Video encoding methods and apparatuses for performing rate-distortion optimization by a hierarchical architecture include receiving input data associated with a current block in a current picture, determining a block partitioning structure to split the current block into coding blocks and determining a corresponding coding mode for each coding block by multiple Processing Element (PE) groups, and entropy encoding the coding blocks in the current block according to the coding modes determined by the PE groups. Each PE group has parallel PEs and is associated with a particular block size. The parallel PEs in each PE group test a number of coding modes on each partition or sub-partition of the current block to derive rate-distortion costs. The block partitioning structure and corresponding coding modes are then decided based on the rate-distortion costs derived by the PE groups.

A data processing method for non-sequential point cloud media includes: acquiring property information of a viewing region corresponding to non-sequential point cloud media; and presenting the non-sequential point cloud media based on the property information of the viewing region corresponding to the non-sequential point cloud media. By introducing first indication information into the property information of the viewing region corresponding to the non-sequential point cloud media, when indicating that a recommended viewing region exists for the non-sequential point cloud media, the presentation of the corresponding non-sequential point cloud media according to property information of the recommended viewing region in the property information of the viewing region can be supported on the basis of an encapsulation structure of the non-sequential point cloud media.

A data processing method for non-sequential point cloud media includes: acquiring property information of a viewing region corresponding to non-sequential point cloud media; and presenting the non-sequential point cloud media based on the property information of the viewing region corresponding to the non-sequential point cloud media. By introducing first indication information into the property information of the viewing region corresponding to the non-sequential point cloud media, when indicating that a recommended viewing region exists for the non-sequential point cloud media, the presentation of the corresponding non-sequential point cloud media according to property information of the recommended viewing region in the property information of the viewing region can be supported on the basis of an encapsulation structure of the non-sequential point cloud media.

An image compression method includes: obtaining a target image and a target code rate corresponding to the target image; determining a first code rate parameter corresponding to the target code rate; and inputting the target image and the first code rate parameter into an image compression model, that has been trained, for processing to obtain a compressed image with the target code rate, wherein the image compression model is obtained by training an initial image compression model based on multiple code rate parameters including the first code rate parameter.

An embodiment of the present specification provides a method and device for processing video data. A method for processing a video signal according to an embodiment of the present specification may comprise the steps of: acquiring a transform index related to one of a plurality of transform combinations including combinations of one or more transform kernels for transforming of a current block of the video signal; deriving a transform combination including a vertical transform and a horizontal transform related to the transform index; and applying each of the vertical transform and horizontal transform of the transform combination to the current block.

Embodiments described herein provided for an instruction and associated logic to enable a processing resource including a tensor accelerator to perform optimized computation of sparse submatrix operations. One embodiment provides hardware logic to apply a numerical transform to matrix data to increase the sparsity of the data. Increasing the sparsity may result in a higher compression ratio when the matrix data is compressed.

Example embodiments provide a system for training a data coding pipeline including a feature extractor neural network, an encoder neural network, and a decoder neural network configured to reconstruct input data based on encoded features. A plurality of losses corresponding to different tasks may be determined for the coding pipeline. Tasks may be performed based on an output of the coding pipeline. A weight update may be determined for at least a subset of the coding pipeline based on the plurality of losses. The weight update may be configured to reduce a number of iterations for fine-tuning the coding pipeline for one of the tasks. This enables faster adaptation of the coding pipeline for one of the tasks after deployment of the coding pipeline. Apparatuses, methods, and computer programs are disclosed. Apparatuses, methods, and computer programs are disclosed.

Example embodiments provide a system for training a data coding pipeline including a feature extractor neural network, an encoder neural network, and a decoder neural network configured to reconstruct input data based on encoded features. A plurality of losses corresponding to different tasks may be determined for the coding pipeline. Tasks may be performed based on an output of the coding pipeline. A weight update may be determined for at least a subset of the coding pipeline based on the plurality of losses. The weight update may be configured to reduce a number of iterations for fine-tuning the coding pipeline for one of the tasks. This enables faster adaptation of the coding pipeline for one of the tasks after deployment of the coding pipeline. Apparatuses, methods, and computer programs are disclosed. Apparatuses, methods, and computer programs are disclosed.

A camera module includes a compressor configured to divide a plurality of pixels included in image data, into a plurality of pixel groups, with respect to each of the plurality of pixel groups into which the plurality of pixels is divided, calculate a representative pixel value of a corresponding pixel group, based on pixel values of multiple pixels included in the corresponding pixel group, generate first compressed data, based on the calculated representative pixel value of each of the plurality of pixel groups, with respect to each of the plurality of pixel groups into which the plurality of pixels is divided, calculate residual values representing differences between the pixel values of the multiple pixels included in the corresponding pixel group and the representative pixel value of the corresponding pixel group, and generate second compressed data, based on the calculated residual values of each of the plurality of pixel groups.

A camera module includes a compressor configured to divide a plurality of pixels included in image data, into a plurality of pixel groups, with respect to each of the plurality of pixel groups into which the plurality of pixels is divided, calculate a representative pixel value of a corresponding pixel group, based on pixel values of multiple pixels included in the corresponding pixel group, generate first compressed data, based on the calculated representative pixel value of each of the plurality of pixel groups, with respect to each of the plurality of pixel groups into which the plurality of pixels is divided, calculate residual values representing differences between the pixel values of the multiple pixels included in the corresponding pixel group and the representative pixel value of the corresponding pixel group, and generate second compressed data, based on the calculated residual values of each of the plurality of pixel groups.