and the regional fea...

Embodiment 3

[0076] Through the training set image {d which contains a set of face images i}, the training set includes the pre-marked face location area {r i} and the coordinates of key feature points of the face Train a dynamically initialized regression model R as follows:

[0077] 3.1, for each input picture, according to the face location area r i The initial position of key feature points before updating can be obtained;

[0078] 3.2, according to the coordinates of the key feature points of the face And the POSIT algorithm can calculate the three-dimensional rotation angle of the face;

[0079] 3.3, according to the known face 3D model x 3D , through steps such as matrix rotation, 3D to 2D plane mapping, and similar transformation, the updated initial position of key feature points is obtained

[0080] 3.4. To train the dynamic initial model R, we refer to the SDM solution method, that is, to solve the optimal solution of the following formula:

[0081] ...

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 present invention provides a training method and system and a detecting method and system for a key human facial feature point detection model. The training method comprises: acquiring a human face position of an input picture; obtaining an initial position of a key feature point before updating according to an average key feature point of a training set and the face position; obtaining an initial position of a key feature point after updating according to a position of an authentic key feature point; according to a difference value between the initial positions of the key feature points before and after updating and a region feature extracted before updating, training a dynamic initialization regression model; and training a cascade regression model according to a distance difference between the initial position of the key feature point after updating and the position of the authentic key feature point and a region feature extracted after updating. The detecting method comprises: calling the dynamic initialization regression model and the cascade regression model in turn for a to-be-detected picture, and calculating a position of a key human facial feature point; and determining whether the key facial feature point is accurate according to a comparison with a preset point. By using the methods and systems provided by the present invention, the accuracy of detecting the key human facial feature point is improved.

Description

technical field [0001] The invention relates to the field of computer vision processing, in particular to a human face key feature point detection model training method and system, detection method and system. Background technique [0002] The key feature points of the face are the basis of face processing technologies such as face recognition and expression recognition. The performance of facial feature point positioning greatly affects the accuracy of face detection methods. Among all facial feature points, salient key feature points such as eyes, mouth, nose tip and eyebrows are the most important, and the distance ratio between them is used to distinguish faces. For general applications, salient key feature points can already meet the needs of processing methods, can align and normalize faces of different shapes and sizes, and provide information for further processing. In addition, the six points of left / right eyes, mouth, nose tip and eyebrows can also be used as the ...

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
IPC IPC(8): G06K9/00
CPCG06V40/161G06V40/168
Inventor 邵枭虎周祥东石宇周曦
Owner CHONGQING INST OF GREEN & INTELLIGENT TECH CHINESE ACADEMY OF SCI
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