Low complexity NN-based in-loop filter architecture with separable convolution

Abstract

Example techniques for filtering video data are described. An example apparatus for at least one of encoding or decoding video data includes: one or more memories configured to store video data; and one or more processors. The one or more processors are configured to receive a picture of the video data and reconstruct the picture of the video data. The one or more processors are further configured to apply a neural network (NN)-based filter to the reconstructed picture of the video data. The NN-based filters include a unified filter. The unified filter includes a head block, a transition block, one or more backbone blocks, and a tail block. At least one of the head block, the transition block, the one or more backbone blocks, or the tail block includes positive multivariate (CP) decomposition with separable convolution.

Description

Cross-references to related applications

[0001] This application claims priority to U.S. Patent Application No. 18 / 738,842, filed June 10, 2024, and U.S. Provisional Patent Application No. 63 / 507,651, filed June 12, 2023, the entire contents of each of which are incorporated herein by reference. U.S. Patent Application No. 18 / 738,842, filed June 10, 2024, claims the benefit of U.S. Provisional Patent Application No. 63 / 507,651, filed June 12, 2023. Technical Field

[0002] This disclosure relates to video encoding and video decoding. Background Technology

[0003] Digital video capabilities can be incorporated into a wide variety of devices, including digital televisions, digital live broadcast systems, wireless broadcasting systems, personal digital assistants (PDAs), laptops or desktop computers, tablets, e-book readers, digital cameras, digital recording devices, digital media players, video game devices, video game consoles, cellular or satellite wireless phones (so-called "smartphones"), video conferencing equipment, video streaming devices, and more. Digital video devices implement video decoding technologies, such as those defined by MPEG-2, MPEG-4, ITU-T H.263, ITU-T H.264 / MPEG-4 (Part 10, Advanced Video Decoding (AVC)), ITU-T H.265 / High Efficiency Video Decoding (HEVC), ITU-T H.266 / Variety Video Decoding (VVC) and extensions to these standards, as well as proprietary video codecs / formats such as AOMedia Video1 (AV1) developed by the Open Media Alliance. By implementing such video decoding technologies, video devices can more efficiently send, receive, encode, decode, and / or store digital video information.

[0004] Video decoding techniques include spatial (intra-picture) prediction and / or temporal (inter-picture) prediction to reduce or remove redundancy inherent in video sequences. For block-based video decoding, video slices (e.g., video pictures or portions of video pictures) can be divided into video blocks, which may also be referred to as decoding tree units (CTUs), decoding units (CUs), and / or decoding nodes. Video blocks in a slice after intra-frame decoding (I) of a picture are encoded using spatial predictions about reference samples in adjacent blocks within the same picture. Video blocks in a slice after inter-frame decoding (P or B) of a picture can use spatial predictions about reference samples in adjacent blocks within the same picture or temporal predictions about reference samples in other reference pictures. A picture may be called a frame, and a reference picture may be called a reference frame. Summary of the Invention

[0005] Generally, this disclosure describes techniques for neural network (NN)-based in-loop filtering for video decoding. More specifically, this disclosure describes simplified examples of NN-based in-loop filters. A unified architecture (UA) for NN-based in-loop filters has been developed. However, UAs are power and processing intensive and are therefore unsuitable for use on power- and / or processing-constrained devices, such as battery-powered mobile devices. The techniques disclosed herein provide NN-based in-loop filters operable on such constrained devices.

[0006] In one example, a method includes: receiving an image of video data; reconstructing the image of the video data; and applying a neural network (NN)-based filter to the reconstructed image of the video data, wherein the NN-based filter includes a unified filter comprising a head block, a transition block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, the transition block, one or more backbone blocks, or the tail block comprises a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0007] In another example, a device includes: one or more memories configured to store video data; and one or more processors configured to: receive images of the video data; reconstruct images of the video data; and apply neural network (NN)-based filters to the reconstructed images of the video data, wherein the NN-based filters include a unified filter comprising a head block, a transform block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, transform block, one or more backbone blocks, or tail block comprises a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0008] In another example, an apparatus includes: a component for receiving an image of video data; a component for reconstructing the image of the video data; and a component for applying a neural network (NN)-based filter to the reconstructed image of the video data, wherein the NN-based filter includes a unified filter comprising a head block, a transition block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, the transition block, one or more backbone blocks, or the tail block comprises a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0009] In another example, a computer-readable storage medium is encoded with instructions that, when executed, cause one or more programmable processors to: receive a picture of video data; reconstruct the picture of the video data; and apply a neural network (NN)-based filter to the reconstructed picture of the video data, wherein the NN-based filter includes a unified filter comprising a head block, a transition block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, transition block, one or more backbone blocks, or tail block comprises a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0010] Details of one or more examples are set forth in the accompanying drawings and the following description. Other features, objects, and advantages will be apparent from the description, drawings, and claims. Attached Figure Description

[0011] Figure 1 This is a block diagram illustrating an example video encoding and decoding system that can perform the techniques of this disclosure.

[0012] Figure 2 This is a block diagram illustrating an example of a hybrid video decoding framework.

[0013] Figure 3 An example hierarchical prediction structure with a group of pictures (GOP) size of 16 is shown.

[0014] Figure 4 This is a block diagram illustrating an example of a Convolutional Neural Network (CNN) based filter with four layers.

[0015] Figure 5 This is a block diagram illustrating an example CNN-based filter with padded input samples and supplementary data.

[0016] Figure 6 This is a block diagram illustrating another example of a CNN-based filter with padded input samples and supplementary data.

[0017] Figure 7 This is an example Figure 6 A conceptual diagram of attention residual blocks.

[0018] Figure 8 This is a block diagram illustrating another example of a CNN-based filter with padded input samples and supplementary data.

[0019] Figure 9 This is a block diagram illustrating an example of a simplified CNN-based filter architecture with padded input samples and supplementary data.

[0020] Figure 10 This is an example Figure 9 A conceptual diagram of an example residual block structure.

[0021] Figure 11 This is a block diagram illustrating another example of a simplified CNN-based filter architecture with padded input samples and supplementary data.

[0022] Figure 12 This is an example Figure 11 A conceptual diagram of an example filter block structure.

[0023] Figure 13 This is a block diagram illustrating an example CNN decomposition architecture.

[0024] Figure 14 This is a block diagram illustrating an example multi-scale feature extraction backbone network with two-dimensional convolution.

[0025] Figure 15 This is a block diagram illustrating an example unified filter with a joint model (joint luminance and chrominance).

[0026] Figure 16 This is a block diagram illustrating the example backbone block.

[0027] Figure 17 This is a block diagram (luminance) illustrating an example unified filter with separate luminance and chrominance models.

[0028] Figure 18 This is a block diagram (chroma) illustrating an example unified filter with separate luminance and chrominance models.

[0029] Figure 19 This is a block diagram illustrating an example low-complexity in-loop filter network architecture according to one or more aspects of this disclosure.

[0030] Figure 20 This is a block diagram illustrating an example backbone block according to one or more aspects of this disclosure.

[0031] Figure 21 This is a block diagram illustrating an example CP decomposition according to one or more aspects of this disclosure.

[0032] Figure 22 This is a flowchart illustrating an example of an NN-based in-loop filtering technique according to one or more aspects of this disclosure.

[0033] Figure 23 This is a block diagram illustrating an example video encoder that can perform the techniques of this disclosure.

[0034] Figure 24 This is a block diagram illustrating an example video decoder that can perform the techniques of this disclosure.

[0035] Figure 25 This is a flowchart illustrating an example method for encoding the current block according to the technology of this disclosure.

[0036] Figure 26 This is a flowchart illustrating an example method for decoding the current block according to the technology of this disclosure. Detailed Implementation

[0037] Video decoding (e.g., video encoding and / or video decoding) typically involves predicting video data blocks based on already decoded video data blocks within the same frame (e.g., intra-frame prediction) or based on already decoded video data blocks in different frames (e.g., inter-frame prediction). In some cases, the video encoder also computes residual data by comparing the predicted block to the original block. Therefore, the residual data represents the difference between the predicted block and the original block. To reduce the number of bits required to signal the residual data, the video encoder transforms and quantizes the residual data and signals the transformed and quantized residual data in the encoded bitstream. Compression achieved through the transformation and quantization process can be lossy, meaning that the transformation and quantization process may introduce distortion into the decoded video data.

[0038] The video decoder decodes the residual data and adds it to the predicted blocks to produce reconstructed video blocks that more closely match the original video blocks compared to individual predicted blocks. Due to losses introduced by the transformation and quantization of the residual data, the reconstructed blocks may exhibit distortion or artifacts. A common type of artifact or distortion is called block artifacts, where the boundaries of the blocks used to decode the video data are visible.

[0039] To further improve the quality of the decoded video, the video decoder can perform one or more filtering operations on the reconstructed video blocks. Examples of these filtering operations include deblocking filtering, Sample Adaptive Offset (SAO) filtering, and Adaptive Loop Filtering (ALF). The parameters of these filtering operations can be determined by the video encoder and explicitly signaled in the encoded video bitstream, or they can be implicitly determined by the video decoder without needing to be explicitly signaled in the encoded video bitstream.

[0040] This disclosure describes an in-loop filtering (ILF) architecture based on a neural network (NN) for video decoding purposes. The techniques disclosed herein can improve decoding performance under constraints of complexity and memory requirements. In some examples, the architecture of this disclosure can be based on a convolutional neural network (CNN). In some CNN ILFs, complexity reduction can be achieved by replacing conventional multidimensional convolutions with decomposition (e.g., utilizing separable convolutions).

[0041] Generally, this disclosure describes techniques for neural network-based in-loop filtering for video decoding. Video decoders (e.g., encoders and / or decoders) may have complexity and memory requirement constraints, such as when such video decoders are implemented in mobile devices. The described techniques can improve decoding performance under complexity and memory requirement constraints, resulting in higher quality video playback than existing techniques under such constraints. The reduction in architectural complexity can be achieved by utilizing positive quasi-multivariate (CP) decomposition using separable convolutions instead of conventional convolutions using multidimensional kernels, removing or simplifying activation layers, and / or removing or shifting multi-scale branches. The techniques described in this disclosure are related to neural network-based in-loop filtering; however, they are applicable to any neural network-based video decoding tool that consumes input data with certain statistical properties. The techniques can be used in the context of advanced video codecs, such as extensions to VVC, or next-generation video decoding standards, and / or any other video codec.

[0042] Figure 1 This is a block diagram illustrating an example video encoding and decoding system 100 capable of performing the techniques of this disclosure. The techniques of this disclosure generally involve decoding (encoding and / or decoding) video data. Typically, video data includes any data used for processing video. Thus, video data may include unencoded raw video, encoded video, decoded (e.g., reconstructed) video, and video metadata, such as signaling data.

[0043] like Figure 1 As shown, in this example, system 100 includes a source device 102 that provides encoded video data to be decoded and displayed by a destination device 116. Specifically, source device 102 provides the video data to destination device 116 via computer-readable medium 110. Source device 102 and destination device 116 can be or may include any of a wide range of devices, such as desktop computers, laptop computers, mobile devices, tablet computers, set-top boxes, mobile phones such as smartphones, televisions, cameras, display devices, digital media players, video game consoles, video streaming devices, broadcast receiver devices, etc. In some cases, source device 102 and destination device 116 may be configured for wireless communication and are therefore referred to as wireless communication devices.

