Abstract: Techniques related to video coding include multiple channel video coding with cross-channel referencing.

Abstract: Lossy or lossless compression and transmission, comprising the steps of: (i) receiving an input image; (ii) encoding it to produce a y latent representation; (iii) encoding the y latent representation to produce a z hyperlatent representation; (iv) quantizing the z hyperlatent representation to produce a quantized z hyperlatent representation; (v) entropy encoding the quantized z hyperlatent representation into a first bitstream, (vi) processing the quantized z hyperlatent representation to obtain a location entropy parameter ?y, an entropy scale parameter ?y, and a context matrix Ay of the y latent representation; (vii) processing the y latent representation, the location entropy parameter ?y and the context matrix Ay, to obtain quantized latent residuals; (viii) entropy encoding the quantized latent residuals into a second bitstream; and (ix) transmitting the bitstreams.

Abstract: The present invention relates to a method and apparatus for performing encoding and decoding using a variably-sized quantization coefficient group, in quantization coefficient group encoding and decoding of video compression technology.

Abstract: Methods and apparatus for video processing, including coding and decoding, are described. One example video processing method includes performing a conversion between a video having one or more video layers and a bitstream of the video according to a format rule, and wherein the format rule specifies that a value of a variable indicating whether to output a picture in an access unit is determined based on a flag indicating whether to output another picture in the access unit.

Abstract: Devices, systems and methods for digital video coding, which includes matrix-based intra prediction methods for video coding, are described. In a representative aspect, 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 in which a prediction block of the current video block is determined by performing, on previously coded samples of the video, a boundary downsampling operation, followed by a matrix vector multiplication operation, and selectively followed by an upsampling operation, where the performing the conversion includes deriving, according to a rule, boundary samples by applying a left bit shift operation or a right bit shift operation on a sum of at least one reference boundary sample, and where the rule determines whether to apply the left bit shift or the right bit shift operation.

Abstract: A method for deblocking at least one boundary of a coding unit. The method includes determining that the coding unit uses a sub-block transform, wherein the sub-block transform generates a transform sub-block boundary within the coding unit, thereby forming at least a transform sub-block within the coding unit; as a result of determining that the coding unit uses the sub-block transform, determining a maximum filter length based on a dimension of the transform sub-block within the coding unit, wherein the maximum filter length indicates a maximum number of samples to modify when deblocking a boundary of the coding unit; and deblocking the boundary of the coding unit based on the determined maximum filter length.

Abstract: A method of video processing includes determining a classification of samples of a block in a video picture of a video according to a rule, wherein the video picture is divided into multiple regions, and wherein the rule disallows use of neighboring samples of a current sample across one or more boundaries of the multiple regions for the classification. The one or more boundaries of the multiple regions includes a virtual boundary for an adaptive loop filtering (ALF) process. The method also includes performing a conversion between the block and a bitstream representation of the video by selectively applying the ALF process according to the classification.

Abstract: An image processing device including an acquiring section configured to acquire quantization matrix parameters from an encoded stream in which the quantization matrix parameters defining a quantization matrix are set within a parameter set which is different from a sequence parameter set and a picture parameter set, a setting section configured to set, based on the quantization matrix parameters acquired by the acquiring section, a quantization matrix which is used when inversely quantizing data decoded from the encoded stream, and an inverse quantization section configured to inversely quantize the data decoded from the encoded stream using the quantization matrix set by the setting section.

Abstract: An image decoding method performed by a decoding device, according to the present document, comprises the steps of: obtaining image information; and generating a reconstructed picture on the basis of the image information.

Abstract: A method for encoding an image using intra prediction can include determining an intra prediction mode of a current block; generating a mode group indicator and a prediction mode index; constructing a most probable mode (MPM) group including three intra prediction modes determined using a left and an above intra prediction mode of the current block; determining whether all reference pixels of the current block are available; generating reference pixels if one or more reference pixels of the current block are unavailable; adaptively filtering the reference pixels based on the intra prediction mode and the size of the current block, wherein if the size of the current block is 4×4, the reference pixels are not filtered, and if the size of the current block is larger than 4×4, the number of intra prediction mode where the reference pixels are filtered increases as the current block becomes larger; generating the prediction block using the reference pixels.

