Face privacy protection method for defending deep counterfeiting

The invention discloses a face privacy protection method for defending deep forgery, and belongs to the field of computer vision face privacy protection.The method comprises the steps that disturbance is added to an original face image through a disturbance generator, and a protected face image is g...

Full description

Saved in:
Bibliographic Details
Main Authors LEE KYOUNG-JI, LUO CHANGJIANG, ZHANG HUA, CAO XIAOCHUN
Format Patent
LanguageChinese
English
Published 14.06.2024
Subjects
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
ELECTRIC DIGITAL DATA PROCESSING
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:The invention discloses a face privacy protection method for defending deep forgery, and belongs to the field of computer vision face privacy protection.The method comprises the steps that disturbance is added to an original face image through a disturbance generator, and a protected face image is generated; two face images are subjected to face changing through face changing models, high-dimensional image features and identity information vectors of the two face images are calculated, different types of face changing models are selected to carry out alternate training on a disturbance generator, and the trained disturbance generator is used for generating a protected face image. According to the method, antagonistic disturbances which are high in aggressiveness and difficult to perceive can be generated, the disturbances are effective to an unknown model, and powerful protection is provided for face images on social media. 本发明公开了一种防御深度伪造的人脸隐私保护方法，属于计算机视觉人脸隐私保护领域，本发明将原始人脸图像通过扰动生成器添加扰动，生成受保护的人脸图像；将两张人脸图像经过换脸模型
Bibliography:Application Number: CN202410300112