[0044] exist Figure 1In the example, source device 102 includes a video source 104, memory 106, video encoder 200, and output interface 108. Destination device 116 includes an input interface 122, video decoder 300, memory 120, and display device 118. According to this disclosure, the video encoder 200 of source device 102 and the video decoder 300 of destination device 116 can be configured to apply techniques for in-loop filtering based on neural networks (NNs). Therefore, source device 102 represents an example of a video encoding device, while destination device 116 represents an example of a video decoding device. In other examples, the source device and destination device may include other components or arrangements. For example, source device 102 may receive video data from an external video source such as an external camera. Similarly, destination device 116 may interface with an external display device instead of including an integrated display device.

[0045] like Figure 1 The system 100 shown is merely an example. Typically, any digital video encoding and / or decoding device can perform techniques for in-loop filtering based on neural networks (NNs). Source device 102 and destination device 116 are merely examples of such decoding devices, where source device 102 generates decoded video data for transmission to destination device 116. This disclosure refers to a “decoding” device as a device that performs the decoding (e.g., encoding and / or decoding) of data. Thus, video encoder 200 and video decoder 300 represent examples of decoding devices, specifically, a video encoder and a video decoder, respectively. In some examples, source device 102 and destination device 116 may operate in a substantially symmetrical manner, such that each of source device 102 and destination device 116 includes video encoding and decoding components. Therefore, system 100 may support one-way or two-way video transmission between source device 102 and destination device 116, for example, for video streaming, video playback, video broadcasting, or video telephony.

[0046] Typically, video source 104 represents the source of video data (i.e., unencoded raw video data) and provides a sequential series of pictures (also referred to as "frames") of the video data to video encoder 200, which encodes the data for the pictures. Video source 104 of source device 102 may include video capture devices such as cameras, video archives containing previously captured raw video, and / or video feed interfaces for receiving video from video content providers. Alternatively, video source 104 may generate computer graphics-based data as source video, or a combination of live video, archived video, and computer-generated video. In each case, video encoder 200 encodes the captured, pre-captured, or computer-generated video data. Video encoder 200 may rearrange the pictures from the received order (sometimes referred to as "display order") to a decoding order for decoding. Video encoder 200 may generate a bitstream comprising the encoded video data. Then, the source device 102 can output the encoded video data to the computer-readable medium 110 via the output interface 108 for reception and / or retrieval by, for example, the input interface 122 of the destination device 116.

[0047] The memory 106 of source device 102 and the memory 120 of destination device 116 represent general-purpose memory. In some examples, memories 106 and 120 may store raw video data, such as raw video from video source 104 and raw decoded video data from video decoder 300. Additionally or alternatively, memories 106 and 120 may store software instructions executable by, for example, video encoder 200 and video decoder 300. Although memories 106 and 120 are shown separately from video encoder 200 and video decoder 300 in this example, it should be understood that video encoder 200 and video decoder 300 may also include internal memory for functionally similar or equivalent purposes. Furthermore, memories 106 and 120 may store encoded video data, such as output from video encoder 200 and input to video decoder 300. In some examples, portions of memories 106 and 120 may be allocated as one or more video buffers, for example, to store raw decoded and / or encoded video data.

[0048] Computer-readable medium 110 may represent any type of medium or device capable of transmitting encoded video data from source device 102 to destination device 116. In one example, computer-readable medium 110 represents a communication medium that enables source device 102 to directly transmit encoded video data to destination device 116 in real time, for example, via a radio frequency network or a computer-based network. According to a communication standard such as a wireless communication protocol, output interface 108 may modulate the transmitted signal including the encoded video data, and input interface 122 may demodulate the received transmitted signal. The communication medium may include any wireless or wired communication medium, such as radio frequency (RF) spectrum or one or more physical transmission lines. The communication medium may form part of a packet-based network such as a local area network, a wide area network, or a global network (such as the Internet). The communication medium may include a router, switch, base station, or any other equipment that may be useful for facilitating communication from source device 102 to destination device 116.

[0049] In some examples, source device 102 can output encoded data from output interface 108 to storage device 112. Similarly, destination device 116 can access encoded data from storage device 112 via input interface 122. Storage device 112 may include any data storage medium of various distributed or locally accessed data storage media, such as hard disk drives, Blu-ray discs, DVDs, CD-ROMs, flash memory, volatile or non-volatile memory, or any other suitable digital storage medium for storing encoded video data.

[0050] In some examples, source device 102 can output encoded video data to file server 114 or another intermediate storage device that can store the encoded video data generated by source device 102. Destination device 116 can access the stored video data from file server 114 via streaming or download.

[0051] File server 114 can be any type of server device capable of storing encoded video data and sending the encoded video data to destination device 116. File server 114 may represent a web server (e.g., for a website), a server configured to provide file transfer protocol services (such as File Transfer Protocol (FTP) or FLUTE-based file delivery protocol), a Content Delivery Network (CDN) device, a Hypertext Transfer Protocol (HTTP) server, a Multimedia Broadcast Multicast Service (MBMS) or Enhanced MBMS (eMBMS) server, and / or a Network Attached Storage (NAS) device. File server 114 may additionally or alternatively implement one or more HTTP streaming protocols, such as HTTP-based Dynamic Adaptive Streaming (DASH), HTTP Live Streaming (HLS), Real-Time Streaming Protocol (RTSP), HTTP Dynamic Streaming, etc.

[0052] Destination device 116 can access encoded video data from file server 114 via any standard data connection, including an internet connection. This may include a wireless channel (e.g., Wi-Fi connection), a wired connection (e.g., digital subscriber line (DSL), cable modem, etc.), or a combination of both, suitable for accessing encoded video data stored on file server 114. Input interface 122 can be configured to operate according to any or more of the various protocols discussed above for retrieving or receiving media data from file server 114, or other such protocols for retrieving media data.

[0053] Output interface 108 and input interface 122 can represent a wireless transmitter / receiver, a modem, a wired networking component (e.g., an Ethernet card), a wireless communication component operating according to any of the various IEEE 802.11 standards, or other physical components. In examples where output interface 108 and input interface 122 include wireless components, output interface 108 and input interface 122 can be configured to transmit data such as encoded video data according to cellular communication standards such as 4G, 4G-LTE (Long Term Evolution), Advanced LTE, 5G, etc. In some examples where output interface 108 includes a wireless transmitter, output interface 108 and input interface 122 can be configured according to other wireless standards such as the IEEE 802.11 specification, the IEEE 802.15 specification (e.g., ZigBee), etc. ™ ),Bluetooth ™Standards are used to transmit data such as encoded video data. In some examples, source device 102 and / or destination device 116 may include corresponding system-on-chip (SoC) devices. For example, source device 102 may include an SoC device for performing functions belonging to video encoder 200 and / or output interface 108, and destination device 116 may include an SoC device for performing functions belonging to video decoder 300 and / or input interface 122.

[0054] The technology disclosed herein can be applied to video decoding to support any multimedia application in a variety of multimedia applications, such as over-the-air television broadcasting, cable television transmission, satellite television transmission, internet streaming video transmission (such as HTTP-based Dynamic Adaptive Streaming (DASH)), digital video encoded onto data storage media, decoding of digital video stored on data storage media, or other applications.

[0055] The input interface 122 of the destination device 116 receives an encoded video bitstream from a computer-readable medium 110 (e.g., a communication medium, storage device 112, file server 114, etc.). The encoded video bitstream may include signaling information defined by the video encoder 200 and also used by the video decoder 300, such as syntax elements having values ​​describing the characteristics and / or processing of video blocks or other decoded units (e.g., slices, pictures, picture groups, sequences, etc.). The display device 118 displays a decoded picture of the decoded video data to the user. The display device 118 may represent any of a variety of display devices, such as a liquid crystal display (LCD), a plasma display, an organic light-emitting diode (OLED) display, or another type of display device.

[0056] Despite Figure 1Not shown, but in some examples, both the video encoder 200 and the video decoder 300 may be integrated with the audio encoder and / or audio decoder (e.g., audio codec), and may include appropriate MUX-DEMUX units or other hardware and / or software to process multiplexed streams that include both audio and video in a common data stream. Example audio codecs may include AAC, AC-3, AC-4, ALAC, ALS, AMBE, AMR, AMR-WB (G.722.2), AMR-WB+, aptX (various versions), ATRAC, BroadVoice (BV16, BV32), CELT, Enhanced AC-3 (E-AC-3), EVS, FLAC, G.711, G.722, G.722.1, G.722.2 (AMR-WB), G.723.1, G.726, G.728, G.729, G.729.1, GSM-FR, HE-AAC, iLBC, iSAC, LA Lyra, Monkey's Audio, MP1, MP2 (MPEG-1, 2 Audio Layer II), MP3, Musepack, Nellymoser Asao, OptimFROG, Opus, Sac, Satin, SBC, SILK, Siren 7, Speex, SVOPC, True Audio (TTA), TwinVQ, USAC, Vorbis (Ogg), WavPack and Windows Media Aud.

[0057] Both the video encoder 200 and the video decoder 300 can be implemented as any of a variety of suitable encoder and / or decoder circuits, such as one or more microprocessors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), discrete logic, software, hardware, firmware, or any combination thereof. When the technology is partially implemented in software, the device may store instructions for the software in a suitable non-transitory computer-readable medium and execute the instructions in hardware using one or more processors to perform the technology of this disclosure. Each of the video encoder 200 and the video decoder 300 may be included in one or more encoders or decoders, either of which may be integrated as part of a combined encoder / decoder (CODEC) in the respective device. Devices including the video encoder 200 and / or the video decoder 300 may implement the video encoder 200 and / or the video decoder 300 in processing circuitry such as integrated circuits and / or microprocessors. Such devices may be wireless communication devices (such as cellular phones) or any other type of device described herein.

[0058] The video encoder 200 and video decoder 300 may operate according to a video decoding standard such as ITU-T H.265 (also known as High Efficiency Video Decoding (HEVC)) or its extensions (such as Multi-View and / or Scalable Video Decoding Extensions)). Alternatively, the video encoder 200 and video decoder 300 may operate according to other proprietary or industry standards such as ITU-T H.266 (also known as Multi-Functional Video Decoding (VVC)). In other examples, the video encoder 200 and video decoder 300 may operate according to proprietary video codecs / formats such as AOMedia Video 1 (AV1), extensions to AV1, and / or subsequent versions of AV1 (e.g., AV2)). In other examples, the video encoder 200 and video decoder 300 may operate according to other proprietary formats or industry standards. However, the technology disclosed herein is not limited to any particular decoding standard or format. Typically, the video encoder 200 and the video decoder 300 can be configured to perform the techniques of this disclosure in conjunction with any video decoding technique that uses neural networks, such as NN-based video decoding.

[0059] Typically, video encoder 200 and video decoder 300 perform block-based decoding of images. The term "block" generally refers to a structure that includes data to be processed (e.g., encoded, decoded, or otherwise used during encoding and / or decoding). For example, a block may include a two-dimensional matrix of samples of luminance and / or chrominance data. Typically, video encoder 200 and video decoder 300 decode video data represented in YUV (e.g., Y, Cb, Cr) format. That is, instead of decoding the red, green, and blue (RGB) data used for images, video encoder 200 and video decoder 300 decode the luminance and chrominance components, where the chrominance components may include both red hue chrominance components and blue hue chrominance components. In some examples, video encoder 200 converts the received RGB format data to a YUV representation before encoding, and video decoder 300 converts the YUV representation to RGB format. Alternatively, preprocessing and postprocessing units (not shown) may perform these conversions.

[0060] This disclosure generally relates to the decoding (e.g., encoding and decoding) of images to include processes of encoding or decoding data of the image. Similarly, this disclosure may relate to the decoding of blocks of images to include processes of encoding or decoding data for the blocks (e.g., prediction and / or residual decoding). Encoded video bitstreams typically include a series of values ​​for syntax elements representing decoding decisions (e.g., decoding modes) and the partitioning of images into blocks. Therefore, references to the decoding of images or blocks should generally be understood as the decoded values ​​of the syntax elements that form the images or blocks.

