Granted Patent B2

US 12,469,110 · App. 17/554,437 · Granted Nov 11, 2025

Inventors: Stanley Baran (Chandler, AZ); Charu Srivastava (Danville, CA); Srikanth Potluri (Folsom, CA); Michael Rosenzweig (Queen Creek, AZ); Archie Sharma (Folsom, CA)

Assignee: Intel Corporation

G06T5/50G06T5/20H04N23/90H04R1/406H04R3/005H04R3/04G06T2207/10016G06T2207/20081G06T2207/20084

Abstract

Claims (50)

1 . A multi-camera style transfer system comprising:

at least one memory;

instructions in the system; and

at least one processor circuit to be programmed by the instructions to at least:

access a first live video feed from a first camera and a second live video feed from a second camera, the first and second live video feeds of a same scene from different positions;

apply a style transfer function to a second image from the second live video feed to generate a stylized image based on a reference image, the reference image corresponding to a first image from the first live video feed;

adjust one or more parameter settings of a video filter based on the stylized image; and

filter the second live video feed via the video filter with the adjusted parameter settings to generate a filtered version of the second live video feed.

2 . The multi-camera style transfer system of claim 1 , wherein the style transfer function is a photorealistic style transfer function.

3 . The multi-camera style transfer system of claim 1 , wherein the style transfer function is applied by a machine learning model.

4 . The multi-camera style transfer system of claim 3 , wherein the style transfer function is a Neural Network (NN) style transfer function.

5 . The multi-camera style transfer system of claim 1 , wherein the parameter settings correspond to visual characteristics including at least one of color temperature, tone, exposure, white balance, hue, saturation, or brightness.

6 . The multi-camera style transfer system of claim 1 , wherein one or more of the at least one processor circuit is to adjust the one or more parameter settings of the video filter by:

filtering the second image with the video filter with first parameter settings to generate a filtered image; and

applying a loss function to:

determine a difference between the stylized image and the filtered image; and

change the first parameter settings to second parameter settings based on the difference between the stylized image and the filtered image.

7 . The multi-camera style transfer system of claim 1 , wherein one or more of the at least one processor circuit is to determine whether to update the reference image.

8 . The multi-camera style transfer system of claim 7 , wherein one or more of the at least one processor circuit is to determine whether to update the reference image based on a comparison of a parameter to a threshold.

9 . The multi-camera style transfer system of claim 8 , wherein the parameter includes at least one of a time limit, a difference between a current image from the first live video feed and the reference image, or a scene change in the first live video feed.

10 . The multi-camera style transfer system of claim 1 , wherein one or more of the at least one processor circuit is to:

identify a first segment in the reference image;

identify a second segment in the second image;

apply the style transfer function to the second segment in the second image to generate the stylized image;

adjust the one or more parameter settings of the video filter based on the stylized image; and

filter a corresponding segment in the second live video feed using the video filter with the adjusted parameter settings.

11 . A non-transitory computer readable storage medium comprising instructions to cause at least one processor to at least:

access a first live video feed from a first camera and a second live video feed from a second camera, the first and second cameras live video feeds of a same scene from different positions;

apply a style transfer function to a second image from the second live video feed to generate a stylized image based on a reference image, the reference image corresponding to a first image from the first live video feed;

adjust one or more parameter settings of a video filter based on the stylized image; and

filter the second live video feed via the video filter with the adjusted parameter settings to generate a filtered version of the second live video feed.

12 . The non-transitory computer readable storage medium of claim 11 ,

wherein the style transfer function is a photorealistic style transfer function.

13 . The non-transitory computer readable storage medium of claim 11 , wherein the style transfer function is applied by a machine learning model.

14 . The non-transitory computer readable storage medium of claim 13 , wherein the style transfer function is a Neural Network (NN) style transfer function.

15 . The non-transitory computer readable storage medium of claim 11 , wherein the parameter settings correspond to visual characteristics including at least one of color temperature, tone, exposure, white balance, hue, saturation, or brightness.

16 . The non-transitory computer readable storage medium of claim 11 , wherein the instructions are to cause the at least one processor to adjust the one or more parameter settings of the video filter by:

filtering the second image of the video filter with first parameter settings to generate a filtered image; and

applying a loss function to:

determine a difference between the stylized image and the filtered image; and

change the first parameter settings to second parameter settings based on the difference between the stylized image and the filtered image.

17 . The non-transitory computer readable storage medium of claim 11 , wherein the instructions are to cause the at least one processor to determine whether to update the reference image.

18 . The non-transitory computer readable storage medium of claim 17 , wherein the instructions are to cause the at least one processor to determine whether to update the reference image based on a comparison of a parameter to a threshold.

19 . The non-transitory computer readable storage medium of claim 18 , wherein the parameter includes at least one of a time limit, a difference between a current image from the first live video feed and the reference image, or a scene change in the first live video feed.

20 . The non-transitory computer readable storage medium of claim 11 , wherein the instructions are to cause the at least one processor to:

identify a first segment in the reference image;

identify a second segment in the second image;

apply the style transfer function to the second segment in the second image to generate the stylized image;

adjust the one or more parameter settings of the video filter based on the stylized image; and

filter a corresponding segment in the second live video feed using the video filter with the adjusted parameter settings.

Assignments (1)

Continuity (1)

References Cited (14)