Granted Patent B2

US 12,536,764 · App. 17/971,530 · Granted Jan 27, 2026

Inventors: Nhu Quynh Pham Nguyen (Austin, TX); Kishore Venkat Rao Goka (Leander, TX); Joshua Raglin (Austin, TX); Dhaval Patel (Austin, TX); Sakshi Gupta (Round Rock, TX); Venkateswarlu Lakkamraju (Tuam, IE); Krishna Balusu (Round Rock, TX)

G06V10/245G06T7/13G06T7/62G06T7/70G06V10/25G06V40/161G06T2207/30201

Abstract

Claims (62)

1 . A method comprising:

obtaining, by applying a head detection model to an image in an incoming video stream, a plurality of bounding boxes from the image;

identifying, using a left edge of a buffer around a leftmost one of the bounding boxes and a right edge of a buffer around a rightmost one of the bounding boxes, a group horizontal size that extends along a horizontal axis from the left edge to the right edge;

identifying, using a top edge of a buffer around a topmost one of the bounding boxes and a bottom edge of a buffer around a bottommost one of the bounding boxes, a group vertical size that extends along a vertical axis from the top edge to the bottom edge;

dividing, to calculate a ratio of a group size in the image, a length of the group horizontal size by a length of the group vertical size; and

identifying, by comparing the ratio of the group size in the image to an aspect ratio of an outgoing video stream, an orientation of a group in the image.

2 . The method of claim 1 , wherein obtaining head detection information comprises:

extracting the image from the incoming video stream;

processing the image to obtain a plurality of head detection bounding boxes for the image of the incoming video stream; and

adding a buffer around each head detection bounding box of the plurality of head detection bounding boxes to obtain the head detection information.

3 . The method of claim 2 , wherein the plurality of head detection bounding boxes are around a plurality of members of the group.

4 . The method of claim 1 , wherein identifying the group vertical size comprises:

calculating the group vertical size as a distance from the top edge to the bottom edge, wherein determining the orientation of the group is based on the group vertical size.

5 . The method of claim 4 , wherein identifying the group horizontal size comprises:

calculating the group horizontal size as a distance from the left edge to the right edge,

wherein determining the orientation of the group is performed based on a comparison of the group vertical size to the group horizontal size.

6 . The method of claim 1 , further comprising:

creating a group frame definition by calculating a group vertical center as a center of a vertical line from the top edge to the bottom edge,

wherein the group vertical center is used to create the group frame definition.

7 . The method of claim 6 , wherein creating the group frame definition comprises:

positioning a group framing template to align with the group vertical center based on the orientation indicating that the group is vertical.

8 . The method of claim 1 , further comprising:

creating a group frame definition by calculating a group horizontal center as a center of a horizontal line from the left edge to the right edge,

wherein the group horizontal center is used to create the group frame definition.

9 . The method of claim 8 , further comprising:

positioning a group framing template to align with the group horizontal center based on the orientation indicating that the group is horizontal.

10 . The method of claim 9 , further comprising:

adding a buffer around the group framing template on the positioned group framing template to generate the group frame definition.

11 . The method of claim 1 , further comprising:

making a detection of a member of the group moving outside of a first group frame defined by a group frame definition;

initiating a timer responsive to the detection; and

triggering a reframing responsive to not detecting the member in a second group frame defined by the group frame definition at an elapse of a threshold amount of time according to the timer.

12 . A method comprising:

obtaining, using a head detection model and for an image in a video stream, head detection information on the video stream;

identifying a group vertical size of a group on a vertical axis using the head detection information;

identifying a group horizontal size of the group on a horizontal axis using the head detection information;

determining an orientation of the group using the group horizontal size and group vertical size;

creating a group frame definition using the orientation;

outputting the video stream using the group frame definition;

making a detection of a group size decreasing, wherein the group size is at least one of the group vertical size and the group horizontal size;

initiating a timer responsive to the detection; and

triggering a reframing responsive to the group size remaining decreased at an elapse of a threshold amount of time according to the timer.

13 . The method of claim 1 , further comprising:

making a detection of a member of the group leaving a room;

initiating a timer responsive to the detection; and

triggering a reframing responsive to not detecting the member in the room at an elapse of a threshold amount of time according to the timer.

14 . A system comprising:

a video processor to:

obtain, by applying a head detection model to an image in an incoming video stream, a plurality of bounding boxes from the image;

identify, using a left edge of a buffer around a leftmost one of the bounding boxes and a right edge of a buffer around a rightmost one of the bounding boxes, a group horizontal size that extends along a horizontal axis from the left edge to the right edge;

identify, using a top edge of a buffer around a topmost one of the bounding boxes and a bottom edge of a buffer around a bottommost one of the bounding boxes, a group vertical size that extends along a vertical axis from the top edge to the bottom edge;

divide, to calculate a ratio of a group size in the image, a length of the group horizontal size by a length of the group vertical size; and

identify, by comparing the ratio of the group size in the image to an aspect ratio of an outgoing video stream, an orientation of a group in the image.

15 . The system of claim 14 , further comprising:

an image framing processor to output the outgoing video stream.

16 . The system of claim 15 , wherein the outgoing video stream comprises the group in the image.

17 . The system of claim 14 , wherein the video processor comprises:

a group framing template to position the incoming video stream.

18 . The system of claim 14 , wherein the group in the image is a group of people.

19 . The system of claim 14 , further comprising:

a camera to capture the incoming video stream.

20 . The system of claim 19 , wherein the video processor is within a housing of the camera.

Assignments (2)

Continuity (2)

References Cited (14)