[0061] HEVC defines various blocks, including decoding units (CUs), prediction units (PUs), and transform units (TUs). According to HEVC, a video decoder (such as a video encoder 200) divides the decoding tree unit (CTU) into CUs based on a quadtree structure. That is, the video decoder divides the CTU and CU into four equal, non-overlapping squares, and each node of the quadtree has zero or four child nodes. Nodes without child nodes are called "leaf nodes," and the CU of such leaf nodes may include one or more PUs and / or one or more TUs. The video decoder may further divide the PUs and TUs. For example, in HEVC, a residual quadtree (RQT) represents the partitioning of the TU. In HEVC, the PU represents inter-frame prediction data, while the TU represents residual data. The CU after intra-frame prediction includes intra-frame prediction information, such as intra-frame mode indication.

[0062] As another example, video encoder 200 and video decoder 300 can be configured to operate according to VVC. According to VVC, the video decoder (such as video encoder 200) partitions the image into multiple CTUs. Video encoder 200 can partition the CTUs according to a tree structure (such as a quadtree-binary tree (QTBT) structure or a multi-type tree (MTT) structure). The QTBT structure removes the concept of multiple partition types, such as the separation between CUs, PUs, and TUs in HEVC. The QTBT structure includes two levels: a first level partitioned according to the quadtree partitioning, and a second level partitioned according to the binary tree partitioning. The root node of the QTBT structure corresponds to the CTU. The leaf nodes of the binary tree correspond to the CUs.

[0063] In the MTT partitioning structure, blocks can be divided using quadtree (QT) partitioning, binary tree (BT) partitioning, and one or more types of ternary tree (TT) partitioning (also known as triplet tree (TT)). A ternary tree or triplet tree partition is a partition in which a block is divided into three sub-blocks. In some examples, a ternary tree or triplet tree partition divides a block into three sub-blocks without dividing the original block through the center. Partition types in MTT (e.g., QT, BT, and TT) can be symmetric or asymmetric.

[0064] When operating according to the AV1 codec, the video encoder 200 and video decoder 300 can be configured to decode video data in blocks. In AV1, the largest decoded block that can be processed is called a superblock. In AV1, a superblock can be 128x128 luma samples or 64x64 luma samples. However, in subsequent video decoding formats (e.g., AV2), superblocks can be defined by different (e.g., larger) luma sample sizes. In some examples, the superblock is the top level of a block quadtree. The video encoder 200 can further divide the superblock into smaller decoded blocks. The video encoder 200 can use square or non-square partitions to divide the superblock and other decoded blocks into smaller blocks. Non-square blocks can include N / 2xN blocks, NxN / 2 blocks, N / 4xN blocks, and NxN / 4 blocks. The video encoder 200 and video decoder 300 can perform separate prediction and transform processing for each decoded block.

[0065] AV1 also defines video data tiles. A tile is a rectangular array of superblocks that can be decoded independently of other tiles. That is, the video encoder 200 and video decoder 300 can encode and decode the decoding blocks within a tile separately without using video data from other tiles. However, the video encoder 200 and video decoder 300 can perform filtering across tile boundaries. The tile size can be uniform or non-uniform. Tile-based decoding enables parallel processing and / or multithreading in the encoder and decoder implementations.

[0066] In some examples, the video encoder 200 and the video decoder 300 may use a single QTBT or MTT structure to represent each of the luma and chroma components, while in other examples, the video encoder 200 and the video decoder 300 may use two or more QTBT or MTT structures, such as one QTBT / MTT structure for the luma component and another QTBT / MTT structure for the two chroma components (or two QTBT / MTT structures for the respective chroma components).

[0067] The video encoder 200 and the video decoder 300 can be configured to use quadtree partitioning, QTBT partitioning, MTT partitioning, superblock partitioning or other partitioning structures.

[0068] In some examples, a CTU includes a decoded tree block (CTB) of luminance samples, two corresponding CTBs of chrominance samples of an image with three sample arrays, or a CTB of samples of an image decoded using three separate color planes and a syntax structure for decoding the samples. A CTB can be an NxN sample block of some value N, such that a partitioning method divides the components into CTBs. A component is an array or a single sample from one of the three arrays (luminance and two chrominance) constituting a 4:2:0, 4:2:2, or 4:4:4 color format image, or an array or a single sample constituting an array or array constituting a monochrome format image. In some examples, a decoded block is an MxN sample block of values ​​M and N, such that a partitioning method divides the CTB into decoded blocks.

[0069] Blocks (e.g., CTUs or CUs) can be grouped in various ways within an image. As an example, a brick can refer to a rectangular area of ​​a row of CTUs within a specific tile in an image. A tile can be a rectangular area of ​​CTUs within a specific tile column and a specific tile row in an image. A tile column refers to a rectangular area of ​​a CTU having a height equal to the height of the image and a width specified by syntax elements (e.g., such as in an image parameter set). A tile row refers to a rectangular area of ​​a CTU having a height specified by syntax elements (e.g., such as in an image parameter set) and a width equal to the width of the image.

[0070] In some examples, a tile can be divided into multiple bricks, each brick comprising one or more CTU rows within the tile. A tile that is not divided into multiple bricks can also be called a brick. However, bricks that are a true subset of a tile cannot be called a tile. Bricks in an image can also be arranged in slices. A slice can be an integer number of bricks in an image that can be uniquely contained within a single Network Abstraction Layer (NAL) unit. In some examples, a slice comprises multiple complete tiles or a consecutive sequence of complete bricks comprising only one tile.

[0071] This disclosure uses "NxN" and "N by N" interchangeably to refer to the sample size of a block (such as a CU or other video block) in the vertical and horizontal dimensions, for example, 16x16 samples or 16 by 16 samples. Typically, a 16x16 CU will have 16 samples in the vertical direction (y=16) and 16 samples in the horizontal direction (x=16). Similarly, an NxN CU typically has N samples in the vertical direction and N samples in the horizontal direction, where N represents a non-negative integer value. Samples in a CU may be arranged in rows and columns. Furthermore, a CU does not necessarily need to have the same number of samples in the horizontal direction as it does in the vertical direction. For example, a CU may include NxM samples, where M is not necessarily equal to N.

[0072] The video encoder 200 encodes video data representing prediction and / or residual information, as well as other information, for use in the control unit (CU). The prediction information indicates how the CU should be predicted to form a prediction block for the CU. The residual information typically represents the sample-by-sample difference between a sample of the CU before encoding and the prediction block.

[0073] To predict the Cubic Frame (CU), the video encoder 200 typically forms prediction blocks for the CU through inter-frame prediction or intra-frame prediction. Inter-frame prediction typically refers to predicting the CU from data in a previously decoded image, while intra-frame prediction typically refers to predicting the CU from data in a previously decoded image within the same frame. To perform inter-frame prediction, the video encoder 200 can use one or more motion vectors to generate prediction blocks. The video encoder 200 can typically perform a motion search to identify reference blocks that closely match the CU, for example, based on the differences between the CU and a reference block. The video encoder 200 can use sum of absolute differences (SAD), sum of squared differences (SSD), mean absolute difference (MAD), mean squared difference (MSD), or other such difference calculations to compute difference metrics to determine whether a reference block closely matches the current CU. In some examples, the video encoder 200 can use unidirectional or bidirectional prediction to predict the current CU.

[0074] Some examples of VVC also provide an affine motion compensation mode, which can be viewed as an inter-frame prediction mode. In affine motion compensation mode, the video encoder 200 can determine two or more motion vectors representing non-translational motion, such as zooming in or out, rotation, perspective motion, or other irregular motion types.

[0075] To perform intra-frame prediction, the video encoder 200 can select an intra-frame prediction mode to generate prediction blocks. Some examples of VVC provide sixty-seven intra-frame prediction modes, including various directional modes, as well as planar and DC modes. Typically, the video encoder 200 selects an intra-frame prediction mode that describes the neighboring samples of the current block (e.g., the block of the CU), from which it predicts samples for the current block. Assuming the video encoder 200 decodes the CTU and CU in raster scan order (from left to right, from top to bottom), such samples are typically located above, to the upper left, or to the left of the current block within the same frame as the current block.

[0076] The video encoder 200 encodes data representing the prediction mode of the current block. For example, for inter-frame prediction modes, the video encoder 200 may encode data indicating which of the various available inter-frame prediction modes is used, as well as the motion information for the corresponding mode. For example, for unidirectional or bidirectional inter-frame prediction, the video encoder 200 may encode motion vectors using Advanced Motion Vector Prediction (AMVP) or merging modes. The video encoder 200 may use similar modes to encode motion vectors used for affine motion compensation modes.

[0077] AV1 includes two common techniques for encoding and decoding blocks of video data. These two common techniques are intra-frame prediction (e.g., intra-frame prediction or spatial prediction) and inter-frame prediction (e.g., inter-frame prediction or temporal prediction). In the context of AV1, when using intra-frame prediction modes to predict blocks of video data for the current frame, the video encoder 200 and video decoder 300 do not use video data from other frames of the video data. For most intra-frame prediction modes, the video encoder 200 encodes blocks of the current frame based on the difference between sample values ​​in the current block and predicted values ​​generated from reference samples in the same frame. The video encoder 200 determines the predicted values ​​generated from the reference samples based on the intra-frame prediction mode.

[0078] After prediction (such as intra-frame or inter-frame prediction for a block), the video encoder 200 can compute residual data for the block. The residual data (such as a residual block) represents the sample-by-sample difference between the block and the prediction block used to form the block, which is formed using the corresponding prediction mode. The video encoder 200 can apply one or more transforms to the residual block to produce transformed data in the transform domain rather than the sample domain. For example, the video encoder 200 can apply a Discrete Cosine Transform (DCT), an integer transform, a wavelet transform, or a conceptually similar transform to the residual video data. Additionally, the video encoder 200 can apply a secondary transform after the first transform, such as the Mode Correlated Inseparable Secondary Transform (MDNSST), the Signal Correlation Transform, the Karhunen-Loeve Transform (KLT), etc. The video encoder 200 produces transform coefficients after applying one or more transforms.

[0079] As noted above, after any transform that produces the transform coefficients, the video encoder 200 may perform quantization on the transform coefficients. Quantization generally refers to a process in which the transform coefficients are quantized to reduce the amount of data used to represent them, thereby providing further compression. By performing the quantization process, the video encoder 200 may reduce the bit depth associated with some or all of the transform coefficients. For example, the video encoder 200 may round an n-bit value down to an m-bit value during quantization, where n is greater than m. In some examples, to perform quantization, the video encoder 200 may perform a bitwise right shift on the value to be quantized.

[0080] After quantization, the video encoder 200 can scan the transform coefficients to generate a one-dimensional vector from a two-dimensional matrix containing the quantized transform coefficients. The scan can be designed to place higher-energy (and therefore lower-frequency) transform coefficients before the vector and lower-energy (and therefore higher-frequency) transform coefficients after the vector. In some examples, the video encoder 200 can utilize a predefined scan order to scan the quantized transform coefficients to produce a serialized vector, and then entropy-encode the quantized transform coefficients of that vector. In other examples, the video encoder 200 can perform an adaptive scan. After scanning the quantized transform coefficients to form a one-dimensional vector, the video encoder 200 can entropy-encode the one-dimensional vector, for example, according to context-adaptive binary arithmetic decoding (CABAC). The video encoder 200 can also entropy-encode the values ​​of syntax elements describing metadata associated with the encoded video data, which is used by the video decoder 300 when decoding the video data.

