Granted Patent B2

US 12,456,300 · App. 17/832,637 · Granted Oct 28, 2025

Inventors: Renjie Tao (Sunnyvale, CA); Ling Tsou (Lawndale, CA)

Assignee: Zoom Communications, Inc.

G06V20/46G06V40/1347

Abstract

Claims (66)

1. A method, comprising:

receiving video content of a communication session comprising a plurality of participants;

extracting frames from the video content;

classifying the frames of the video content;

identifying one or more distinguishing frames comprising a presentation slide, wherein identifying the one or more distinguishing frames comprises:

removing a thumbnail of a participant video feed from each of the frames;

inverting colors of each of the frames to obtain inverted frames to enhance text detection speed;

obtaining a pixel value summation for each of the inverted frames;

calculating a pixel value summation difference between adjacent inverted frames; and

determining a distinguishing frame when the pixel value summation difference meets a threshold;

for each distinguishing frame comprising the presentation slide, detecting a title within the frame, wherein detecting the title within the frame is based on one or more title detection rules, wherein one or more candidate titles are determined prior to determining the title, wherein one of the title detection rules comprises determining that a number of candidate titles determined for the frame does not exceed a threshold number of candidate titles, and wherein one of the title detection rules comprises determining that a font size for the title meets or exceeds a threshold ratio of font size relative to other text within the frame;

formatting, by a processing engine the titles for each distinguishing frame into a JavaScript Object Notation (JSON) file format; and

transmitting, by the processing engine, to one or more client devices, the titles for each of the distinguishing frames, which are extracted, via optical character recognition (OCR) technology, the distinguishing frames comprising the presentation slide in the JSON file format.

2. The method of claim 1 , wherein detecting the title within the frame comprises using one or more artificial intelligence (AI) models.

3. The method of claim 1 , wherein detecting the title within the frame comprises a plurality of You Only Look Once (YOLO) techniques.

4. The method of claim 1 , wherein detecting the title within the frame comprises:

dividing the frame into one or more grids of residual blocks.

5. The method of claim 4 , wherein detecting the title within the frame further comprises:

generating one or more segregated bounding boxes within the grids of residual blocks.

6. The method of claim 5 , wherein detecting the title within the frame further comprises:

determining, via intersection of union (IOU) techniques, a top bounding box with highest prediction confidence for the title from the segregated bounding boxes.

7. The method of claim 1 , wherein one of the title detection rules comprises determining that a position of the title within the frame matches with one or more prespecified title positions.

8. The method of claim 7 , wherein the prespecified title positions comprise one or more of: center, left, and top title positions corresponding to areas of the frame.

9. A communication system comprising:

one or more processors configured to:

receive video content of a communication session comprising a plurality of participants;

extract frames from the video content;

classify the frames of the video content;

identify one or more distinguishing frames comprising a presentation slide, wherein the one or more processors are further configured to:

remove a thumbnail of a participant video feed from each of the frames;

invert colors of each of the frames to obtain inverted frames to enhance text detection speed;

obtain a pixel value summation for each of the inverted frames;

calculate a pixel value summation difference between adjacent inverted frames; and

determine a distinguishing frame when the pixel value summation difference meets a threshold;

for each distinguishing frame comprising the presentation slide, detect a title within the frame, wherein the title is detected within the frame based on one or more title detection rules, wherein one or more candidate titles are determined prior to determining the title, wherein one of the title detection rules determines that a number of candidate titles determined for the frame does not exceed a threshold number of candidate titles, and wherein one of the title detection rules determines that a font size for the title meets or exceeds a threshold ratio of font size relative to other text within the frame;

format, via a processing engine, the titles for each distinguishing frame into a JavaScript Object Notation (JSON) file format; and

transmit, via the processing engine, to one or more client devices, the titles for each of the distinguishing frames, which are extracted via optical character recognition (OCR) technology, the distinguishing frames comprising the presentation slide in the JSON file format.

10. The communication system of claim 9 , wherein the one or more processors are further configured to:

filter out frames of the video content which are classified as a black frame, face frame, or demo frame.

11. The communication system of claim 9 , wherein the one or more processors are further configured to:

filter out frames which cannot be distinguished from neighboring frames based on the content of the frame.

12. The communication system of claim 9 , wherein the one or more processors are further configured to:

determine a layout analysis of each distinguishing frame comprising a presentation slide.

13. The communication system of claim 9 , wherein the title is detected within the frame using one or more artificial intelligence (AI) models.

14. The communication system of claim 9 , wherein the title is detected within the frame using a plurality of You Only Look Once (YOLO) techniques.

15. A non-transitory computer-readable medium comprising instructions, that when executed by a processor, cause the processor to perform operations comprising:

receiving video content of a communication session comprising a plurality of participants;

extracting frames from the video content;

classifying the frames of the video content;

identifying one or more distinguishing frames comprising a presentation slide, wherein identifying the one or more distinguishing frames comprises:

removing a thumbnail of a participant video feed from each of the frames;

inverting colors of each of the frames to obtain inverted frames to enhance text detection speed;

obtaining a pixel value summation for each of the inverted frames;

calculating a pixel value summation difference between adjacent inverted frames; and

determining a distinguishing frame when the pixel value summation difference meets a threshold;

for each distinguishing frame comprising the presentation slide, detecting a title within the frame, wherein detecting the title within the frame is based on one or more title detection rules, wherein one or more candidate titles are determined prior to determining the title, wherein one of the title detection rules comprises determining that a number of candidate titles determined for the frame does not exceed a threshold number of candidate titles, and wherein one of the title detection rules comprises determining that a font size for the title meets or exceeds a threshold ratio of font size relative to other text within the frame;

formatting, by a processing engine, the titles for each distinguishing frame into a JavaScript Object Notation (JSON) file format; and

transmitting, by the processing engine, to one or more client devices, the titles for each of the distinguishing frames, which are extracted via optical character recognition OCR) technology, the distinguishing frames comprising the presentation slide in the JSON file format.

16. The non-transitory computer-readable medium of claim 15 , wherein the operations further comprise:

determining a layout analysis of each distinguishing frame comprising the presentation slide.

17. The non-transitory computer-readable medium of claim 15 , wherein detecting the title within the frame comprises using one or more artificial intelligence (AI) models.

18. The non-transitory computer-readable medium of claim 15 , wherein detecting the title within the frame comprises a plurality of You Only Look Once (YOLO) techniques.

19. The non-transitory computer-readable medium of claim 15 , wherein the operations further comprise:

dividing the frame into one or more grids of residual blocks.

20. The non-transitory computer-readable medium of claim 19 , wherein the operations further comprise:

generating one or more segregated bounding boxes within the grids of residual blocks.

Assignments (2)

Continuity (1)

References Cited (26)