Abstract: Techniques are described herein for processing video data. For instance, a technique can include encoding one or more frames of video data using a video encoder, the video encoder including at least a quantization process; determining an actual bit rate of the encoded one or more frames; predicting an estimated bit rate using an encoder proxy, the encoder proxy including a statistical model for estimating a bit rate of the encoded one or more frames; determining, using the encoder proxy, a gradient of the estimated bit rate; and training the encoder proxy to predict the estimated bit rate based on the actual bit rate, the estimated bit rate, and the gradient.

Abstract: A system-on-chip which includes a video codec including a deblocking filter includes a motion estimator that calculates a motion vector of an input image, a motion compensator that compensates for a motion of the input image by using the motion vector, and a parameter generator that allows image data, in which the motion is compensated, to be transferred to and filtered by the deblocking filter of the video codec.

Abstract: A method includes receiving an input video stream and scaling the input video stream into two or more spatial layers. For each spatial layer, the method also includes generating a temporal layer prediction pattern by: obtaining a temporal base layer for a corresponding spatial layer; identifying, based on the temporal base layer, a plurality of temporal layers and a plurality of temporal time slots during a temporal period; and aligning the temporal base layer for the corresponding spatial layer with one of the temporal time slots during the temporal period. Each temporal time slot is associated with one of the temporal base layer or one of the plurality of temporal layers for the corresponding spatial layer. The temporal base layer for each corresponding spatial layer is aligned with a different temporal time slot than each other temporal base layer for each other corresponding spatial layer.

Abstract: A method of decoding an image with latent feature-domain intra-prediction is performed by at least one processor and includes receiving a set of latent blocks and for each of the blocks in the set of latent blocks: predicting a block, based on a set of previously recovered blocks; receiving a selection signal indicating a currently recovered block, based on the selection signal performing one of (1) and (2): (1) generating a compact residual, a set of residual context parameters, a decoded residual, and generating a first decoded block; (2) generating a second decoded block, based on a compact representation block and a set of context parameters. The method further includes generating a set of recovered blocks comprising each of the currently recovered blocks; generating a recovered latent image by merging all the blocks in the set of recovered blocks; and decoding the recovered latent image, to obtain a reconstructed image.

Abstract: In a method for video decoding, coded information of a current block is received. Dimension information and partition information of the current block is determined based on the coded information, where the dimension information indicates at least one of a height, a width, and a size of the current block, and the partition information indicates a binary tree depth of a coding unit partition of the current block. A partition mode of the current block is determined as one of an intra sub-partition mode and a coding unit partition based on (i) comparisons between values of the height, the width, and the size of the current block and respective threshold values, or (i) a ratio between the value of the height and the value of the width of the current block. The current block is decoded based on the determined partition mode of the current block.

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 accessing a bitstream 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 of filter parameters from a bitstream identified by said decoded index decoding for a current block an index.

Abstract: A decoding method is disclosed that comprises: —decoding (S120) a block of a picture, said block being intra predicted from a plurality of reference blocks; —determining (S130) a frontier inside the decoded block based on a frontier existing between said plurality of reference blocks; and —filtering (S140) with a de-artifacting filter along the frontier determined inside the decoded block.

Abstract: A serverless computing system is configured to provide access to a machine learning model by at least associating an endpoint, comprising code that accesses the machine learning model, with an extension that interfaces between a serverless compute architecture and the endpoint. A request to perform an inference is received by the system and processed by using the serverless compute architecture to execute a compute function. The compute function cases the extension to interface with the endpoint to cause the machine learning model to perform the inference.