[0081] To perform CABAC, the video encoder 200 can assign context within a context model to the symbols to be transmitted. Context may involve, for example, whether the neighboring values ​​of a symbol are zero. Probability determination can be based on the context assigned to the symbols.

[0082] The video encoder 200 may further generate syntax data for the video decoder 300, such as block-based syntax data, image-based syntax data, and sequence-based syntax data, for example, in image headers, block headers, and slice headers, or generate other syntax data such as sequence parameter sets (SPS), image parameter sets (PPS), or video parameter sets (VPS). The video decoder 300 may also decode such syntax data to determine how to decode the corresponding video data.

[0083] In this way, the video encoder 200 can generate a bitstream that includes encoded video data, such as syntax elements describing the partitioning of images into blocks (e.g., CUs) and prediction and / or residual information for the blocks. Finally, the video decoder 300 can receive the bitstream and decode the encoded video data.

[0084] Typically, the video decoder 300 performs the reverse process of the video encoder 200 to decode the encoded video data of the bitstream. For example, the video decoder 300 can use CABAC to decode the values ​​of syntax elements used for the bitstream in a manner substantially similar to but reversed by the CABAC encoding process of the video encoder 200. Syntax elements can define partitioning information for dividing a picture into CTUs and defining the CUs of each CTU according to a corresponding partitioning structure such as a QTBT structure. Syntax elements can further define prediction and residual information for video data blocks (e.g., CUs).

[0085] The residual information can be represented, for example, by quantized transform coefficients. The video decoder 300 can inversely quantize and inverse transform the quantized transform coefficients of the block to reconstruct the residual block for the block. The video decoder 300 uses a signaling prediction mode (intra-frame prediction or inter-frame prediction) and associated prediction information (e.g., motion information for inter-frame prediction) to form a prediction block for the block. The video decoder 300 can then combine the prediction block and the residual block (on a sample-by-sample basis) to reconstruct the original block. The video decoder 300 can perform additional processing, such as performing a deblocking process to reduce visual artifacts along the block boundaries.

[0086] This disclosure may generally relate to "signaling" certain information (such as syntax elements). The term "signaling" can generally refer to the communication of values ​​and / or other data of syntax elements used to decode encoded video data. That is, video encoder 200 may signal the values ​​of syntax elements in the bitstream. Generally, signaling refers to generating values ​​in the bitstream. As noted above, source device 102 may transmit the bitstream to destination device 116 substantially in real time or not in real time (such as when syntax elements are stored in storage device 112 for later retrieval by destination device 116).

[0087] According to the technology disclosed herein, a method for decoding video data includes: receiving an image of the video data; reconstructing the image of the video data; and performing a neural network (NN)-based filtering process on the reconstructed image of the video data, wherein the NN-based filtering process includes performing positive multivariate (CP) decomposition using separable convolution.

[0088] Video decoding standards include ITU-T H.261, ISO / IEC MPEG-1 Visual, ITU-T H.262 or ISO / IEC MPEG-2 Visual, ITU-T H.263, ISO / IEC MPEG-4 Visual, and ITU-T H.264 (also known as ISO / IEC MPEG-4 AVC), High-Efficiency Video Decoding (HEVC), or ITU-T H.265, including its Range Extension, Multi-View Extension (MV-HEVC), and Scalable Extension (SHVC). Multi-Functional Video Decoding (VVC), or ITU-T H.266, was recently developed by the Joint Video Experts Group (JVET) of the ITU-T Video Decoding Experts Group (VCEG) and the ISO / IEC Moving Picture Experts Group (MPEG).

[0089] The VVC specification has been finalized and is hereafter referred to as VVC FDIS, available at phenix.int-evry.fr / jvet / doc_end_user / documents / 19_Teleconference / wg11 / JVET-S2001-v17.

[0090] Figure 2 This is a conceptual diagram illustrating a hybrid video decoding framework. Video decoding standards since H.261 are based on the so-called hybrid video decoding principle, which... Figure 2 As shown in the diagram. The term hybrid refers to a combination of two methods for reducing redundancy in video signals: predictive and transform decoding with predictive residual quantization. Prediction and transform reduce redundancy in video signals by decorrelation, while quantization reduces the data represented by the transform coefficients by decreasing their precision, ideally by removing only irrelevant details. This hybrid video decoding design principle is also used in two recent standards, ITU-T H.265 / HEVC and ITU-T H.266 / VVC.

[0091] like Figure 2 As shown, a modern hybrid video decoder 130 typically performs block partitioning, motion compensation or inter-picture prediction, intra-picture prediction, transform, quantization, entropy decoding, and post-loop / intra-loop filtering. Figure 2 In the example, the video decoder 130 includes a summing unit 134, a transform unit 136, a quantization unit 138, an entropy decoding unit 140, an inverse quantization unit 142, an inverse transform unit 144, a summing unit 146, a loop filter unit 148, a decoded picture buffer (DPB) 150, an intra-frame prediction unit 152, an inter-frame prediction unit 154, and a motion estimation unit 156.

[0092] Generally, the video decoder 130 receives input video data 132 while encoding video data. Block partitioning is used to divide the received video data images into smaller blocks for use in the prediction and transformation processes. Early video decoding standards used a fixed block size, typically 16×16 samples. More recent standards such as HEVC and VVC employ tree-based partitioning structures to provide flexible partitioning.

[0093] Motion estimation unit 156 and inter-frame prediction unit 154 can predict input video data 132, for example, based on previously decoded data from DPB 150. Motion compensation, or inter-picture prediction, utilizes the redundancy present between pictures in the video sequence (hence the term "inter-picture"). According to block-based motion compensation used in modern video codecs, predictions are obtained from one or more previously decoded pictures (i.e., reference pictures). The corresponding regions used to generate inter-frame predictions are indicated by motion information, including motion vectors and reference picture indices.

[0094] In recent video codecs, a hierarchical prediction structure within group of pictures (GOPs) is applied to improve decoding efficiency. Figure 3 An example hierarchical prediction structure 162 with a group of pictures (GOP) size of 16 is shown.

[0095] Refer again Figure 2 In-picture prediction utilizes the spatial redundancy present in the picture (hence the term "in-picture") by deriving predictions for blocks from spatially adjacent (reference) samples that have already been decoded / decoded. Angle prediction, DC prediction, and planar or on-plane prediction are used in modern video codecs, including AVC, HEVC, and VVC.

[0096] Hybrid video decoding standards apply block transforms to prediction residuals (regardless of whether the prediction residuals come from inter-picture or intra-picture predictions). Early standards (including H.261 / 262 / 263) used Discrete Cosine Transform (DCT). In HEVC and VVC, more transform kernels besides DCT can be applied to handle different statistical information in specific video signals.

[0097] Quantization aims to reduce the precision of an input value or set of input values ​​in order to reduce the amount of data required to represent those values. In hybrid video decoding, quantization is typically applied to individual transformed residual samples (e.g., transform coefficients) resulting in integers at the coefficient level. In recent video decoding standards, the stride is derived from a so-called quantization parameter (QP) that controls fidelity and bit rate. A larger stride reduces the bit rate but also degrades quality, causing, for example, block artifacts and blurred details in video images.

[0098] The summing unit 134 calculates the residual data as the difference between the input video data 132 and the predicted data from the intra-frame prediction unit 152 or the inter-frame prediction unit 154. The summing unit 134 provides the residual block to the transform unit 136, which applies one or more transforms to the residual block to generate a transform block. The quantization unit 138 quantizes the transform block to form quantized transform coefficients. The entropy decoding unit 140 entropy-encodes the quantized transform coefficients and other syntax elements (such as motion information or intra-frame prediction information) to generate an output bitstream 158.

[0099] Simultaneously, the inverse quantization unit 142 inverse quantizes the quantized transform coefficients, and the inverse transform unit 144 inverse transforms the transform coefficients to reproduce the residual block. The summation unit 146 combines the residual block with the prediction block (on a sample-by-sample basis) to produce a decoded video data block. The loop filter unit 148 applies one or more filters (e.g., at least one of a neural network-based filter, a neural network-based loop filter, a neural network-based post-loop filter, an adaptive in-loop filter, or a predefined adaptive in-loop filter) to the decoded block to produce a filtered decoded block.

[0100] The entropy decoding unit 140 can perform context-adaptive binary arithmetic decoding (CABAC) on the encoded video. CABAC is used in recent video codecs such as AVC, HEVC, and VVC due to its high efficiency.

[0101] The loop filter unit 148 can perform post-loop or in-loop filtering. Post-loop / in-loop filtering is a filtering process (or a combination of such processes) applied to the reconstructed image to reduce decoding artifacts. The input to the filtering process is generally the reconstructed image, which is a combination of the reconstructed residual signal (including quantization errors) and the prediction. Figure 2 As shown, the reconstructed image after in-loop filtering is stored in the decoded image buffer (DPB) 150 and used as a reference for inter-image prediction of subsequent images. Decoding artifacts are largely determined by the QP (Quick Pointer). Therefore, QP information is typically used in the design of the filtering process. In HEVC, the in-loop filter includes deblocking filtering and Sample Adaptive Offset (SAO) filtering. In the VVC standard, an Adaptive Loop Filter (ALF) is introduced as a third filter. The ALF filtering process is shown below:

[0102] (1)

[0103] in These are samples before the filtering process. These are the sample values ​​after the filtering process. Represents the filter coefficients. It is a clipping function, and This represents the clipping parameters. Variables k and l are... and The value varies between these values, where L represents the filter length. (Pruning function) It corresponds to the function The pruning operation introduces nonlinearity to make ALF more efficient by reducing the influence of neighboring sample values ​​that are very different from the current sample value. In VVC, the filtering parameters can be signaled in the bitstream and can be selected from a predefined filter bank. The ALF filtering process can also be summarized using the following formula:

[0104] (2)

[0105] Figure 4 This is a conceptual diagram illustrating a neural network-based filter 170 with four layers. Various studies have shown that embedding neural networks into, for example... Figure 2 Compression efficiency can be improved in hybrid video decoding frameworks. Neural networks have been used for intra-frame prediction and inter-frame prediction to improve prediction efficiency. In recent years, NN-based intra-loop filtering has also been a well-known research topic. In some examples, the filtering process is applied as a post-filter. In such examples, the filtering process is applied to the output image, and the unfiltered image can be used as a reference image.

[0106] In addition to existing filters such as deblocking filters, sample adaptive offset (SAO), and / or adaptive loop filtering (ALF), NN-based filters 170 can also be applied. NN-based filters can also be applied exclusively, where the NN-based filters are designed to replace all existing filters. Additionally or alternatively, NN-based filters (such as NN-based filter 170) can be designed to supplement, enhance, or replace any or all other filters.

[0107] Figure 4 The neural network-based filtering process can take reconstructed samples (e.g., luminance and chrominance samples, which in some examples can be packed into a 3D volume with 6 planes) as input, and the intermediate outputs are residual samples, which are added back to the input to refine the input samples. The neural network-based filter can use all color components (e.g., Y, U, and V, or Y, Cb, and Cr, e.g., luminance data 172A, blue hue chrominance 172B, and red hue chrominance 172C) as input to take advantage of cross-component correlations. Different color components can share the same filters (including network structure and model parameters), or each component can have its own specific filters.

[0108] The filtering process can also be summarized as follows:

[0109] (3)

[0110] The model structure and parameters of the NN-based filter can be predefined and stored in the video encoder 200 and video decoder 300. The filter can also be signaled in the bitstream.

