US 11,991,368 B2
Video compression using deep generative models
Amirhossein Habibian, Amsterdam (NL); and Taco Sebastiaan Cohen, Amsterdam (NL)
Assigned to QUALCOMM INCORPORATED, San Diego, CA (US)
Filed by Qualcomm Incorporated, San Diego, CA (US)
Filed on Jul. 11, 2022, as Appl. No. 17/862,291.
Application 17/862,291 is a continuation of application No. 16/360,458, filed on Mar. 21, 2019, granted, now 11,388,416.
Prior Publication US 2022/0360794 A1, Nov. 10, 2022
Int. Cl. H04N 19/149 (2014.01); G06N 3/04 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2023.01); G06N 3/084 (2023.01); G06N 20/20 (2019.01); H04N 19/117 (2014.01); H04N 19/172 (2014.01); H04N 19/31 (2014.01)
CPC H04N 19/149 (2014.11) [G06N 3/045 (2023.01); G06N 3/084 (2013.01); G06N 20/20 (2019.01); H04N 19/117 (2014.11); H04N 19/172 (2014.11); H04N 19/31 (2014.11)] 30 Claims
OG exemplary drawing
 
11. A system for compressing video, comprising:
at least one processor configured to:
receive video content for compression;
encode, into a code in a latent code space, the video content through an auto-encoder implemented by a first artificial neural network configured to execute on the at least one processor, wherein the code in the latent code space represent a lossy encoding of the received video content such that decoding the code results in an approximation of the received video content;
generate a losslessly compressed version of the code in the latent code space through a probabilistic model implemented by a second artificial neural network configured to execute on the at least one processor; and
output the losslessly compressed version of the code in the latent code space for transmission; and
a memory coupled to the at least one processor.