Abstract: An electronic device of decoding a current video block in a video picture is provided. The electronic device determines whether a partitioning scheme of the current video block is a single tree partitioning or a dual tree partitioning. The electronic device selects, based on the partitioning scheme, one derivation from candidate derivations to compute a first predictive quantization parameter (QP) for the current video block. A first derivation comprises computing the first predictive QP based at least in part on another QP and an additional value when the partitioning scheme is the single tree partitioning. A second derivation comprises computing the first predictive QP by using a previously-derived QP when the partitioning scheme is the dual tree partitioning. The electronic device computes a second QP based at least in part on the first predictive QP and an offset value, and use the second QP to decode the current video block.

Abstract: A video coding device receives a bitstream including video data. The device determines an intra-prediction mode subset. The intra-prediction mode subset include intra-prediction modes that correlate to a plurality of reference lines for a current image block and excludes intra-prediction modes that correlate to a primary reference line for the current image block. When a first intra-prediction mode is included in the intra-prediction mode subset, the device decodes the first intra-prediction mode by an alternative intra-prediction mode index. When the first intra-prediction mode is not included in the intra-prediction mode subset, the device decodes the first intra-prediction mode by an intra-prediction mode index. The device presents video data including an image block decoded based on the first intra-prediction mode.

Abstract: A method of feature substitution for end-to-end image compression, is performed by at least one processor and includes encoding an input image, using a first neural network, to generate an encoded representation, and quantizing the generated encoded representation, using a second neural network, to generate a compressed representation. The first neural network and the second neural network are trained by determining a rate loss, based on a bitrate of the generated compressed representation, and updating the generated encoded representation, based on the determined rate loss.

Abstract: Methods and apparatus for video processing, including coding and decoding, are described. One example video processing method includes performing a conversion between a video having one or more video layers and a bitstream of the video according to a format rule, and wherein the format rule specifies that a value of a variable indicating whether to output a picture in an access unit is determined based on a flag indicating whether to output another picture in the access unit.

Abstract: A system can include a video processing engine to determine an estimated Quantization Parameter (QP) for a row of Coding Tree Units (CTUs) in a frame of a video. The processing engine can encode the row of CTUs in the frame of the video. A CTU in the row of CTUs can be encoded with a QP equal to the estimated QP. The system includes another video processing engine to determine an estimated QP for another row of CTUs of the frame of the video. The other processing engine can set a running QP to the estimated QP for the row of CTUs prior to the given video processor encoding a last CTU in the given row of CTUs. The other processing engine can further encode the other row of CTUs. CTUs encoded by the video processing engine and the other video processing engine can be stored in memory.

Abstract: Systems, methods and apparatus for video processing are described. The video processing may include video encoding, video decoding, or video transcoding. One example method of video processing includes performing a conversion between a video including one or more video regions and a bitstream of the video according to a format rule. The format rule specifies that a variable X indicates whether B slice is allowed or used in a video region. The format rule further specifies that the variable X is based on values of a reference picture list information present flag and/or a field indicating a number of entries in a reference picture list syntax structure.

Abstract: Training a model for detecting pedestrian objects is described. Annotated data can be received by a training component. In an example, the annotated data can include a first indication of a first object (e.g., a pedestrian) and a second indication of a second object (e.g., a pedestrian object), wherein the first object and the second object comprise a portion of a compound object (e.g., a pedestrian/pedestrian object system). In an example, a training component can train a model to determine an output associated with a second object based at least in part on the annotated data and the model can be transmitted to a vehicle configured to be controlled by output(s) of the model.

Abstract: An encoder includes memory and circuitry. The circuitry, using the memory, (i) selects a mode from among a plurality of modes each for deriving a motion vector, and derives a motion vector for a current block via the selected mode, and (ii) performs inter prediction encoding on the current block, using the derived motion vector, via one of a skip mode and a non-skip mode different from the skip mode. The plurality of modes include a plurality of first modes each for predicting the motion vector for the current block based on an encoded block neighboring the current block without encoding information indicating a motion vector into a stream. When a second mode included in the plurality of first modes is selected, the current block is encoded via the non-skip mode regardless of presence or absence of a residual coefficient.

