A face recognition method and system without face data training

A technology of face recognition and data training, which is applied in the field of face recognition methods and systems without face data training, which can solve the problems of failure to realize real-time update of face data, lack of adaptability to recognition tasks, and low precision of face photo processing, etc. question

Inactive Publication Date: 2018-12-25
CHENGDU REMARK TECH CO LTD +1
View PDF5 Cites 40 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In the prior art, the Chinese invention patent with the application number of 2017108922472 and the application date of 2017.09.27 discloses a stranger recognition method and system, but the recognition method is not adaptable to the recognition task of strangers; The processing accuracy is not high, and the real-time update of face data is not realized

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
  • A face recognition method and system without face data training
  • A face recognition method and system without face data training
  • A face recognition method and system without face data training

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0058] A face recognition method trained on unmanned face data mainly includes step S103: Obtain the comparison value between the face image and the face base database. If the Face ID image can be found in the base map library according to the comparison value, The detected ID is output. If the similarity between the input image and the face base image is greater than the recognition threshold plus 5, then the face quality judgment will be performed. If the face quality standard is met, if the base photo is greater than or equal to 10, the number of recognitions will be deleted If the Face ID cannot be found in the basemap library according to the comparison value, then the face quality judgment will be performed. If the face quality standard is met, the image will be added to the face In the bottom library, then output ID.

[0059] The present invention does not require massive face data training, realizes face recognition without a bottom database, and realizes real-time updati...

Embodiment 2

[0061] This embodiment is further optimized on the basis of embodiment 1, such as figure 1 As shown, if the base photo is less than 10, it is directly added to the face base. If the Face ID cannot be found in the base map library based on the comparison value, if the face quality standard is met, the feature value MD5 code is used to generate the Face ID and added to the face base library, and then the ID is output; if the face is not satisfied The quality standard is judged as unidentified. If the face quality standard is not met, a queue is used to temporarily store the face data that has not been recognized. When the server is idle, the face database is compared and the ID is found, and the cached picture is deleted and the ID is output. When the number of newly added pictures in the face database is greater than 500, face clustering is performed, pictures with high similarity are deleted, and Face IDs with high similarity are merged. After the ID is output, the number of...

Embodiment 3

[0065] This embodiment is optimized on the basis of embodiment 1 or 2, and further includes the following steps:

[0066] Step S101: such as figure 2 As shown, MTCNN is used to detect the facial features of the face, determine the position of the face in the image, and extract one or more captured photos of the face in the video; scale the given picture to different sizes to form an image pyramid to achieve the size constant;

[0067] Step S102: such as Image 6 As shown, the ResNet algorithm is used for face recognition, and the cosine distance between the captured image and the face base image is calculated to form a comparison value.

[0068] The step S101 mainly includes the following steps:

[0069] Step S1011: such as image 3 As shown, the P-Net full convolutional network is used to generate the candidate window and the frame regression vector; the Bounding box regression method is used to correct the candidate window, and the non-maximum value is used to suppress and merge the...

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 discloses a face recognition method without face data training. The invention does not need massive face data training, and automatically recognizes the collected face pictures under thecondition of lacking a face base database, which is convenient to quickly collect and classify the massive human data, reduces the workload of manual recognition, and has good adaptability to the recognition task of strangers. The invention realizes the real-time updating of the data in the human face base, improves the definition of the image in the human face base, and improves the accuracy ofthe human face recognition. The invention also discloses a face recognition system without a face base library. The face feature clustering module of the invention makes up for the problem of inaccurate judgment of the side face of the MTCNN, improves the accuracy of the photograph acquisition and improves the accuracy of the face recognition. So that face recognition technology can be used in a wider and more complex scene.

Description

Technical field [0001] The invention belongs to the field of computer vision and image processing, and in particular relates to a face recognition method and system for human face data training. Background technique [0002] From the discovery of face recognition technology to the application of face recognition technology, it has a history of more than 60 years. In recent years, face recognition technology has been a hot research topic in research fields such as model recognition, image processing, machine vision, and neural networks. The more prominent research results have been applied to many industrial fields, such as security verification systems, credit card verification, criminal identification, banking and customs health, human-computer interaction and other fields. [0003] Compared with other biometric technologies, face recognition technology has relative advantages: 1. Simpler data collection procedures, no manual operation, no contact; 2. Faster speed and easy to use ...

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): G06K9/00G06K9/62G06T7/00
CPCG06T7/0002G06T2207/30168G06V40/172G06V40/50G06F18/214
Inventor 李源王飞
Owner CHENGDU REMARK TECH CO LTD
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