Abstract: In one implementation, we propose a block-based end-to-end image and video compression method that takes non-overlapping or overlapping split blocks of input images or frames of videos as input. Then, the proposed decoder network reconstructs non-overlapped split blocks of the input. We also introduce an intra prediction method to reduce spatial redundancy in the latent space, i.e., one or more previously decoded latent tensors from neighboring blocks are used as references to predict the current block's latent tensor. Additionally, the decoder can selectively complete the pixel reconstruction process for decoded latent blocks without causing any error drift to neighboring blocks since the prediction is made in the latent space. Enabling and disabling the pixel reconstruction can be signaled by the encoder as metadata in the bitstream or decided at the decoding stage using a computer vision task.

Abstract: A block of video data is split using one or more of several possible partition operations by using the partitioning choices obtained through use of a deep learning-based image partitioning. In at least one embodiment, the block is split in one or more splitting operations using a convolutional neural network. In another embodiment, inputs to the convolutional neural network come from pixels along the block's causal borders. In another embodiment, boundary information, such as the location of partitions in spatially neighboring blocks, is used by the convolutional neural network. Methods, apparatus, and signal embodiments are provided for encoding.

Abstract: A video encoder or decoder processes portions of video using intra coding and with variable block shapes. Under the general aspects, intra prediction directions are adapted to the block shape or aspect ratio. The general aspects described adaptively replace several conventional angular intra prediction modes with wide-angle intra prediction modes for non-square blocks. The total number of angular directions for any block can remain the same as in prior video coding standards or can be reduced or expanded. To keep the same number of prediction directions, directions along a shorter rectangular edge of a target block are removed but replaced with addition directions along the longer edge of the rectangular target block. For directions with wide angles, reference sample smoothing is performed. Signaling can be provided to inform a corresponding decoder regarding the prediction modes and directions.

Abstract: Decoding or encoding picture information can involve determining a value associated with a prediction of a pixel of a block of picture information, wherein, the prediction is based on an intra-prediction along a non-diagonal direction, determining, from among a plurality of decoded pixels neighboring the block on the top and the left, at least four reference samples for the block based on the non-diagonal direction, and wherein the determining of the value comprises an interpolation based on a quadratic model and the four reference samples; and decoding or encoding at least a portion of the picture information based on the value.

Abstract: A method for video decoding is disclosed. A directional intra prediction mode is decoded for a block of a picture, the directional intra prediction mode having a direction. Based on the directional intra prediction mode, a first predictor and a second predictor for a sample in the current block are accessed, the first and second predictors being on a line at least approximating the direction. A sample value of the sample is predicted by interpolation using the first and second predictors, responsive to a difference between the second and first predictors, wherein the difference is scaled by at least a ratio that is based on a location of the sample in the block, and wherein a denominator of the ratio is a power of two regardless of the location of the sample. The sample of the block is reconstructed based on the predicted sample value.

Abstract: A decoding method for decoding blocks of a picture part is disclosed that comprises: decoding a plurality of candidate sets of filter parameters from a bitstream; decoding for a current block of said picture an index identifying one of the candidate sets of filter parameters from said bitstream; decoding said current block from said bitstream; and filtering said decoded current block with the candidate set of filter parameters identified by said decoded index.

Abstract: Video signal coding and decoding functions can generate lists of potential candidates to use in coding and decoding, for example, predictors. Video signal coding component candidate undergo operations before potential inclusion in candidate lists. The candidates are checked after being modified by the operations to see if other equal candidates are already in the candidate list. If equal candidates are not in the list, the modified candidates are added to the candidate list. If equal candidates are already in the list, the modified candidates are not added to the list. Operations that can be performed comprise rounding and clipping.

Abstract: Different implementations are described, particularly implementations for video encoding and decoding based on linear weighted intra prediction, also called matrix based intra prediction, are presented. Accordingly, for a block being encoded or decoded in linear weighted intra prediction, obtaining intra predicted samples from at least two matrix-vector products between at least two selected weight matrices of reduced size and a set of neighboring reference samples. Advantageously, such arrangement allows to reduce the amount of memory for storing data and to reduce the complexity of the intra prediction samples computation.

Abstract: Systems, methods, and instrumentalities are disclosed for video encoding and/or video decoding using artificial neural networks (e.g., convolutional neural networks or recurrent neural networks), attention, and/or attention with spatial attributes. For example, an apparatus may be configured to perform one or more of the following: obtaining a context block, a current block, and a latent vector associated with the context block; performing at least one convolution on the context block, the reference block, and the latent vector; generating motion flow data associated with the current block based on the at least one convolution; or generating a bitstream the comprises an indication of the motion flow data. The motion flow data may be quantized. The generated bitstream may comprise an indication of the quantized motion flow data.

Abstract: A block of video data is split using one or more of several possible partition operations by using the partitioning choices obtained through use of a texture-based image partitioning. In at least one embodiment, the block is split in one or more splitting operations using a convolutional neural network. In another embodiment, inputs to the convolutional neural network come from pixels along the block's causal borders. In another embodiment, boundary information, such as the location of partitions in spatially neighboring blocks, is used by the texture analysis. Methods, apparatus, and signal embodiments are provided for encoding.