Abstract: Provided are a video decoding method and apparatus including: in a video encoding and decoding process, determining parity information of a current block based on a width and a height of the current block; determining a lookup table of the current block from among a plurality of predefined lookup tables based on the parity information; determining a dequantization scale value of the current block based on the lookup table of the current block; and performing dequantization on the current block by using the dequantization scale value.

Abstract: A method for compressing video signals based on adaptive compression rate and a circuit system thereof are provided. In the method, a digital signal processor is used to process a video so as to frame-by-frame obtain statistical data, for example, a maximum of compressed data. The maximum of compressed data of a previous frame is used to determine a compression state of a current frame. The compression state of the frame allows the processor to decide a direction to adjust a compression ratio. Next, statistical data of the previous frame is used to decide a stride to adjust the compression ratio. The statistical data can be a maximum of compressed data and a quantization table scale referred to rendering a prediction curve that allows the processor to determine the stride. A compression ratio is then determined according to the direction and the stride of adjustment.

Abstract: A method for decoding a picture from a bitstream. In one embodiment, The method includes: receiving a slice header for a slice of the picture, wherein the slice header comprises a state syntax element; decoding a state value from the state syntax element in the slice header, wherein a) if the state value is not equal to a first value, then the state value indicates that i) the bitstream includes for the picture a picture header comprising a set of picture syntax elements and ii) the slice header does not comprise the set of picture syntax elements and b) if the state value is equal to the first value, then the state value indicates that i) the slice header contains said set of picture syntax elements and ii) the bitstream does not include a picture header for the picture; and using the set of picture syntax elements to decode the slice of the picture.

Abstract: When dependent scalar quantization is used, the choice of the quantizer depends on the decoding of the preceding transform coefficient, and the entropy decoding of a transform coefficient depends on quantizer choice. To maintain high throughput in hardware implementations for transform coefficient entropy coding, several decision schemes of the scaler quantizer are proposed. In one implementation, the state transition and the context model selection are based on only regular coded bins. For example, the state transition can be based on the sum of the SIG, gt1 and gt2 flags, the exclusive-or function of the SIG, gt1 and gt2 flags, or based on only the gt1 or gt2 flag. When a block of transform coefficients is coded, the regular mode bins can be coded first in one or more scan passes, and the remaining bypass coded bins are grouped together in another one or more scan passes.

Abstract: According to an aspect of the disclosure, a method of video decoding performed in a video decoder is provided. In the method, syntax information of a coding unit (CU) is decoded from a coded video bitstream. The syntax information is signaled in a picture header (PH) and includes chroma quantization parameter (QP) offsets in a PH level. The chroma QP offsets include at least one of a Cb offset, a Cr offset, and a CbCr offset. Further, quantization parameters for the CU are determined based on the chroma QP offsets in the PH level and a quantization parameter range offset of the CU.

Abstract: A method of video processing is provided. The method includes: determining, for a conversion between a current video unit of a video and a coded representation of the video, that applicability of a first coding tool and a second coding tool is mutually exclusive; and performing the conversion based on the determining, wherein the first coding tool corresponds to an adaptive color space transformation (ACT) tool; wherein use of the ACT tool comprises: converting, during encoding a representation of a visual signal from a first color domain to a second color domain, or converting during decoding, a representation of a visual signal from the second color domain to the first color domain.

Abstract: An information processing method and device, an apparatus, and a storage medium. The method includes: for inputted source video data, predicting a colour component of a coding block in the source video data according to a prediction mode to obtain a first prediction block, wherein the prediction mode is preset and is a Position-Dependent Prediction Combination (PDCP) mode; determining a difference between the colour component of the coding block and a prediction value of the first prediction block to obtain a residual block; and signalling the residual block and the prediction mode in a bitstream.

Abstract: The present principles relates to a method for intra-predictive encoding a coding unit comprising picture data, said intra-predictive encoding depending on a prediction tree and a transform tree, characterized in that the method further comprises: —obtaining said prediction tree by spatially partitioning the coding unit according to a non-square partition type; —determining said transform tree from said coding unit in order that each of its leaves is embedded into a unique unit of said obtained prediction tree; and —signaling in a signal the size of the leaves of said transform tree and said a non-square partition type.

