Face spoofing detection method and system based on meta-pseudo labels and illumination invariant features

A technology of constant illumination and spoof detection, which is applied in spoof detection, biometric recognition, character and pattern recognition, etc. It can solve the problem of face spoof detection model relying on the generalization of training data, so as to improve learning ability and reduce impact , the effect of improving robustness

Pending Publication Date: 2022-02-18
CHINA-SINGAPORE INT JOINT RES INST +1
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the defects and deficiencies in the prior art, the present invention provides a method and system for face deception detection based on meta-pseudo-labels and illumination-invariant features. The present invention adopts a technical solution based on meta-pseudo-labels and illumination-invariant features , which solves the technical problem that the face spoofing detection model is too dependent on training data and poor generalization, and achieves the technical effect of effectively improving the generalization performance while ensuring the accuracy rate in the library

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
  • Face spoofing detection method and system based on meta-pseudo labels and illumination invariant features
  • Face spoofing detection method and system based on meta-pseudo labels and illumination invariant features
  • Face spoofing detection method and system based on meta-pseudo labels and illumination invariant features

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0108] This embodiment uses the Replay-Attack, CASIA-MFSD, and MSU_MFSD living body detection data sets as examples for training and testing, and introduces the implementation process of this embodiment in detail. Among them, the Replay-Attack dataset contains 1200 videos. Using a MacBook camera with a resolution of 320×240 pixels, the real faces from 50 testers and the generated spoofed faces are collected, and the ratio of 3:3:4 It is divided into training set, verification set and test set; the CASIA-MFSD dataset contains 600 videos, which are collected from 50 The real face of the tester and the deceptive face generated accordingly are divided into a training set and a test set according to 2:3; the MSU_MFSD data set includes 280 videos, and the real faces from 35 testers are collected. The generated spoofed faces, 15 of them are used in the training set and 20 of them are used in the test set. Since the CASIA-MFSD and MSU_MFSD living body detection data sets do not inclu...

Embodiment 2

[0222] Such as Image 6 As shown, the present embodiment provides a face deception detection system based on meta-pseudo-labels and illumination invariant features, including: data preprocessing module, student model and teacher model building module, teacher learning module, student meta-learning module, teacher update module, attention module, illumination invariant feature extraction network building block, illumination invariant feature learning module, verification module and testing module;

[0223] In this embodiment, the data preprocessing module is used to extract the image of the face area to obtain the RGB color channel map, and the RGB color channel map to be trained is randomly cut out image blocks, which are divided into labeled samples and unlabeled samples. The unlabeled sample obtained by performing random data enhancement on the label sample is used as a sample for RGB branch training; the RGB color channel image to be trained is preprocessed by light separat...

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 spoofing detection method and system based on meta-pseudo labels and illumination invariant features. The method comprises the steps that: data is preprocessed to obtain an RGB color channel graph and a PLGF graph; the RGB color channel graph is divided into labeled samples, unlabeled samples and enhanced unlabeled samples, the labeled samples, the unlabeled samples and the enhanced unlabeled samples are sent to a teacher learning module to obtain teacher semi-supervised loss, false labels of the unlabeled samples and enhanced unlabeled loss, and parameters of a student model and a teacher model are updated; the PLGF map is sent to an illumination invariant feature extraction network to obtain a feature vector and a classification vector, and a network model and parameters are stored after triple loss and cross entropy loss are used for supervised training; a threshold value is determined by using a verification set; and test data is loaded to the student model and the illumination invariant feature extraction network to obtain a corresponding RGB classification score and a PLGF classification score, weighted summation is performed to obtain a classification score, and a classification result is judged according to a threshold value. According to the method, the robustness of the face spoofing detection model is improved under the condition that the training samples are insufficient.

Description

technical field [0001] The invention relates to the technical field of face recognition anti-spoofing detection, in particular to a face spoofing detection method and system based on meta-pseudo tags and illumination invariant features. Background technique [0002] Today, the use of facial biometrics in businesses and industries has increased dramatically. For example, face unlocking technology can be used to protect personal privacy in electronic devices, and facial biometrics can be used to authenticate payments. However, using the face as a biometric for authentication is not secure. Facial biometric systems can be vulnerable to spoofing attacks. Face spoofing attack methods can generally be divided into four categories: 1) photo attack: the attacker uses photos printed or displayed on the display to deceive the authentication system; Video deception authentication system; 3) face mask attack, the attacker wears a face mask carefully crafted according to the victim to ...

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): G06V40/40G06V10/56G06K9/62G06V40/16G06V10/764
CPCG06F18/2411G06F18/2415
Inventor 冯浩宇王宇飞胡永健蔡楚鑫葛治中
Owner CHINA-SINGAPORE INT JOINT RES INST
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