Abstract: A method includes decoding a first portion of video data consisting of an information set consisting of information representative of a plurality of candidate sets of sample adaptive offset filter parameters. A set of sample adaptive offset parameters comprises information representative of an offset to be added to a sample value of a current block. A first syntax element representing an index that identifies a first candidate set of the plurality of candidate sets of sample adaptive offset filter parameters is decoded from the video data for the current block. The current block is decoded from a second portion of the video data different from the first portion. The decoded current block is filtered with the first candidate set identified by the index. The first portion is signaled with syntax elements at a level of a block of a group of blocks comprising the current block decoded first in decoding order.

Abstract: A method and apparatus for performing prediction for encoding or decoding uses intra prediction with sub-partitions. The sub-partitions are oriented either horizontally or vertically and can use wide-angle modes different than that of the video block they originate from. Reference samples for a sub-partition are those of the video block when reference samples for the sub-partition are not available, such as due to direction. In an embodiment, when a sub-partition is square, conventional intra prediction directions are used. Reference samples can be used from a block above and right of the video block or left and below the video block using a mapping along a prediction direction, vertically, or horizontally.

Abstract: A method and apparatus for performing intra prediction for encoding or decoding adds multiple prediction modes to those of prior intra prediction schemes. A horizontal mode, a vertical mode, and a diagonal mode are added. Reference pixels for each of the added modes can come from multiple pixel positions away from a block to be encoded or decoded. Reference pixels to the left and above a block are reconstructed from those already coded. Reference pixels from right of, and below, a block are estimated or extrapolated and can be based on other pixels around the block. An index can be sent in a bitstream indicating which prediction was used.

Abstract: A block of video data is split using one or more of several possible partition operations by using the partitioning choices obtained through use of a texture-based image partitioning. In at least one embodiment, the block is split in one or more splitting operations using a convolutional neural network. In another embodiment, inputs to the convolutional neural network come from pixels along the block's causal borders. In another embodiment, boundary information, such as the location of partitions in spatially neighboring blocks, is used by the texture analysis. Methods, apparatus, and signal embodiments are provided for encoding.

Abstract: Methods and apparatuses for encoding/decoding an image or a video using neural network are disclosed. In some embodiments, side-information is decoded from a bitstream that allows for adapting a first neural network-based decoder, the decoded side-information and coded data representative of an image or a video obtained from the bitstream or a separate bitstream are provided as inputs to the first neural-network-based decoder and a reconstructed image or video is obtained from an output of the first neural network-based decoder.

Abstract: A method for video decoding is disclosed. A directional intra prediction mode is decoded for a block of a picture, the directional intra prediction mode having a direction. Based on the directional intra prediction mode, a first predictor and a second predictor for a sample in the current block are accessed, the first and second predictors being on a line at least approximating the direction. A sample value of the sample is predicted by interpolation using the first and second predictors, responsive to a difference between the second and first predictors, wherein the difference is scaled by at least a ratio that is based on a location of the sample in the block, and wherein a denominator of the ratio is a power of two regardless of the location of the sample. The sample of the block is reconstructed based on the predicted sample value.

Abstract: Methods and apparatus for using wide-angle intra prediction for rectangular blocks enable greater prediction angles. Wide-angle intra prediction enables intra prediction direction angles beyond the conventional 45 and ?135 degrees. In one embodiment, when a video block to be coded or decoded is non-square, additional intra prediction directions are enabled in the direction of the longer block edge and more reference samples are available along that edge. An index is used to indicate the prediction direction and can be adapted according to the additional intra predictions in the longer direction, with correspondingly fewer prediction directions along the shorter block edge. This preserves the number of prediction modes that need to be indexed but allows their angles to correspond to the shape of the block.

Abstract: A method for video decoding is disclosed. A first reconstructed version of an image block of an encoded video is accessed. The first reconstructed version of said image block is filtered by a first neural network to form a second reconstructed version of said image block to be used as reference. The second reconstructed version of said image block is filtered by a second neural network to form a third reconstructed version of said image block to be displayed. The first and second neural networks are trained jointly.

Abstract: A method for encoding a block is disclosed. To this aim, a split mode is determined based on a rate-distortion optimization using a texture-based split prediction set obtained for the block. As an example, the split mode is determined by adapting the texture-based split prediction set according to at least one of a binary or triple split mode non-redundancy constraint or a heuristic-based split mode set pruning. The block is finally encoded using the determined split mode.

Abstract: A method and apparatus for performing intra prediction mode partitioning predicts pixels of a video coding block using a plurality of intra prediction modes over a plurality of regions of the video coding block. Predictions can be based on reference arrays using at least one reference line above the block to be coded and/or using at least one reference column to the left of the block to be coded. Different prediction modes are used in different regions of the coding block. In at least one embodiment, the video coding block is split into a plurality of partitions in addition to having a plurality of intra prediction modes over different regions of the block. Signaling enables a decoder to determine the number of regions within a coding block. In at least another embodiment, prediction modes allowable within regions of a video coding block differ by one angular position.