Detecting and framing a subject of interest in a teleconference
View Patent ↗A method of view selection in a teleconferencing environment includes receiving a frame of image data from an optical sensor such as a camera, detecting one or more conference participants within the frame of image data, and identifying an interest region for each of the conference participants. Identifying the interest region comprises estimating head poses of participants to determine where a majority of the participants are looking and determining if there is an object in that area. If a suitable object is in the area at which the participants are looking, such as a whiteboard or another person, the image data corresponding to the object will be displayed on a display device or sent to a remote teleconference endpoint.
1. A method of rendering a view of a subject of interest in a teleconferencing environment, the method comprising:
determining that the subject of interest is a person;
determining a centroid corresponding to the subject of interest;
determining a gaze of the subject of interest relative to a lens of a camera used to capture the view containing the subject of interest, the lens having a central region;
determining that the gaze of the subject of interest is offset from the central region by at least fifteen degrees; and
positioning the subject of interest within the view such that the centroid of the subject of interest is offset from a center of the view.
2. The method of claim 1 , wherein positioning the subject of interest within the view such that the centroid of the subject of interest is offset from a center of the view comprises bounding the subject of interest within a rectangular region having a horizontal width and placing the subject of interest within the rectangular region such that the centroid of the subject of interest is separated horizontally from a boundary towards which the gaze is more directed by a distance corresponding to between two-thirds and one-half of the horizontal width.
3. A teleconferencing endpoint that comprises:
a camera configured to provide a frame of image data;
one or more processors coupled to the camera to render a view of a subject of interest within the frame of image data by:
determining that the subject of interest is a person;
determining a centroid corresponding to the subject of interest;
determining a gaze of the subject of interest relative to a lens of the camera, the lens having a central region;
determining that the gaze of the subject of interest is offset from the central region by at least fifteen degrees; and
positioning the subject of interest within the view such that the centroid of the subject of interest is offset from a center of the view.
4. The teleconferencing endpoint of claim 3 , wherein the processor is further configured to position the subject of interest within the view such that the centroid of the subject of interest is offset from a center of the view by:
bounding the subject of interest within a rectangular region having a horizontal width; and
placing the subject of interest within the rectangular region such that the centroid of the subject of interest is separated horizontally from a boundary of the rectangular region towards which the gaze is more directed by a distance corresponding to between three-fourths and one-half of the horizontal width.
5. A computer readable non-transitory information storage medium storing executable instructions that cause one or more processors to render a view of a subject of interest by configuring the one or more processors to:
determine that the subject of interest is a person;
determine a centroid corresponding to the subject of interest;
determine a gaze of the subject of interest relative to a lens of a camera used to capture the view containing the subject of interest, the lens having a central region;
determine that the gaze of the subject of interest is offset from the central region by at least fifteen degrees; and
position the subject of interest within the view such that the centroid of the subject of interest is offset from a center of the view.
6. The computer readable non-transitory information storage medium of claim 5 , wherein said positioning separates the centroid of the subject of interest horizontally away from a view boundary toward which their gaze is directed by a distance between one half and two thirds of a total width of the view.