[0111] exist Figure 4 In the example, a neural network-based filter may include a series of feature extraction layers followed by an output convolution. Figure 4In a convolutional layer, the feature extraction layer may consist of a 3x3 convolutional (conv) layer followed by a parameter-corrected linear unit (PReLU) layer. The convolutional layer applies a convolution operation to the input data, which involves a filter or kernel sliding across the input data (e.g., a reconstructed sample from input 172) and computing the dot product at each locality. The convolutional operation essentially captures local patterns within the input data. For example, in the context of image processing, these patterns could be edges, textures, or other visual features. The filter or kernel is a small weight matrix that is updated during training. By making this filter slide across the input data (or feature maps from previous layers) and computing the dot product at each locality, the convolutional layer creates a feature map that encodes the spatial hierarchies and patterns detected in the input.

[0112] The output of a convolutional layer is a set of feature maps, each corresponding to a filter that captures a different aspect of the input data. As the data passes through deeper layers of the network, this layer helps the neural network learn increasingly complex and abstract features. Figure 4 In the naming of 3x3 conv 3x3x6x8, the first 3x3 indicates that the convolutional layer has a 3x3 filter size (e.g., a 3x3 matrix). 3x3x6x8 refers to the input and output dimensions of the convolutional layer, where 6 is the number of input channels and 8 is the number of output channels.

[0113] The PReLU layer is an activation function used in neural networks and is introduced as a variant of the ReLU (Modified Linear Unit) activation function. As described above, convolutional layers output feature maps, each corresponding to a filter representing a feature detected in the input. After the convolutional layers, the PReLU layer applies the PReLU activation function to each element of the feature map produced by the convolutional layers. For positive values, the PReLU layer acts like standard ReLU, passing the value through. For negative values, the PReLU layer allows small, linear negative outputs instead of setting them to zero (e.g., like ReLU). This preserves neuronal activity and gradient flow, which is beneficial for the learning of deep networks.

[0114] In summary, when a convolutional layer is followed by a PReLU layer, the convolutional layer first extracts features from the input data through a set of learned filters. The resulting feature map is then passed through the PReLU activation function, which introduces non-linearity and helps avoid the problem of dead neurons by allowing small gradients when the input is negative. This combination is effective on complex patterns in the learning data while maintaining a robust gradient flow, which is particularly beneficial for deeper network architectures.

[0115] When neural network-based filtering is applied to video decoding, the entire video signal (pixel data) may be divided into multiple processing units (e.g., 2D blocks), and each processing unit may be processed separately or combined with other information associated with that pixel block. Possible choices for processing units include frames, slices / tiles, CTUs, or any predefined or signaled shape and size. Generally, neural network-based filtering is performed on reconstructed blocks of video data. Here, reconstructed blocks and samples can refer to both the decoded blocks generated by the video decoder 300 and the blocks reconstructed in the reconstruction loop of the video encoder 200.

[0116] To further improve the performance of neural network-based filtering, different types of input data can be processed together to produce filtered outputs. Input data may include, but is not limited to, reconstructed pixels / samples, predicted pixels / samples, pixels / samples after loop filtering, partitioning structure information, deblocking parameters (e.g., boundary strength (BS)), QP values, slice or image type, or filter applicability or decoding mode maps. Input data can be provided at different granularities. Luminosity reconstruction and predicted samples can be provided at the original resolution, while chromosity samples can be provided at a lower resolution (e.g., for a 4:2:0 representation) or upsampled to luminosity resolution for per-pixel representation. Similarly, QP, BS, partitioning, or decoding mode information can be provided at lower resolutions, including per frame, slice, or processed block with a single value (e.g., QP). In other examples, QP, BS, partitioning, or decoding mode information can be expanded (e.g., duplicated) to achieve per-pixel / sample representation.

[0117] Figure 5 The diagram shows an example of an architecture that utilizes supplementary data. Figure 5 This is a block diagram illustrating an example CNN-based filter with padded input samples and supplementary data. The NN-based filter 171 uses pixels / samples in a processing block combined with supplementary data as input 174. Input 174 may include four sub-blocks of interleaved luminance samples (Yx4) 174A and associated blue hue chromaticity (U) data 174B and red hue chromaticity (V) data 174C. Supplementary data includes quantization parameter (QP) step size 176 and boundary strength (BS) 178. The region of input pixels / samples can be expanded by four padded pixels / samples on each side. The resulting size of the processing volume is (4+64+4) x (4+64+4) x (4 Y+2UV+1QP + 3BS).

[0118] Compared to Figure 4The neural network-based filter 171 may include two or more hidden layers utilizing both 1x1 convolutions and leaky ReLU layers. The leaky ReLU layer, similar to the PReLU layer, allows for small non-zero gradients to be output when the layer is inactive. Instead of outputting zero for negative inputs, the leaky ReLU multiplies them by a small constant. This small slope ensures that even normally inactive neurons still contribute a small amount to the network's learning, reducing the likelihood of the dead ReLU problem.

[0119] To further improve the performance of neural network-based filtering, multi-mode solutions can be designed. For example, for each processing unit, the video encoder 200 can select from a set of modes based on rate-distortion optimization, and the selection can be signaled in the bitstream. Different modes may include different neural network models, different values ​​of input information used as the neural network models, etc. In one example, the video encoder 200 and the video decoder 300 can use a neural network-based filtering solution based on a single neural network model by using different QP values ​​as inputs to the neural network models for different modes.

[0120] An example of the CNN ILF architecture that JVET is developing is now described. One such architecture is a multimodal CNN ILF with attention blocks. In JVET-Z0113, as described above, in April 2022, a multimodal NN-based filtering solution was disclosed regarding NN-based filtering with multimodal design, Y. Li, K. Zhang, L. Zhang, H. Wang, M. Coban, AM Kotra, M. Karczewicz, F. Galpin, K. Andersson, J. Ström, D. Liu, R. Sjöberg, EE1-1.7: Combined testing of EE1-1.6 and EE1-1.3, JVET-Z0113 (hereinafter referred to as "JVET-Z0113").

[0121] In one example, a neural network-based filtering solution with multiple modes can be used, as described above. Figure 6 The structure of a neural network is shown in the diagram. Figure 6 In the first part, different input data types are convolved using multiple 3x3 (conv3x3) kernels to produce feature maps. The convolutional inputs are activated by parameter-corrected linear units (PReLU), and then the results for each data type are concatenated, fused, and subsampled (transformed) to create the output y. This output is then fed through attention residual (AttRes) blocks, each block having... Figure 6The structure is shown. The output from the last attention residual block z is fed into the final part of the network.

[0122] For example, Figure 6 The NN-based filter includes a first part comprising input 3x3 convolutions 510A to 510E and corresponding parameter-corrected linear units (PReLU) 512A to 512E for each input to generate feature maps (e.g., feature extraction segments of the NN filter). A cascade unit 514 cascades the feature maps and provides them to a fusion block 516 and a transformation block 522. Figure 6 The NN-based filter further includes: a set 528 of attention residual (AttRes) blocks 530A to 530N; and a final portion (e.g., a tail segment) comprising 3x3 convolutions 550, PReLU 552, 3x3 convolutions 554, and pixel rearrangement units 556. The AttRes blocks may also be referred to as backbone blocks.

[0123] In the first part (e.g., the feature extraction section), different inputs are received (including quantization parameters (QP) 500, partition information (part) 502, boundary strength (BS) 504, predicted samples (pred) 506, and reconstructed samples (rec) 508). Corresponding 3x3 convolutions 510A to 510E and PReLU 512A to 512E convolve the corresponding inputs and activate them to produce feature maps. A concatenation unit 514 then concatenates the feature maps. A fusion block 516, including a 1x1 convolution 518 and PReLU 520, fuses the concatenated feature maps. A transformation block, including a 3x3 convolution 524 and PReLU 526, subsamples the fused input to create output 188. Output 188 is then fed through a set 528 of attention residual blocks 530A to 530N, which may include a variety of numbers of attention residual blocks, such as eight. The attention blocks are related to... Figure 7 Further explanation. The output 189 from the last of the set 528 of attention residual blocks 530 is fed into the final part of the NN-based filter. In the final part (which may be the tail block), a 3x3 convolution 550, PReLU 552, a 3x3 convolution 554, and a pixel rearrangement unit 556 process the output 189, and an addition unit 558 combines the result with the original input reconstructed sample 508. This ultimately forms the filtered output for rendering and is stored as a reference for subsequent inter-frame predictions, for example, in the decoded image buffer (DPB). In some examples, Figure 6 The NN-based filter uses 96 feature maps.

[0124] Figure 7 This is an example Figure 6 A conceptual diagram of the attention residual block. That is, Figure 7 Attention residual block 530 is depicted, which may include similar to Figure 6 The attention residual blocks 530A to 530N are components of the components. In this example, the attention residual block 530 includes a first 3x3 convolution 532, a parametric rectified linear unit (PReLU) filter 534, a second 3x3 convolution 536, an attention block 538, and an adder unit 540. The adder unit 540 combines the output of the attention block 538 with the output 188 initially received by the convolution 532 to generate an output 189.

[0125] Figure 8 This is an example Figure 7 A conceptual diagram of an example spatial attention layer. (See also:) Figure 8 As shown, the spatial attention layer of the attention residual block 530 includes a 3x3 convolution 706, a PReLU 708, a 3x3 convolution 710, a size expansion unit 712, a 3x3 convolution 720, a PReLU 722, and a 3x3 convolution 724. The 3x3 convolution 706 receives input 702, which corresponds to... Figure 6 The quantization parameters (QP) are 500, partition information (part) is 502, boundary strength (BS) is 504, prediction information (pred) is 506, and reconstructed samples (rec) are 508. A 3x3 convolution with a 720-bit reception Z-axis is used. K 704. The outputs of the size expansion unit 712 and the 3x3 convolution 724 are combined, and then combined with the R value 730 to generate the S value 732. The S value 732 is then combined with the Z... K Values ​​704 are combined to generate output Z. K+1 Value 734.

[0126] Another example of a neural network architecture is a reduced-complexity multimodal CNN ILF. In JVET-AC0155, an alternative design for a neural network architecture was proposed in January 2023, S. Eadie, M. Coban, M. Karczewicz, EE1-1.9: Reduced-complexity CNN-based In-Loop Filtering, JVET-AC0155. For example, it can be used... Figure 6 The example filter has a larger number of low-complexity residual blocks in its backbone, along with a reduction in the number of channels (feature maps) and the removal of the attention module. Figure 9 The proposed CNN-based filtering structure is shown in the figure (e.g., for brightness filtering).

[0127] Figure 9 This is a block diagram illustrating an example of a simplified CNN-based filter architecture with padded input samples and supplementary data. Figure 9The NN-based filters include 3x3 convolutions 810A to 810E and PReLU 812A to 812E, which convolve the corresponding inputs (i.e., QP 800, Part 802, BS 804, Pred 806, and Rec 808) to generate feature maps (e.g., feature extraction segments). A concatenation unit 814 concatenates the convolutional inputs (e.g., feature maps). A fusion block 816 then uses a 1x1 convolution 818 and PReLU 820 to fuse the concatenated feature maps. A transformation block 822 then processes the fused data using a 3x3 convolution 824 and PReLU 826.

[0128] In this example, the NN-based filter comprises a set 828 of residual blocks 830A to 830N (also referred to as backbone blocks), each of which can be configured according to... Figure 10 The residual block structure 830 is structured as discussed below. Residual blocks 830A to 830N are replaceable. Figure 6 AttRes blocks 530A to 530N. Figure 9 The example can be used for luminance (luminance) filtering, but similar modifications can be made for chrominance (chrominance) filtering as discussed below.

