Back to results
US 12,587,646 B2Granted
Network based image filtering for video coding
AssigneeBeijing Dajia Internet Information Technology Co., Ltd.
InventorsWei Chen, Xiaoyu Xiu, Yi-Wen Chen, Hong-Jheng Jhu, Che-Wei Kuo, Xianglin Wang, Bing Yu
Life of this patent
Estimating the term…
Abstract
A method and an apparatus for image filtering in video coding using a neural network are provided. The method includes: loading, a plurality of quantization parameter (QP) map (QpMap) values at a plurality of QpMap channels into the neural network; and adjusting, according to a QP scaling factor, the plurality of QpMap values for the neural network to learn and filter a current input frame to the neural network.
Claims14
A method for image filtering in video coding using a neural network, comprising:
- loading a plurality of quantization parameter (QP) map (QpMap) values at a plurality of QpMap channels into the neural network;
- adjusting, according to a QP scaling factor, the plurality of QpMap values for the neural network to learn and filter a current input frame to the neural network;
- loading a plurality of input patches associated with the current input frame into the neural network at a plurality of input channels;
- and
- combining the plurality of QpMap channels with the plurality of input channels,
- wherein the plurality of input patches comprise a plurality of first input patches loaded at a first input channel, a plurality of second input patches loaded at a second input channel, a plurality of third input patches loaded at a third input channel;
- wherein the plurality of QpMap channels comprise a first QpMap channel for the first input channel, a second QpMap channel for the second input channel, and a third QpMap channel for the third input channel;
- and
- wherein the method further comprises:
- obtaining an adjusted second input channel and an adjusted third input channel by respectively up-sampling the second input channel and the third input channel;
- and
- interleaving the first QpMap channel, the second QpMap channel, and the third QpMap channel with the first input channel, the adjusted second input channel, and the adjusted third channel.
9 more elements
An apparatus for image filtering in video coding using a neural network, comprising:
- one or more processors;
- and
- a memory configured to store instructions executable by the one or more processors,
- wherein the one or more processors, upon execution of the instructions, are configured to perform operations of:
- loading a plurality of quantization parameter (QP) map (QpMap) values at a plurality of QpMap channels into the neural network;
- adjusting, according to a QP scaling factor, the plurality of QpMap values for the neural network to learn and filter a current input frame to the neural network;
- loading a plurality of input patches associated with the current input frame into the neural network at a plurality of input channels;
- and
- combining the plurality of QpMap channels with the plurality of input channels,
- wherein the plurality of input patches comprise a plurality of first input patches loaded at a first input channel, a plurality of second input patches loaded at a second input channel, a plurality of third input patches loaded at a third input channel;
- wherein the plurality of QpMap channels comprise a first QpMap channel for the first input channel, a second QpMap channel for the second input channel, and a third QpMap channel for the third input channel;
- and
- wherein the one or more processors are configured to perform further operations of:
- obtaining an adjusted second input channel and an adjusted third input channel by respectively up-sampling the second input channel and the third input channel;
- and
- interleaving the first QpMap channel, the second QpMap channel, and the third QpMap channel with the first input channel, the adjusted second input channel, and the adjusted third channel.
13 more elements
A non-transitory computer-readable storage medium storing computer-executable instructions that, when executed by one or more computer processors, causing the one or more computer processors to perform operations of:
- loading a plurality of quantization parameter (QP) map (QpMap) values at a plurality of QpMap channels into the neural network;
- adjusting, according to a QP scaling factor, the plurality of QpMap values for the neural network to learn and filter a current input frame to the neural network;
- loading a plurality of input patches associated with the current input frame into the neural network at a plurality of input channels;
- and
- combining the plurality of QpMap channels with the plurality of input channels,
- wherein the plurality of input patches comprise a plurality of first input patches loaded at a first input channel, a plurality of second input patches loaded at a second input channel, a plurality of third input patches loaded at a third input channel;
- wherein the plurality of QpMap channels comprise a first QpMap channel for the first input channel, a second QpMap channel for the second input channel, and a third QpMap channel for the third input channel;
- and
- wherein the one or more computer processors are caused to perform further operations of:
- obtaining an adjusted second input channel and an adjusted third input channel by respectively up-sampling the second input channel and the third input channel;
- and
- interleaving the first QpMap channel, the second QpMap channel, and the third QpMap channel with the first input channel, the adjusted second input channel, and the adjusted third channel.
9 more elements
Dependencies are inferred from the claim text.