US 12,309,433 B2
On padding methods for neural network-based in-loop filter
Yue Li, San Diego, CA (US); Li Zhang, San Diego, CA (US); and Kai Zhang, San Diego, CA (US)
Assigned to Lemon Inc.
Filed by Lemon Inc., Grand Cayman (KY)
Filed on May 13, 2022, as Appl. No. 17/744,060.
Claims priority of provisional application 63/191,121, filed on May 20, 2021.
Prior Publication US 2022/0394309 A1, Dec. 8, 2022
Int. Cl. H04N 19/00 (2014.01); G06T 9/00 (2006.01); H04N 19/124 (2014.01); H04N 19/184 (2014.01); H04N 19/82 (2014.01)
CPC H04N 19/82 (2014.11) [G06T 9/002 (2013.01); H04N 19/124 (2014.11); H04N 19/184 (2014.11)] 19 Claims
OG exemplary drawing
 
1. A method of processing video data, comprising:
determining, in real time, padding dimensions for padding samples to be applied to a video unit of a video for in-loop filtering, wherein d1, d2, d3, and d4 represent the padding dimensions corresponding to top, bottom, left, and right boundaries of the video unit, respectively; and
performing a conversion between the video unit and a bitstream of the video based on the padding dimensions that were determined,
wherein at least one of the padding dimensions is based on a size of the video unit, a color format, a channel type, a slice type, or a partitioning tree type, and
wherein a padding method used to generate the padding samples outside the video unit is based on the channel type, the slice type, or on which temporal layer the video unit belongs to, and wherein a neural network (NN) filter is applied to the padding samples.