Face anonymization using a generative adversarial network

Patent metadata

Field	Value
Publication number	US-2023328039-A1
Application number	US-202318131147-A
Country	US
Kind code	A1
Filing date	Apr 5, 2023
Priority date	Apr 6, 2022
Publication date	Oct 12, 2023
Grant date	—

Abstract

A generative adversarial network performs face anonymization. In a first step, an input image showing a face to be anonymized is received. Furthermore, an input vector with control data for face anonymization is received. The generative adversarial network then generates an output image in which the face is anonymized in accordance with the control data of the input vector.

First claim

1 . A method for face anonymization, the method comprising:
receiving an input image showing a face to be anonymized; receiving an input vector with control data for face anonymization; and generating by a generative adversarial network, an output image in which the face is anonymized in accordance with the control data of the input vector.

2 . The method according to claim 1 , wherein the input vector comprises control data for individual facial attributes.

3 . The method according to claim 2 , wherein the control data specifies whether a facial attribute shall be modified, kept, or optionally modified.

4 . The method according to claim 3 , wherein the control data further specifies an amount of modification of a facial attribute.

5 . The method according to claim 1 , wherein the generative adversarial network comprises a generator sub-network, a discriminator sub-network, and an identity classifier sub-network.

6 . The method according to claim 5 , wherein the generator sub-network is trained to generate new faces with different facial attributes.

7 . The method according to claim 5 , wherein the discriminator sub-network is trained to evaluate if generated new faces are realistic and natural.

8 . The method according to claim 5 , wherein the identity classifier sub-network is trained to evaluate if face identities of new faces have been changed.

9 . The method according to claim 1 , wherein the generative adversarial network uses an adversarial loss function, a reconstruction loss function, an attribute loss function, and an identity loss function.

10 . The method according to claim 1 , wherein the input vector is selected based on an application scenario.

11 . An apparatus for face anonymization, the apparatus comprising a generative adversarial network configured to:
receive an input image showing a face to be anonymized; receive an input vector with control data for face anonymization; and generate an output image in which the face is anonymized in accordance with the control data of the input vector

12 . The apparatus according to claim 11 , wherein the input vector comprises control data for individual facial attributes.

13 . The apparatus according to claim 12 , wherein the control data specifies whether a facial attribute shall be modified, kept, or optionally modified.

14 . The apparatus according to claim 13 , wherein the control data further specifies an amount of modification of a facial attribute.

15 . The apparatus according to claim 11 , wherein the generative adversarial network comprises a generator sub-network, a discriminator sub-network, and an identity classifier sub-network.

16 . The apparatus according to claim 15 , wherein the generator sub-network is trained to generate new faces with different facial attributes.

17 . The apparatus according to claim 15 , wherein the discriminator sub-network is trained to evaluate if generated new faces are realistic and natural.

18 . The apparatus according to claim 15 , wherein the identity classifier sub-network is trained to evaluate if face identities of new faces have been changed.

19 . The apparatus according to claim 11 , wherein the generative adversarial network uses an adversarial loss function, a reconstruction loss function, an attribute loss function, and an identity loss function.

20 . The apparatus according to claim 11 , wherein the input vector is selected based on an application scenario.

Assignees

• Elektrobit Automotive Gmbh
• Univ Nanyang Tech

Inventors

• Wang Yi Estelle
• Zhai Liming
• Guo Qing
• Dahan Jeremy

Classifications

• H04L63/0421 Primary

  Anonymous communication, i.e. the party's identifiers are hidden from the other party or parties, e.g. using an anonymizer · CPC title

• G06T7/0002

  Inspection of images, e.g. flaw detection · CPC title

• G06T11/60

  Creating or editing images; Combining images with text · CPC title

• G06F21/6254

  by anonymising data, e.g. decorrelating personal data from the owner's identification · CPC title

• G06T2207/30168

  Image quality inspection · CPC title