[0129] Figure 9 The number of residual blocks and channels included in set 828 can be configured differently. That is, N can be set to different values, and the number of channels in residual block structure 830 can be set to a number other than 160 to achieve different performance-complexity tradeoffs. These modifications can be used to perform chroma filtering to process chroma channels.

[0130] The set 828 of residual blocks 830A to 830N has N instances of residual block structure 830. In one example, N can be equal to 32, such that there are 32 residual block structures. Residual blocks 830A to 830N can use 64 feature maps, which are relative to... Figure 6 The number of feature maps used in the example is reduced from 96.

[0131] exist Figure 9 In the final part, the output of set 828 is processed by 3x3 convolution 850, PreLU 852, 3x3 convolution 854, and pixel rearrangement unit 856, and the addition unit 858 combines the result with the original input reconstructed sample (REC) 808. This ultimately forms a filtered output for rendering and is stored as a reference for subsequent inter-frame predictions, for example, in the decoded picture buffer (DPB).

[0132] In one example, Figure 9In the example, the number of residual blocks and feature maps (convolutions) used are M=24 and 64, respectively. In residual blocks 830A to 830N, the number of channels first increases to 160 before the activation layer and then decreases to 64 after the activation layer. For different performance complexity tradeoffs, the number of residual blocks and channels can be configured differently (M is set to another value, and the number of channels in the residual block can be set to a number other than 160). Chroma filtering can follow... Figure 6 The concept involves modifying its core as described above to handle chroma channels.

[0133] In another NN architecture, residual blocks can be replaced by filter blocks, such as... Figure 11 The following will be discussed later in this disclosure. In this example, bypass branches around the convolutional and activation layers in the residual blocks from the previous solution are removed. The number of channels and the number of filter blocks can be configurable, for example, 64 channels and 24 filter blocks, with 160 channels before and after activation. This results in a network complexity of 605.93kMAC and 1.5M parameters for an intra-frame luma model.

[0134] Another example neural network architecture is the multimodal CNN ILF utilizing separable convolutions. Further complexity reduction of the CCN ILF architecture can be achieved by using separable convolutions instead of 2D convolutions (3x3). In JVET-AD0023, EE1 test 1.3.5, the low-rank convolution approximation decomposes the 3x3xMxN convolution into pixel-wise convolutions (1x1xMxR), two separable convolutions (3x1xRxR, 1x3xRxR), and another pixel-wise convolution (1x1xRxN) for the residual blocks of the architecture described in JVET-AC0155. In this example, R is the approximate rank, and the performance / complexity of the approximation is absolved.

[0135] Figure 10 This is an example Figure 9 A conceptual diagram of an example residual block structure 830 is provided. In this example, the residual block structure 830 includes a first 1x1 convolution 832, which increases the number of input channels to 160 before the activation layer (PReLU 834) processes the input channels. PReLU 834 can then reduce the number of channels to 64 through this processing. A second 1x1 convolution 836 then processes the reduced channels, followed by a 3x3 convolution 838. Finally, a combining unit 840 combines the output of the 3x3 convolution 838 with the original input received by the residual block structure 830.

[0136] In another neural network architecture, residual blocks can be replaced by filter blocks (also known as backbone blocks), such as Figure 11 As shown. In this example, bypass branches around all layers in each residual block are removed, as... Figure 12 As shown. The number of channels and filter blocks is configurable, for example, 64 channels and 24 filter blocks, with 160 channels before and after activation. In some examples, for intra-frame luma models, the complexity of such networks is 605.93kMAC and the number of parameters is 1.5M.

[0137] Figure 11 This illustrates alternatives to the technology according to this disclosure. Figure 6 A conceptual diagram of another example filter block structure for an attention residual block set. Figure 11 The NN-based filters include 3x3 convolutions 1010A to 1010E and PReLU 1012A to 1012E, which convolve the corresponding inputs (i.e., QP 1000, Part 1002, BS 1004, Pred 1006, and Rec 1008) to form feature maps (e.g., feature extraction segments). Concatenation unit 1014 concatenates the feature maps. Then, fusion block 1016 uses 1x1 convolution 1018 and PReLU 1020 to fuse the concatenated inputs. Transformation block 1022 then uses 3x3 convolution 1024 and PReLU 1026 to process the fused data.

[0138] In this example, the NN-based filtering unit comprises a set 1028 of N filter blocks 1030A to 1030N (also referred to as the backbone block), each of which may have Figure 12 The structure of filter block 1030 is discussed below. Filter block structure 1030 is substantially similar to residual block structure 830, except that the combination unit 840 is omitted from filter block structure 1030, so that the input is not combined with the output. Instead, the output of each residual block structure can be directly fed to subsequent blocks.

[0139] exist Figure 11 In the final part, 3x3 convolution 1050, PreLU 1052, 3x3 convolution 1054, and pixel rearrangement unit 1056 process the output of set 1028, and addition unit 1058 combines the result with the original input reconstructed sample (REC) 1008. This ultimately forms a filtered output for rendering and is stored as a reference for subsequent inter-frame predictions, for example, in the decoded picture buffer (DPB).

[0140] In one example Figure 11 The architecture is implemented using parameters K=64, M=160 and R=51, and the total number of 24 residual blocks results in a network complexity of 356.43kMAC and 1.07M parameters for the intra-frame luminance model.

[0141] Figure 12 This is an example Figure 11 A conceptual diagram of an example filter block structure 1030 is provided. In this example, the filter block structure 1030 includes a first 1x1 convolution 1032, which increases the number of input channels to 160 before the activation layer (PReLU 1034) processes the input channels. The PReLU filter 1034 can then reduce the number of channels to 64 through this processing. A second 1x1 convolution 1036 then processes the reduced number of channels, followed by a 3x3 convolution 1038. As discussed above, with... Figure 10 Compared to the residual block structure 830, the filter block structure 1030 does not include a combination unit.

[0142] Convolutions with 3x3 kernels are prevalent in neural network-based filters. In the architecture described above, 3x3xNxM convolutions are utilized across multiple segments and blocks, where the 3x3 kernel slides in the spatial (2D) domain. However, multidimensional convolutions (such as 2D kernel convolutions) introduce significant complexity. According to the techniques disclosed herein, video encoder 200 and video decoder 300 can be configured to utilize separable convolutions instead of multidimensional convolutions (e.g., 3x3xNxM convolutions), or to utilize separable convolutions instead of multidimensional convolutions. For example, in any segment of an neural network-based filter, two separable one-dimensional convolutions can be used instead of a 3x3 convolution. Using separable convolutions reduces computational complexity and memory bandwidth requirements.

[0143] To avoid overcomputation and reduce the parameter set derived from multidimensional convolutions (such as 3x3 convolutions in CNN architectures described above or similar, or 2D convolutional quantities with higher-dimensional kernels)), this disclosure describes a technique for video encoder 200 and video decoder 300 configured to utilize separable convolutions (e.g., 1D separable convolutions) generated by low-complexity approximations, rather than multidimensional (e.g., 2D) convolutions that slide in spatial directions. While the techniques of this disclosure are described with reference to 3x3 convolutions, the decomposition techniques of this disclosure can be used for multidimensional convolutions of any size (e.g., 4x4, 5x5, or larger). Typically, a multidimensional convolution has a kernel size of n1xn2 in the spatial dimension, where n1 and n2 are positive integers. The values ​​of n1 and n2 can be the same or different. A multidimensional convolution can further have a size of K in the depth dimension (e.g., n1xn2xK). Additionally, utilizing the number of output channels M, a multidimensional convolution can be represented as a 4-dimensional tensor of n1xn2xKxM.

[0144] In one example of this disclosure, the low-rank convolution approximation decomposes a 3x3xMxN convolution into a pixel-wise convolution (1x1xMxR), two separable convolutions (3x1xRxR, 1x3xRxR), and another pixel-wise convolution (1x1xRxN). Here, R is the rank of the approximation and can be used to adjust the performance / complexity of the approximation. The value of R can be an integer. In some examples, R can be derived as a function (ratio) of M or N, or max(M,N). In some examples, R can be set to equal A*max(M,N), where A is less than 1 (e.g., 0.2, 0.5, 0.8), A is greater than 1 (e.g., 1.0, 1.2), or other values.

[0145] In a general example, multidimensional convolution can be approximated by multiple separable convolutions, which can be performed by applying a first convolution of size n1x1 to the output of the first convolution and then applying a second convolution of size 1xn2.

[0146] Figure 13 This is a block diagram illustrating an example multi-scale feature extraction backbone network with two-dimensional convolution. Figure 13 Examples can be similar to Figure 10 The example uses 3x1xKxR convolution and 1x3xRxK convolution to approximate 3x3xKxK convolution.

[0147] exist Figure 13 In the example, residual block 1420 comprises a 1x1xKxM convolution 1402, followed by a PReLU 1404. The output of PReLU 1404 is fed into a 1x1xMxK convolution 1406. The 3x3xKxK convolution 1408 of residual block 1420 is approximated by a 3x1xKxR convolution 1400 and a subsequent 1x3xRxK convolution 1410. The output of the 1x3xRxK convolution 1410 can be fed into a combining unit 1412, which combines the output of the 1x3xRxK convolution 1410 with the input to the 1x1xKxM convolution 1402. R is the positive quasi-rank of the decomposition. A lower rank implies a greater reduction in complexity.

[0148] Figure 14 This is a block diagram illustrating an example unified filter with a joint model (joint luminance and chrominance). A method utilizing... (The sentence is incomplete and requires more context to translate accurately.) Figure 14Multi-scale feature extraction of a 2D convolutional network is illustrated in Y. Li, S. Eadie, D. Rusanovskyy, M. Karczewicz, EE1-related: Multi-scale component testing of EE1-1.6 and combined testing of EE1-1.3.5, JVET-AD0211 (JVET-AD0211). A 3x3 convolution can be decomposed into a 3x1xC1xR convolution followed by a 1x3xRxC2 convolution, where C1 and C2 are the number of input and output channels, respectively, and R is the approximate rank (see V. Lebedev, Y. Ganin, M. Rakhuba, I. Oseledets, V. Lempitsky, Speeding up Convolution Neural Networks Using Fine-tuned CP-Decomposition), ICLR 2015, available online: https: / / arxiv.org / pdf / 1412.6553. In some examples, the parameter R can be derived proportionally to R = C1 x C2 / (C1 + C2), thus controlling the complexity of the approximation.

[0149] Figure 14 The proposed architecture is shown where 3x3 convolutional blocks are replaced by separable convolutions of 3x1 and 1x3. In this example, the residual block structure 1430 includes a first 1x1 convolution 1432 before a first activation layer (PReLU 1434), and a 3x3 convolution 1440 with the first 1x1 convolution 1432 and PReLU 1434, followed by a second activation layer (PReLU 1442). The second 1x1 convolution 1436 then processes the combined output of PReLU 1434 and PReLU 1442, followed by a 3x3 convolution 1438. However, in Figure 14 In the example, a 3x3 convolution 1440 can be approximated using multiple separable convolutions, as shown below. Figure 14 The 3x1 convolution 1450 and 1x3 convolution 1452 are shown in the figure. Similarly, the 3x3 convolution 1438 can be approximated by multiple separable convolutions, as shown in the figure. Figure 14 The 3x1 convolution is 1460 and the 1x3 convolution is 1462.

[0150] For example, Figure 14 The illustrated architecture can be implemented using parameters R1=8, R2=44, M1=160 and M2=16, and with a total of 24 residual blocks, the network complexity will be 358.43kMAC and the number of parameters will be 1.07M for the intra-frame luminance model.

