Virtual privacy curtain

Patent metadata

Field	Value
Publication number	US-12356114-B2
Application number	US-202217967201-A
Country	US
Kind code	B2
Filing date	Oct 17, 2022
Priority date	Oct 17, 2022
Publication date	Jul 8, 2025
Grant date	Jul 8, 2025

Abstract

Presented herein are techniques to process an image for a video conference. A method includes obtaining an image from a camera, identifying a first person and a second person in the image, labeling pixels in the image belonging to the first person and to the second person, determining a first distance between the first person and the camera and a second distance between the second person and the camera, and based on the first distance and the second distance, selectively removing or blurring at least one of the first person and the second person from the image, before transmitting the image to an endpoint of a video conference, by manipulating the pixels in the image belonging to the at least one of the first person and the second person.

First claim

What is claimed is:

1. A method comprising:
obtaining an image from a camera;
identifying a first person and a second person in the image;
labeling pixels in the image as belonging to the first person and to the second person;
determining a first distance between the first person and the camera and a second distance between the second person and the camera;
determining at least one of a first pose of the first person and a second pose of the second person based on a plurality of keypoints for each of the first person and the second person; and
based on the first distance and the second distance, and the at least one of the first pose of the first person and the second pose of the second person, selectively removing or blurring at least one of the first person and the second person from the image, before transmitting the image to an endpoint of a video conference, by manipulating the pixels in the image belonging to the at least one of the first person and the second person.

2. The method of claim 1 , wherein labeling pixels in the image is performed by a machine learning model that identifies the plurality of keypoints for each of the first person and the second person.

3. The method of claim 2 , further comprising determining the first distance and the second distance using the machine learning model and the plurality of keypoints.

4. The method of claim 2 , further comprising determining the first pose of the first person and the second pose of the second person based on the plurality of keypoints.

5. The method of claim 4 , further comprising, based on the at least one of the first pose of the first person and the second pose of the second person, determining that the at least one of the first person and the second person is an intended participant of the video conference.

6. The method of claim 1 , further comprising applying a smoothing process between successive images from the camera.

7. The method of claim 6 , wherein the smoothing process comprises applying a predetermined delay before determining the first distance between the first person and the camera and the second distance between the second person and the camera on a subsequent image from the camera.

8. The method of claim 1 , further comprising establishing, by manipulating the pixels in the image, a virtual wall between the first person and the second person based on the first distance and the second distance.

9. The method of claim 8 , wherein the virtual wall occludes at least a portion of the first person or a portion of the second person from the image before transmitting the image to the endpoint of the video conference.

10. The method of claim 1 , wherein identifying the first person and the second person in the image, and labeling pixels in the image as belonging to the first person and to the second person comprises segmenting the image, pixel by pixel.

11. An apparatus comprising:
a network interface configured to enable network communications;
a memory configured to store logic instructions; and
a processor, when executing the logic instructions, configured to:
obtain an image from a camera;
identify a first person and a second person in the image;
label pixels in the image as belonging to the first person and to the second person;
determine a first distance between the first person and the camera and a second distance between the second person and the camera;
determine at least one of a first pose of the first person and a second pose of the second person based on a plurality of keypoints for each of the first person and the second person; and
based on the first distance and the second distance, and the at least one of the first pose of the first person and the second pose of the second person, selectively remove or blur at least one of the first person and the second person from the image, before transmitting the image to an endpoint of a video conference, by manipulating the pixels in the image belonging to the at least one of the first person and the second person.

12. The apparatus of claim 11 , wherein the processor is further configured to label pixels in the image using a machine learning model that identifies the plurality of keypoints for each of the first person and the second person.

13. The apparatus of claim 12 , wherein the processor is further configured to determine the first distance and the second distance using the machine learning model and the plurality of keypoints.

14. The apparatus of claim 12 , wherein the processor is further configured to determine the first pose of the first person and the second pose of the second person based on the plurality of keypoints.

15. The apparatus of claim 14 , wherein the processor is further configured to, based on the at least one of the first pose of the first person and the second pose of the second person, determine that the at least one of the first person and the second person is an intended participant of the video conference.

16. The apparatus of claim 11 , wherein the processor is further configured to apply a smoothing process between successive images from the camera.

17. The apparatus of claim 16 , wherein the smoothing process comprises applying a predetermined delay before determining the first distance between the first person and the camera and the second distance between the second person and the camera on a subsequent image from the camera.

18. A non-transitory computer readable storage media encoded with instructions that, when executed by a processor, cause the processor to:
obtain an image from a camera;
identify a first person and a second person in the image;
label pixels in the image as belonging to the first person and to the second person;
determine a first distance between the first person and the camera and a second distance between the second person and the camera;
determine at least one of a first pose of the first person and a second pose of the second person based on a plurality of keypoints for each of the first person and the second person; and
based on the first distance and the second distance, and the at least one of the first pose of the first person and the second pose of the second person, selectively remove or blur at least one of the first person and the second person from the image, before transmitting the image to an endpoint of a video conference, by manipulating the pixels in the image belonging to the at least one of the first person and the second person.

19. The non-transitory computer readable storage media of claim 18 , encoded with instructions that, when executed by the processor, cause the processor to label pixels in the image using a machine learning model that identifies the plurality of keypoints for each of the first person and the second person.

20. The non-transitory computer readable storage media of claim 19 , encoded with instructions that, when executed by the processor, cause the processor to:
determine the first pose of the first person and the second pose of the second person based on the plurality of keypoints.

Assignees

• Cisco Tech Inc

Inventors

• Cazzani Dario
• Reed William
• Mosure Mitchell

Classifications

• G06T5/70 Primary

  Denoising; Smoothing · CPC title

• G06T7/10

  Segmentation; Edge detection (motion-based segmentation G06T7/215) · CPC title

• G06V10/761

  Proximity, similarity or dissimilarity measures · CPC title

• G06T7/73

  using feature-based methods · CPC title

• G06V2201/07

  Target detection · CPC title