IP Library Granted Patent US 12,167,169
Granted Patent B1
US 12,167,169 · App. 17/933,186 · Granted Dec 10, 2024

Systems and methods to optimize video streaming using digital avatars

Inventors: Siddha Ganju (Santa Clara, CA); Ruthie Lyle (Durham, NC); Naveen Kumar Rai (Sunnyvale, CA); Ronay Ak (Sarasota, FL); Andrew Russell (Weston, FL)
Assignee: NVIDIA Corporation
H04N7/157G06T13/205G06T13/40G06V10/82G06V40/20H04N7/147H04N7/152G06V2201/07
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,167,169
App. No.
17/933,186
Granted
Dec 10, 2024
Kind
B1
Abstract

A digital avatar system can process video streams and generate synthetic video with a digital avatar. The digital avatar provides the appearance of a participant from the video stream talking and one or more of performing various behaviors or actions consistent with the participant's behavior when they are live streamed. A digital avatar system can detect triggering events during a live stream and automatically switch to an avatar mode.

Claims (53)

1. A method comprising:

causing a presentation, at a first computing device, of a first video feed and audio data of a participant in a video streaming session, wherein the first video feed and the audio data is provided using sensors connected to a second computing device;

detecting a triggering event;

responsive to the triggering event, generating a second video feed depicting a digital avatar corresponding to the participant by:

retrieving an image associated with the participant from the first video feed;

generating an initialized avatar image using the image;

processing the initialized avatar image and the audio data, the processing including an animation of the initialized avatar image that is synchronized with the audio data; and

causing a presentation, using the first computing device, of the second video feed.

2. The method according to claim 1 , wherein detecting the triggering event includes using a Large Language Model.

3. The method according to claim 1 , wherein retrieving the image associated with the participant is further comprised of selecting an image from a frame of the first video feed of the participant.

4. The method according to claim 1 , wherein detecting the triggering event includes at least one of: detecting movement of the participant within the first video feed, determining and monitoring a relative position of the participant within a frame of the first video feed, detecting using object recognition, an object within the first video feed, and receiving a signal from an IoT device.

5. The method according to claim 1 , wherein the animation of the second video feed includes a realistic rendering of the participant.

6. The method according to claim 1 , wherein the method provides a seamless video feed of the participant in the video streaming session by dynamically switching between a recorded video feed of the participant and a synthetic video feed of the participant during the video streaming session.

7. The method according to claim 1 , wherein

detecting, after causing the presentation of the second video feed, a second triggering event; and

responsive to the second triggering event, causing a presentation, using the first computing device, of the first video feed of the participant.

8. The method according to claim 1 , wherein generating the second video feed further comprises:

generating animated facial data, using a first neural network, from the audio data; and

generating the second video feed, using a second neural network, from the animated facial data and from the selected image of the participant.

9. The method according to claim 1 , wherein displaying the second video feed includes replacing the first video feed with the second video feed.

10. A method comprising:

receiving, at a first computing device, a first video feed and audio data of a participant in a video streaming session, wherein the first video feed is provided using sensors connected to a second computing device;

detecting a triggering event;

responsive to the triggering event, generating a second video feed depicting a digital avatar corresponding to the participant by:

retrieving an image associated with the participant from the first video feed;

generating an initialized avatar image using the image;

processing the initialized avatar image and the audio data, the processing including an animation of the initialized avatar image that is synchronized with the audio data; and

sending the second video feed to a third computing device.

11. The method according to claim 10 , wherein detecting the triggering event includes using a Large Language Model.

12. The method according to claim 10 , wherein retrieving the image associated with the participant is further comprised of selecting an image from a frame of the first video feed of the participant.

13. The method according to claim 10 , wherein detecting the triggering event includes at least one of: detecting low lighting conditions in the first video feed, detecting a low battery condition for one of the first computing device or the second computing device, detecting that no video data is being received from the second computing device, and detecting a low bandwidth condition for the first computing device or the second computing device.

14. The method according to claim 10 , wherein generating the second video feed further comprises:

generating animated facial data, using a first neural network, from the audio data; and

generating the digital avatar video, using a second neural network, from the animated facial data and from the selected image of the participant.

15. The method according to claim 10 , wherein generating the second video feed further comprises:

retrieving a set of behavior data; and

processing the initialized avatar image using the set of behavior data.

16. The method according to claim 10 , wherein the method provides a seamless video feed of the participant in the video streaming session by dynamically switching between a recorded video feed of the participant and a synthetic video feed of the participant during the video streaming session.

17. The method according to claim 10 , wherein detecting the triggering event includes detecting that the participant is in a sensitive geographic location.

18. A system, comprising:

one or more processors to:

cause a presentation, at a first computing device, of a first video feed and audio data of a participant in a video streaming session, wherein the first video feed and the audio data is provided using sensors connected to a second computing device;

detect a triggering event;

responsive to the triggering event, generate a second video feed depicting a digital avatar corresponding to the participant by:

retrieving an image associated with the participant from the first video feed;

generating an initialized avatar image using the image;

processing the initialized avatar image and the audio data, the processing including an animation of the initialized avatar image that is synchronized with the audio data; and

cause a presentation, using the first computing device, of the second video feed.

19. The system according to claim 18 , wherein the system provides a seamless video feed of the participant in the video streaming session by dynamically switching between a recorded video feed of the participant and a synthetic video feed of the participant during the video streaming session.

20. The system according to claim 18 , wherein the one or more processors further:

select an image of the participant;

generate animated facial data, using a first neural network, from the audio data; and

generate the second video feed, using a second neural network, from the animated facial data and from the selected image of the participant.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 19, 2022
From: GANJU, SIDDHA; LYLE, RUTHIE; RAI, NAVEEN KUMAR; AK, RONAY; RUSSELL, ANDREW
To: NVIDIA CORPORATION
Reel/Frame 061135/0380 →
Cited By (3)
US 12,367,641 US 12,456,180 US 12,568,288