[0151] Figure 15This is a block diagram illustrating an example unified filter with a joint model (joint luminance and chrominance). Figure 15 NN-based filters may include IPB 1500 (which provides information about whether a block is inter-frame prediction or intra-frame prediction), QP SLICE 1502 (which provides quantization parameters for slices), QP BASE 1504 (which provides the quantization parameters for the sequence), Boundary Strength (BS) 1506, Predicted Samples (PRED) 1508, and Reconstructed Samples (REC) 1509. The video decoder 300 can apply corresponding 3x3 convolutions 1510A to 1510F to each input in the input. The video decoder 300 can apply corresponding PReLUs 1512A to 1512F to the output of the 3x3 convolutions 1510A to 1510F. In some examples, backbone blocks 1530A to 1530N can be residual blocks.

[0152] The fusion block 1516 uses a 1x1 convolution 1518 and PReLU 1520 to fuse the feature maps. Then, the transformation block 1522 uses a 3x3 convolution 1524 and PReLU 1526 to process the fused data.

[0153] The video decoder 300 may apply a set 1528 of backbone blocks. The set 1528 may include multiple filter blocks 1530A to 1530N (also referred to herein as backbone blocks). In some examples, N=24, resulting in 24 filter blocks. The video decoder 300 may apply a 3x3 convolution 1550 to the output of the set 1528. The video decoder 300 may apply a PReLU 1552 to the output of the 3x3 convolution 1550. The video decoder 300 may apply a 3x3 convolution 1554 to the output of the PReLU 1552. The video decoder 300 may crop 1558 of the output of the 3x3 convolution 1554 to generate filtered reconstructed UV samples of the image. The video decoder 300 may perform pixel rearrangement 1556 on the output of the 3x3 convolution 1554 and crop 1560 of the output of the pixel rearrangement 1556 to generate filtered reconstructed Y samples of the image. For example, pixel rearrangement 1556 can upsample its input so that the output of pixel rearrangement 1556 has a size of w*2, h*2, c / 4, where w is the width, h is the input height, and c is the number of input channels to the input of pixel rearrangement 1556. In some examples, Figure 15 The example parameters include one or more of the following: d1=192, d2=32, d3=16, d4=16, d5=16, C=64, and d6=48.

[0154] Figure 16 This is a block diagram illustrating the example backbone block. Figure 16The backbone block 1600 can be an example of any of the filter blocks 1530A to 1530N. The backbone block 1600 may include multi-scale branches 1620 that perform convolutions in parallel. For example, a 1x1 convolution 1602 may be performed in parallel with a 3x1 convolution 1604 followed by a 1x3 convolution 1606. The multi-scale branches 1620 may perform multi-scale feature extraction.

[0155] Video decoder 300 can apply PReLU 1608 to the output multi-scale branch 1620. Video decoder 300 can apply a 1x1 convolution 1610 to the output of PReLU 1608. Video decoder 300 can apply a 1x3 convolution 1612 to the output of a 1x1 convolution 1610. Video decoder 300 can apply a 3x1 convolution 1614 to the output of a 1x3 convolution 1612. In some examples, Figure 16 The example parameters include C=64, C1=160, C21=32, C22=32, and C31=64.

[0156] Figure 17 This is a block diagram (luminance) illustrating an example unified filter with separate luminance and chrominance models. Figure 17 NN-based filters may include IPB 1700, QP BASE 1702, BS 1704, PRED 1706 and reconstructed brightness samples (REC) EXT Y)1708. The video decoder 300 can apply the corresponding 3x3 convolutions of 1710A to 1710E to each input in the input. The video decoder 300 can apply the corresponding PReLU of 1712A to 1712E to the output of the 3x3 convolutions of 1710A to 1710E.

[0157] The fusion block 1716 uses a 1x1 convolution 1718 and PReLU 1720 to fuse the feature maps. Then, the transformation block 1722 uses a 3x3 convolution 1724 and PReLU 1726 to process the fused data.

[0158] The video decoder 300 can apply a set 1728 of backbone blocks. The set 1728 can include multiple filter blocks 1730A to 1730N (also referred to as backbone blocks). In some examples, N=20, resulting in 20 filter blocks. The video decoder 300 can apply a 3x3 convolution 1750 to the output of the set 1728. The video decoder 300 can apply a PReLU 1752 to the output of the 3x3 convolution 1750. The video decoder 300 can apply a 3x3 convolution 1754 to the output of the PReLU 1752. The video decoder 300 can crop the output of the 3x3 convolution 1754 to generate filtered reconstructed UV samples of the image. The video decoder 300 can perform pixel rearrangement 1756 on the output of the 3x3 convolution 1754 and crop the output of the pixel rearrangement 1756 to generate filtered reconstructed Y samples of the image.

[0159] One or more of the filter blocks 1730A to 1730N in the set 1728 can be Figure 16 An example of a 1600-size backbone block. In some examples, Figure 17 The example parameters and Figure 16 Examples together may include one or more of the following: d1=192, d2=32, d3=16, d4=16, d5=16, C=64, C1=160, C 21 =32、C 22 =32、C 31 =64, N=20 and d6=48.

[0160] Figure 18 This is a block diagram (chroma) illustrating an example unified filter with separate luminance and chrominance models. Figure 18 The input to the NN-based filter may include REC EXT Y 1800 (which can be downsampled by 2x), QP BASE 1804, BS 1806, PRED 1808 and reconstructed brightness samples (REC) EXT (UV) 1809. The video decoder 300 can apply the corresponding 3x3 convolutions of 1810A to 1810E to each input in the input. The video decoder 300 can apply the corresponding PReLUs of 1812A to 1812E to the output of the 3x3 convolutions of 1810A to 1810E.

[0161] The fusion block 1816 uses a 1x1 convolution 1818 and PReLU 1820 to fuse the feature maps. Then, the transformation block 1822 uses a 3x3 convolution 1824 and PReLU 1826 to process the fused data.

[0162] The video decoder 300 can apply a set 1828 of backbone blocks. The set 1828 can include multiple filter blocks 1830A to 1830N (also referred to as backbone blocks). In some examples, N=16, resulting in 24 filter blocks. The video decoder 300 can apply a 3x3 convolution 1850 to the output of the set 1828. The video decoder 300 can apply a PReLU 1852 to the output of the 3x3 convolution 1850. The video decoder 300 can apply a 3x3 convolution 1854 to the output of the PReLU 1852. The video decoder 300 can crop 1858 of the output of the 3x3 convolution 1854 to generate filtered reconstructed UV samples of the image. The video decoder 300 can perform pixel rearrangement 1856 on the output of the 3x3 convolution 1854 and crop 1860 of the output of the pixel rearrangement 1856 to generate filtered reconstructed Y samples of the image.

[0163] One or more of the filter blocks 1830A to 1830N from the 1828 series can be used. Figure 16 The backbone architecture. In some examples, Figure 18 The example parameters and Figure 16 A group may include one or more of the following: d1=192, d2=32, d3=16, d4=16, d5=16, C=64, C1=160, C 21 =32、C 22 =32、C 31 =64, N=16 and d6=48.

[0164] A multi-scale feature extraction backbone with two-component decomposition has been integrated into the unified model in EE. Additionally, the specifications from EE include two model versions: 1) a unified model that combines luma and chroma, such as... Figure 15 As shown; and 2) models that separate luminance and chromaticity, as shown Figure 17 and Figure 18 As shown.

[0165] The unified filter (UF) architecture adopted by JVET (such as Figures 15 to 18 The figure shown represents a relatively high-performance level for video decoding applications, operating at over 400kMAC / pixel for decoding YCbCr 4:2:0 data. To run in-loop filters with video codecs on currently state-of-the-art computationally and / or power-constrained devices, the complexity of NN-based tools is limited to the level below 20kMAC / pixel. Therefore, Figures 15 to 18 The UF architecture cannot be implemented on such computing and / or power-constrained devices.

[0166] To achieve the desired level of complexity (e.g., <20kMAC / pixel) and retain most of the decoding performance provided by the UF architecture, according to the techniques of this disclosure, positive quasi-multivariate (CP) decomposition of separable convolutions is applied to the UF architecture to replace 3x3 convolutions in the backbone network and / or other blocks. The techniques of this disclosure also remove multi-scale feature extraction branches from residual blocks, remove activation layers (e.g., input feature extraction) from head blocks, and employ other parameter tuning.

[0167] Figure 19 This is a block diagram illustrating an example low-complexity in-loop filter network architecture according to one or more aspects of this disclosure. For example, Figure 19 The architecture can be used to implement low-complexity in-loop filters for joint YCbCr 4:2:0 processing. Figure 19 The example also applies to the separate component Y,CbCr processing. Figure 19 An example could be a uniform filter. A uniform filter can include a filter with at least one quality input (such as a QP input), making the filter usable for processing different quality levels. Figure 19 It can be seen that, Figure 19 The example filter is one with QP SLICE 1902 Input and QP BASE A unified filter with 1904 inputs. A unified NN-based filter (such as...) Figure 19 The unified NN-based filter can be trained using a variety of image and / or video data at different quality levels.

[0168] Figure 19 The architecture includes a header block 1980, a transition block 1922, a set 1928 of one or more backbone blocks 1930, and a tail block 1986. The inputs to the header block 1980 include IPB 1900 and QP. SLICE 1902, QP BASE 1904, BS 1906, PRED 1908 and REC 1909.

[0169] Head block 1980 includes blocks preceding fusion block 1916. Figure 19The header block 1980 comprises 3x3 convolutions 1910A through 1910F applied to the corresponding input. Although shown as 3x3 convolutions, in some examples, any or any combination of 3x3 convolutions 1910A through 1910F may be replaced by CP decomposition, as discussed further herein. The header block 1980 is shown with optional activation units 1912A through 1912F. In some examples, there are no activation units in the header block 1980. In some examples, one or more of the activation units 1912A through 1912F are rectified linear units (ReLU). In some examples, the header block 1980 may include one or more multi-scale branches (not shown for simplicity), such as... Figure 16 Multiscale branching 1620.

[0170] Fusion block 1916 can use 1x1 convolution 1918 and PReLU 1920 to fuse feature maps. Transformation block 1922 can include a block of filter 1980 that appears after fusion block 1916 and before any backbone or filter block (e.g., before filter blocks 1930A to 1930N). Transformation block 1922 uses 3x3 convolution 1924 and PReLU 1926 to process the fused data. In some examples, 3x3 convolution 1924 can be replaced by CP decomposition, as discussed further herein. Set 1928 can include filter blocks 1930A to 1930N (which may also be referred to as backbone blocks). In some examples, filter blocks 1930A to 1930N can provide Figure 19 The filtering function of the filter. In some examples, filter blocks 1930A to 1930N may be located after transition block 1922 and before tail block 1986. In some examples, one or more of filter blocks 1930A to 1930N may include CP decomposition, as discussed further herein. In some examples, set 1928 (e.g., filter blocks 1930A to 1930N) does not include multi-scale branches, such as... Figure 16 Multi-scale branches 1620. Removing or omitting multi-scale branches can reduce the complexity of NN-based filters.

[0171] Tail block 1986 may be included after all backbone blocks or filter blocks (e.g., after filter blocks 1930A to 1930N). Figure 19 The filter block 1986 may include a 3x3 convolution 1950, a PReLU 1952, a 3x3 convolution 1954, a cropping unit 1958, a pixel rearrangement unit 1956, and a cropping unit 1960. In some examples, one or more of the 3x3 convolution 1950 or 3x3 convolution 1954 may be replaced by CP decomposition, as discussed further herein.

