US 12,301,785 B1
Selective data encoding and machine learning video synthesis for content streaming systems and applications
Pratyush Mahapatra, Santa Clara, CA (US); and Ruthie Lyle, Durham, NC (US)
Assigned to NVIDIA Corporation, Santa Clara, CA (US)
Filed by NVIDIA Corporation, Santa Clara, CA (US)
Filed on Oct. 12, 2022, as Appl. No. 18/045,915.
Int. Cl. H04N 19/103 (2014.01); H04N 19/146 (2014.01); H04N 19/156 (2014.01)
CPC H04N 19/103 (2014.11) [H04N 19/146 (2014.11); H04N 19/156 (2014.11)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
receiving a first video content including a plurality of frames;
determining a first display configuration for a target display;
detecting a plurality of facial expressions within the plurality of frames;
identifying a first subset of facial expressions in the plurality of facial expressions that are visible on the target display based on the first display configuration and a second subset of facial expressions in the plurality of facial expressions that are not visible on the target display based on the first display configuration;
generating a first encoded file for an inference engine associated with the target display, wherein the inference engine comprises a neural network for generating synthetic video from encoded files, wherein the first encoded file includes information associated with the first subset of facial expressions and wherein the first encoded file does not include information associated with the second subset of facial expressions; and
sending the first encoded file to the inference engine.