Articles for disrupting automated visual object tracking processes

Patent metadata

Field	Value
Publication number	US-11941823-B2
Application number	US-202017438865-A
Country	US
Kind code	B2
Filing date	Mar 13, 2020
Priority date	Mar 14, 2019
Publication date	Mar 26, 2024
Grant date	Mar 26, 2024

Abstract

System and method for producing an adversarial article that may be used to disrupt an automated visual tracking process. An input module receives input related to a specific automated visual tracking process. Based on that input, a pattern-design module generates an adversarial pattern. The adversarial pattern may then be applied to an article, which may be any kind of physical or virtual object. The tracker's normal processing modes are disrupted when the tracker attempts to process an image containing the adversarial article(s). The tracker may be mounted on an autonomous vehicle, a mobile robot, or other mobile or stationary camera surveillance system.

First claim

What is claimed is:

1. A method for producing an adversarial article, said method comprising the steps of:
(a) receiving input related to a specific automated visual tracking process;
(b) based on said input, generating an adversarial pattern for said specific automated visual tracking process; and
(c) applying said adversarial pattern to an article to thereby produce said adversarial article,
wherein a normal processing mode of said specific automated visual tracking process is disrupted when an image of said adversarial article is processed by said automated visual tracking process.

2. The method according to claim 1 , wherein said specific automated visual tracking process is performed by a system mounted on an autonomous vehicle.

3. The method according to claim 1 , wherein, based on said input, at least one characteristic of said specific automated visual tracking process is determined, and wherein said adversarial pattern is based on said at least one characteristic.

4. The method according to claim 1 , wherein generating said adversarial pattern in step (b) is automatic.

5. The method according to claim 1 , wherein generating said adversarial pattern in step (b) is performed using a machine-learning-based module.

6. The method according to claim 1 , wherein said article comprises at least one of: a poster; a banner; a wall; a vehicle; a garment; an electronic display; a virtual object; a two-dimensional surface; and a three-dimensional object.

7. The method according to claim 1 , wherein said adversarial pattern is applied to said article in step (c) using at least one of pigment and light.

8. A system for producing air adversarial article, said system comprising:
an input module for receiving input related to a specific automated visual tracking process;
a pattern-design module for generating an adversarial pattern for said specific automated visual tracking process, wherein said generating is based on said input; and
an article,

wherein said adversarial pattern is applied to said article to thereby produce said adversarial article, and
wherein a normal processing mode of said specific automated visual tracking process is disrupted when an image of said adversarial article is processed by said automated visual tracking process.

9. The system according to claim 8 , wherein said specific automated visual tracking process is performed by a system mounted on an autonomous vehicle.

10. The system according to claim 8 , wherein said input module determines at least one characteristic of said specific automated visual tracking process, and wherein said adversarial pattern is based on said at least one characteristic.

11. The system according to claim 8 , wherein said pattern-design module comprises a machine learning-based module.

12. The system according to claim 8 , wherein said adversarial pattern is hidden within a source image.

13. The system according to claim 8 , wherein said article comprises at least one of: a poster; a banner; a wall; a vehicle; a garment; an electronic display; a virtual object; a two-dimensional surface; and a three-dimensional object.

14. The system according to claim 8 , wherein said adversarial pattern is applied to said article using at least one of pigment and light.

15. Non-transitory computer-readable media having encoded thereon computer-readable and computer-executable instructions that, when executed, implement a method for producing an adversarial article, the method comprising the steps of:
(a) receiving input related to a specific automated visual tracking process;
(b) based on said input, generating an adversarial pattern for said specific automated visual tracking process; and
(c) applying said adversarial pattern to an article to thereby produce said adversarial article,
wherein a normal processing mode of said specific automated visual tracking process is disrupted when an image of said adversarial article is processed by said automated visual tracking process.

16. The computer-readable media of claim 15 , wherein said specific automated visual tracking process is performed by a drone-mounted system.

17. The computer-readable media according to claim 15 , wherein generating said adversarial pattern in step (b) is automatic.

18. The computer-readable media according to claim 15 , wherein generating said adversarial pattern in step (b) is performed using a machine-learning-based module.

19. The computer-readable media according to claim 15 , wherein said article comprises at least one of: a poster; a banner; a wall; a vehicle; a garment; an electronic display; a virtual object; a two-dimensional surface; and a three-dimensional object.

20. The computer-readable media according to claim 15 , wherein said adversarial pattern is applied to said article in step (c) using at least one of pigment and light.

Assignees

• Element Ai Inc
• Servicenow Canada Inc

Inventors

• Xu Anqi
• Dia Ousmane
• Wiyatno Rey Reza

Classifications

• G06N3/094

  Adversarial learning · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/0475

  Generative networks · CPC title

• G06N3/09

  Supervised learning · CPC title