Abstract: As system is provided for performing a method of receiving a superposition state defined by a sum of a plurality of addends, wherein each addend of the plurality of addends is a product between a corresponding coefficient of a plurality of coefficients and a corresponding state of a plurality of states encoded with block unary encoding. The system may identify at least one error state, of the plurality of states, having a string value that is not a block unary code string of a set of block unary code strings. The system may compute an updated superposition state based on the plurality of states without the error state.

Abstract: Provided are a video decoding method and apparatus including: in a video encoding and decoding process, determining parity information of a current block based on a width and a height of the current block; determining a lookup table of the current block from among a plurality of predefined lookup tables based on the parity information; determining a dequantization scale value of the current block based on the lookup table of the current block; and performing dequantization on the current block by using the dequantization scale value.

Abstract: A method for colour component prediction, applied to an encoder and includes: determining a neighbouring reference sample set of a current block, wherein the neighbouring reference sample set comprises at least one reference sample; determining a preset parameter value corresponding to the current block, including: determining a bitdepth value corresponding to a to-be-predicted colour component of the current block; and performing a shift on 1 by the bitdepth value minus 1 to obtain the preset parameter value. The method further includes: determining an input sample matrix by means of a first preset calculation model based on the neighbouring reference sample set and the preset parameter value; and performing colour component prediction on the current block according to the input sample matrix to obtain a prediction block of the current block.

Abstract: Provided are a video decoding method and apparatus including: in a video encoding and decoding process, determining parity information of a current block based on a width and a height of the current block; determining a lookup table of the current block from among a plurality of predefined lookup tables based on the parity information; determining a dequantization scale value of the current block based on the lookup table of the current block; and performing dequantization on the current block by using the dequantization scale value.

Abstract: A scene-based image processing method includes: performing decoding after acquiring video data to obtain multiple image frames, and identifying key frame images by performing key frame detection (S10); analyzing similarity between the content of two adjacent key frame images, and determining whether the difference in the content of the two adjacent key frame images is within a preset threshold (S20); and when the difference in the content of the two adjacent key frame images is within the preset threshold, determining that the two adjacent key frame images are a same scene, uniformly performing image feature statistics compilation, overlaying feature information, and performing image enhancement on a same scene by using a similar enhancement processing curve (S30).

Abstract: Provided are a video decoding method and apparatus including: in a video encoding and decoding process, determining parity information of a current block based on a width and a height of the current block; determining a lookup table of the current block from among a plurality of predefined lookup tables based on the parity information; determining a dequantization scale value of the current block based on the lookup table of the current block; and performing dequantization on the current block by using the dequantization scale value.

Abstract: A video coding device may be configured to perform intra prediction coding according to one or more of the techniques described herein.

Abstract: Aspects of the disclosure provide a method and an apparatus including processing circuitry for video decoding. The processing circuitry decodes prediction information of a current block in a current picture from a coded video bitstream. The prediction information indicates a prediction mode for the current block being an inter prediction mode or an intra block copy (IBC) mode. The processing circuitry selects a base vector from a candidate list including a block vector candidate used in the IBC mode and determines an offset vector based on offset information that includes directions and sizes for constructing offset vectors. The processing circuitry determines a vector of the current block based on the base vector and the offset vector. The vector is a block vector when the prediction mode is the IBC mode and the vector is a motion vector when the prediction mode is the inter prediction mode.

Abstract: An encoding method with multiple image block division manners is disclosed, including: determining a division manner and a division direction of an image block; dividing the image block to obtain image subblocks sequentially arranged horizontally or vertically; determining whether the image subblocks need subdivision, and if subdivision is not needed, predicting the encoding object in the frame according to the image subblocks, to obtain residual data; performing transformation, quantization, and entropy encoding for the residual data so as to obtain coded residual data; and writing the division manner of the image block, the division direction of the image block, an identifier indicating whether the image subblocks need subdivision, and the coded residual data into a bitstream. By applying the encoding method, better prediction accuracy can be achieved when the image block presents a small change of pixel value in the horizontal or vertical direction.

