An iterative training of neural networks for video coding and decoding using intra prediction is provided that finds a tradeoff between an extreme genericity and an extreme specialization to a codec for the trained neural networks. At the first iteration, the set of neural networks is trained following a partitioning approach. Then, for several iterations, the set of neural networks is inserted into the codec, and pairs of a block and its context are extracted from the partitioning of images via the codec with a single additional neural network-based mode then, the neural networks are retrained on these pairs. This way, from the second iteration, the neural networks learn an intra prediction diverging from that in the codec while still being valuable for the codec in terms of rate-distortion performance.

A method, apparatus, and non-transitory computer-readable medium for adaptive neural image compression by meta-learning using substitute QF settings, which includes generating one or more substitute quality factors via a plurality of iterations using the original quality factors, wherein the substitute quality factors are a modified version of the original quality factors and are associated with a single instance of neural network loop filtering model. The approach may further include determining a neural network based loop filter comprising neural network based loop filter parameters and a plurality of layers, wherein the neural network based loop filter parameters include shared parameters and adaptive parameters, and may further include generating enhanced video data, based on the one or more substitute quality factors and the input video data, using the neural network based loop filter.

Disclosed are a method for determining a color difference component quantization parameter and a device using the method. Method for decoding an image can comprise the steps of: decoding a color difference component quantization parameter offset on the basis of size information of a transform unit; and calculating a color difference component quantization parameter index on the basis of the decoded color difference component quantization parameter offset. Therefore, the present invention enables effective quantization by applying different color difference component quantization parameters according to the size of the transform unit when executing the quantization.

A video system that applies constraints on block partitioning is provided. The system receives a partitioning control parameter from a bitstream specifying a maximum block size for enabling ternary-tree split that is constrained to be 64 or smaller. The system receives data from a bitstream for a block of pixels to be decoded as a current block of a current picture of a video. The system splits the current block into one or more partitions recursively, wherein ternary split is disallowed for a partition of the current block unless the partition is less than or equal to the maximum block size. The system reconstructs the one or more partitions of the current block.

A system and method for transmission of a video stream are provided. The system may include: an encoder adapted to generate a video stream comprising a plurality of encoded frames, encoded according to at least one encoding parameter; a comparator in communication with the encoder, the comparator adapted to compare encoded frames of the plurality of encoded frames with input frames to determine a fitness metric reflective of visual quality of the encoded frames; and a controller in communication with the comparator, the controller adapted to adjust the at least one encoding parameter based on the fitness metric.

A video encoder for encoding a video into a data stream using motion compensated prediction for inter predicted blocks, comprising a hypothesis number control, configured to control a number of prediction hypotheses of the inter predicted blocks within a predetermined portion of the video to meet a predetermined criterion.

An apparatus for decoding a current block in a scan order identifies a current entropy coded transform coefficient from the current block; determines, based on the scan order and using a scan order distance table, a first scan order distance that is a difference between a first scan order location corresponding to the current entropy coded transform coefficient and a second scan order location corresponding to a first context coefficient; identifies, using the first scan order distance, a first location into a context coefficient register; identifies, at the first location of the context coefficient register, a first context coefficient value; identifies a probability distribution using at least the first context coefficient value; entropy decodes the current entropy coded transform coefficient using the probability distribution to obtain an entropy decoded current transform coefficient; and includes the entropy decoded current transform coefficient in an output video stream.

An apparatus for decoding a current block in a scan order identifies a current entropy coded transform coefficient from the current block; determines, based on the scan order and using a scan order distance table, a first scan order distance that is a difference between a first scan order location corresponding to the current entropy coded transform coefficient and a second scan order location corresponding to a first context coefficient; identifies, using the first scan order distance, a first location into a context coefficient register; identifies, at the first location of the context coefficient register, a first context coefficient value; identifies a probability distribution using at least the first context coefficient value; entropy decodes the current entropy coded transform coefficient using the probability distribution to obtain an entropy decoded current transform coefficient; and includes the entropy decoded current transform coefficient in an output video stream.

A method for video processing includes performing a conversion between a current video block of a video and a bitstream representation of the current video block using a matrix based intra prediction (MIP) mode, where the conversion includes performing the upsampling operation in which the final prediction block is determined by using a reduced prediction block of the current video block and by using reconstructed neighboring samples of the current video block according to a rule, and where the reduced prediction block is obtained by performing the matrix vector multiplication operation on reduced boundary samples of the current video block.

Several systems, methods and integrated circuits capable of reducing blocking artifacts in video data are disclosed. In an embodiment, a system for reducing blocking artifacts in video data includes a processing module and a deblocking module. The deblocking module comprises a luma deblocking filter and a chroma deblocking filter configured to filter an edge between adjacent blocks associated with the video data, where a block of the adjacent blocks corresponds to one of a prediction block and a transform block. The processing module is communicatively associated with the deblocking module and is operable to configure at least one filter coefficient corresponding to the chroma deblocking filter based on one or more filter coefficients corresponding to the luma deblocking filter. The processing module is further configured to cause the chroma deblocking filter to filter the edge between the adjacent blocks based on the configured at least one filter coefficient.