US 12,383,838 B2
Game event recognition for user generated content
Suresh Yerva, Beglauru (IN); and Stephen Holmes, Fort Collins, CO (US)
Assigned to Nvidia Corporation, Santa Clara, CA (US)
Filed by Nvidia Corporation, Santa Clara, CA (US)
Filed on May 15, 2023, as Appl. No. 18/317,639.
Application 18/317,639 is a continuation of application No. 17/075,377, filed on Oct. 20, 2020, granted, now 11,648,481.
Prior Publication US 2023/0277944 A1, Sep. 7, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. A63F 13/77 (2014.01); A63F 13/537 (2014.01); H04N 21/439 (2011.01); H04N 21/478 (2011.01)
CPC A63F 13/77 (2014.09) [A63F 13/537 (2014.09); H04N 21/4394 (2013.01); H04N 21/4781 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
analyzing, by a neural network, a first region of one or more frames of video content associated with a first element, wherein the first region is associated with a first level of a region hierarchy comprising multiple levels, each level defining one or more regions of the video content, the neural network trained to recognize objects using feature data extracted from video frames;
in response to determining that the first element in the first region has a first state associated with a type of event, analyzing, using the neural network, at least one second region of the one or more frames of video content, the second region associated with a second element and a second level of the region hierarchy;
wherein the analyzing is triggered by a determination that the second level is lower in the region hierarchy than the first level;
wherein at least one region associated with a lower level of the region hierarchy is analyzed in response to determining that an element in at least one region associated with a higher level of the region hierarchy is in a specific state associated with the type of event;
determining, using the neural network, that the second element has at least one second state associated with the type of event;
in response to determining that the first element is in the first state and the second element is in the least one second state associated with the type of event, identifying that the type of event occurred; and
providing a portion of the video content, representative of the type of event, for display on a client device.