Abstract: A method for transforming high dynamic range (HDR) video data into standard dynamic range (SDR) video data and encoding the SDR video data so that the HDR video data may be recovered at the decoder includes generating a tone map describing a transformation applied to the HDR video data to generate the SDR video data. The generated tone map describes the transformation as the multiplication of each HDR pixel in the HDR video data by a scalar to generate the SDR video data. The tone map is then modeled as a reshaping transfer function and the HDR video data is processed by the reshaping transfer function to generate the SDR video data. The reshaping transfer function is then inverted and described in a self-referential metadata structure. The SDR video data is then encoded including the metadata structure defining the inverse reshaping transfer function.

Abstract: Systems, methods and apparatus for video processing are described. The video processing may include video encoding, video decoding, or video transcoding. One example method of video processing includes performing a conversion between a video including one or more video regions and a bitstream of the video according to a format rule. The format rule specifies that a variable X indicates whether B slice is allowed or used in a video region. The format rule further specifies that the variable X is based on values of a reference picture list information present flag and/or a field indicating a number of entries in a reference picture list syntax structure.

Abstract: A method and an apparatus for quantizing an activation volume of a deep neural network are disclosed. The method includes: obtaining an activation volume of a network layer in the deep neural network (S101), wherein elements in the activation volume are arranged in three directions: height, width, and depth; dividing depth segments in the activation volume in which a difference among element features is smaller than a preset threshold into a same slice group along the depth direction of the activation volume, so as to obtain a plurality of slice groups (S102); quantizing each slice group respectively by using a quantization parameter corresponding to each slice group obtained through a quantization formula (S103). The quantization error can be reduced through the above method.

Abstract: An image processing apparatus includes a memory configured to store compressed data; and frame buffer compressing circuitry which includes encoder circuitry configured to compress at least some of source data to generate the compressed data and transmit the compressed data to the memory, and decoder circuitry configured to read and decompress the compressed data from the memory, in which the compressed data includes a payload and a header including actual compressed data and flag, the frame buffer compressing circuitry is configured to reflect a result obtained by comparing an accumulated compressibility corresponding to the compressed data with a reference compressibility in the flag, and is configured to perform compression or decompression in a lossy mode or a lossless mode depending on the flag.

Abstract: Techniques are described for optimizing streaming video encoding profiles.

Abstract: A method for decoding image data can include generating a prediction block based on a prediction mode; generating a quantization block by inversely scanning quantization coefficient information; generating a transform block by inversely quantizing the quantization block using a quantization parameter; generating a residual block by inversely transforming the transform block; generating a reconstructed picture by using the prediction block and the residual block; and applying a deblocking filter on the reconstructed picture, wherein it is determined whether the deblocking filtering is applied between two adjacent blocks P and Q containing samples p0 and q0 respectively by using a boundary quantization parameter when a boundary strength is not zero, the boundary quantization parameter is set to an average value of a quantization parameter of block P and a quantization parameter of block Q, and the quantization parameter is derived by adding a residual quantization parameter and a quantization parameter predicto

Abstract: Composite components, such as gas turbine engine airfoils, and methods of forming composite components are provided. For example, a composite component of a gas turbine engine comprises a plurality of continuous fiber layers, each continuous fiber layer formed from a plurality of continuous fibers disposed in a matrix material, and a chopped fiber layer comprising a plurality of chopped fibers formed as a filmed sheet. The chopped fiber layer is laid up with the plurality of continuous fiber layers to form the composite component. A method for forming a composite component comprises laying up a plurality of continuous fiber layers, each continuous fiber layer comprising continuous fibers disposed in a matrix material; laying up a chopped fiber layer with the plurality of continuous fiber layers, the layup of the plurality of continuous fiber layers and the chopped fiber layer forming a reinforced layup; and curing the reinforced layup.