[0172] This article is aimed at Figure 19The parameters illustrated in the examples are merely examples of possible choices. The performance of such architectures can vary with different configurations. Figure 19 In the examples, the following elements are described: activation in the head block; multi-scale branching; and CP decomposition. It should be noted that all, any, or any combination of such elements may be utilized according to the techniques of this disclosure.

[0173] Figure 19 Examples of example architectures and parameters are shown, such as the number of inputs, features, channels, and the number of residual blocks. In some examples, Figure 19 The example parameters include one or more of the following: d1=12, d2=8, d3=4, d4=2, d5=2, C=24, N=11, and d6=24. For the illustrated configuration, the complexity is approximately 19.55kMAC.

[0174] For example, regarding the activations in header block 1980, in some examples, activation units 1912A to 1912F are removed for the input group; these could be PReLUs in the current UF architecture. Removing PReLUs reduces the complexity of the NN-based filter. For example, header block 1980 may not include a corresponding PReLU for at least one input to header block 1980. In some examples, any or any combination of activation units 1912A to 1912E may be removed from header block 1980. For example, header block 1980 may not include a corresponding activation layer for at least one input to header block 1980. In some examples, one or more activation units 1912A to 1912F may be Modified Linear Units (ReLUs). Replacing PReLUs with ReLUs reduces the complexity of the NN-based filter. For example, one or more PReLUs in the current UF architecture may be replaced by ReLUs. In some examples, activation units 1912A through 1912F of the head block 1980 are removed to allow only linear operations to be performed before the fusion operation. For example, the output of one or more of the 3x3 convolutions 1910A through 1910F is fed into a 1x1 convolution 1918 without passing through any activation layer. Removing activation layers reduces the complexity of the NN-based filter.

[0175] For example, regarding multi-scale branching, in some examples, the multi-scale branching of the current UF architecture (e.g., as...) Figure 16 The multi-scale branches are removed from the backbone residual blocks (e.g., filter blocks 1930A to 1930N) so that parallel convolutions are not performed on the backbone blocks. Removal or omission of multi-scale branches can reduce the complexity of NN-based filters. In some examples, multi-scale branches are removed from the backbone blocks (e.g., within the backbone blocks). Figure 16The multi-scale branch (as shown) is moved to the head block 1980 to facilitate multi-scale feature extraction at a relatively low computational cost (e.g., lower computational cost than if the multi-scale branch were implemented in the backbone block). For example, the head block 1980 may include at least one multi-scale branch similar to multi-scale branch 1620. In some examples, the multi-scale branch may be applied to REC 1909, for example, before the 3x3 convolution 1910F.

[0176] For example, regarding CP decomposition, CP decompensation can be used to decompose multidimensional convolutions, for example, by using: 1) a 1x1 convolution and 2) multiple separate convolutions (e.g., 1x3 and 3x1 convolutions) in the horizontal and vertical dimensions, followed by 3) a 1x1 convolution combining extracted features from previous layers. In some examples, the first 1x1 convolution of the CP decomposition may be combined with the previous layer, thus skipping the first 1x1 convolution. For example, the CP decomposition may include at least one 1x1 convolution and separable 3x1 convolutions in the horizontal and vertical dimensions. For example, separate convolutions in the horizontal and vertical dimensions may include convolutions in the horizontal dimension and convolutions in the vertical dimension. Convolutions in the horizontal dimension may include one of a 1x3 convolution or a 3x1 convolution. Convolutions in the vertical dimension may include the other of a 1x3 convolution or a 3x1 convolution. In some examples, at least one 1x1 convolution includes a first 1x1 convolution preceding convolutions in the horizontal and vertical dimensions, and a second 1x1 convolution following convolutions in the horizontal and vertical dimensions. In some examples, the first 1x1 convolution may be omitted if there is no channel expansion or contraction. In some examples, if the convolution kernel size is NxM, the spatial decomposition of the kernel can be approximated by Nx1 convolutions followed by 1xM convolutions.

[0177] In some examples, the video encoder 200 or video decoder 300 may employ separable decompositions (e.g., CP decompositions) in set 1928 and / or tail block 1986. In some examples, the video encoder 200 or video decoder 300 may apply separable decompositions (e.g., CP decompositions) to all blocks (e.g., head block 1980, transition block 1922, set 1928, and tail block 1986). For example, the CP decomposition may be a specific type of separable decomposition and may use multiple convolutions to approximate higher-order multidimensional convolutions. For example, the CP decomposition may approximate a 3x3 convolution by using a 1x1 convolution, a 1x3 convolution, a 3x1 convolution, and another 1x1 convolution.

[0178] Figure 20 This is a block diagram illustrating an example backbone block according to one or more aspects of this disclosure. Backbone block 2000 may be... Figure 19Examples of any of the filter blocks 1930A to 1930N in set 1928. The backbone block 2000 differs from the UF backbone block 1600 in that backbone block 2000 does not include multi-scale branches like those in backbone block 1600, and the multidimensional convolutions replaced by convolutions 1610, 1612, and 1614 are replaced by CP decomposition in backbone block 2000. The CP decomposition in backbone block 2000 includes 1x1 convolution 2006, 1x3 convolution 2008, 3x1 convolution 2010, and 1x1 convolution 2012. In some examples, Figure 20 The parameters include C=24, C1=72 and C21=24.

[0179] Figure 21 This is a block diagram illustrating an example CP decomposition according to one or more aspects of this disclosure. CP decomposition 2100 may include 1x1 convolution 2106, 1x3 convolution 2108, 3x1 convolution 2110, and 1x1 convolution 2112. It should be understood that CP decomposition 2100 may replace any or any combination of the 3x3 convolutions 1910A to 1910F of the head block 1980, the 3x3 convolution 1924 of the transition block 1922, and the 3x3 convolution 1950 and / or 3x3 convolution 1954 of the tail block 1986. In some examples, Figure 21 The parameters include C=24, C1=72 and C21=24.

[0180] The techniques described herein are applicable to in-loop filters with a wide range of complexities by selecting different parameter configurations. The techniques disclosed herein reduce computational complexity and memory bandwidth requirements while providing competitive performance, such as in devices with complexity and / or memory bandwidth constraints. The examples described in this document relate to neural network-based in-loop filtering; however, such examples are applicable to any neural network-based video decoding tool that consumes input data with certain statistical properties, such as static content or sparse representation.

[0181] Figure 22 This is a flowchart illustrating an example of an in-loop filtering technique based on an neural network according to one or more aspects of this disclosure. The video decoder 300 may receive image data (2200) for reconstructing video data. For example, the video encoder 200 may send encoded video data of images in a bitstream to the video decoder 300, and the video decoder 300 may receive encoded video data of images in a bitstream.

[0182] The video decoder 300 can reconstruct images (2202) from the video data. For example, the video decoder 300 can decode encoded video data to reconstruct images of the video data.

[0183] The video decoder 300 can apply neural network (NN)-based filters to the reconstructed images from video data, wherein the NN-based filters include a unified filter comprising a head block, a transform block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, transform block, one or more backbone blocks, or tail block comprises a positive quasi-multivariate (CP) decomposition (2204) utilizing separable convolutions. For example, the video decoder 300 can use NN-based filters (such as...) Figure 19 The reconstructed image is filtered using a neural network-based filter. The neural network-based filter may include a head block 1980, a transition block 1922, one or more backbone blocks (e.g., set 1928), and a tail block 1986. At least one of the head block 1980, transition block 1922, one or more backbone blocks, or tail block 1986 includes CP decomposition utilizing separable convolutions. The one or more backbone blocks may be residual blocks.

[0184] For example, head block 1980 may include a block of NN-based filters 19 preceding the fusion block of the fused feature map. Transition block 1922 may include a block of NN-based filters that appears after the fused feature map (e.g., after the fusion block and therefore after the head block) and before any backbone or filter blocks. Backbone blocks (e.g., filter blocks 1930A to 1930N) provide filtering functionality and include a block of NN-based filters between transition block 1922 and tail block 1986. Tail block 1986 may include a block of NN-based filters that appears after all backbone blocks.

[0185] In some examples, the CP decomposition includes at least one 1x1 convolution, a horizontal convolution, and a vertical convolution. In some examples, the horizontal convolution includes either a 1x3 convolution or a 3x1 convolution, and the vertical convolution includes the other of a 1x3 convolution or a 3x1 convolution. For example, the CP decomposition may include at least one 1x1 convolution, a 1x3 convolution, and a 3x1 convolution. In some examples, at least one 1x1 convolution includes a first 1x1 convolution occurring before the horizontal and vertical convolutions, and a second 1x1 convolution occurring after the horizontal and vertical convolutions. It should be noted that the order of the convolutions can be changed and two linear layers can be combined. In some examples, the second 1x1 convolution combines extracted features from the previous layer.

[0186] In some examples, one or more backbone blocks and tail blocks 1986 each include a corresponding CP decomposition using separable convolutions. In other words, one or more backbone blocks include a CP decomposition using separable convolutions, and the tail block includes a CP decomposition using separable convolutions. In some examples, head block 1980, transition block 1922, one or more backbone blocks 1930, and tail block 1986 each include a corresponding CP decomposition using separable convolutions. In some examples, one or more backbone blocks 1930 do not include multi-scale branches. For example, multi-scale branches can be implemented using methods such as... Figure 16 The parallel convolution shown is used to perform multi-scale feature extraction.

[0187] In some examples, head block 1980 does not include a corresponding parameter-corrected linear unit activation (PReLU) to at least one input of head block 1980. In some examples, head block 1980 does not include a corresponding activation layer to at least one input of the head block. In some examples, head block 1980 does not include any activation layer. In some examples, head block 1980 includes at least one corresponding modified linear unit activation (ReLU) to at least one input of head block 1980. In some examples, head block 1980 includes at least one multi-scale branch. In some examples, one or more backbone blocks of the NN-based filter each include 24 channels. For example, C=24.

[0188] Figure 23 This is a block diagram illustrating an example video encoder 200 that can perform the techniques of this disclosure. Figure 23 This disclosure is provided for illustrative purposes and should not be construed as a limitation on the techniques extensively illustrated and described herein. For illustrative purposes, this disclosure describes the video encoder 200 in accordance with the techniques of VVC and HEVC. However, the techniques of this disclosure can be performed by video encoding devices configured for other video decoding standards and video decoding formats, such as AV1 and subsequent formats of AV1 video decoding.

[0189] exist Figure 23In the example, the video encoder 200 includes a video data memory 230, a mode selection unit 202, a residual generation unit 204, a transform processing unit 206, a quantization unit 208, an inverse quantization unit 210, an inverse transform processing unit 212, a reconstruction unit 214, a filter unit 216, a decoded picture buffer (DPB) 218, and an entropy encoding unit 220. Any or all of the video data memory 230, mode selection unit 202, residual generation unit 204, transform processing unit 206, quantization unit 208, inverse quantization unit 210, inverse transform processing unit 212, reconstruction unit 214, filter unit 216, DPB 218, and entropy encoding unit 220 can be implemented in one or more processors or in processing circuitry. For example, the units of the video encoder 200 can be implemented as one or more circuit or logic elements as part of hardware circuitry, or as part of a processor, ASIC, or FPGA. Furthermore, the video encoder 200 may include additional or alternative processors or processing circuitry to perform these and other functions.

[0190] Video data storage 230 can store video data to be encoded by components of video encoder 200. Video encoder 200 can receive data from, for example, video source 104 (…). Figure 1 The video data memory 230 receives video data stored in the video data memory 230. The DPB 218 can act as a reference picture memory, storing reference video data for use when the video encoder 200 predicts subsequent video data. The video data memory 230 and DPB 218 can be formed from any of a variety of memory devices, such as dynamic random access memory (DRAM) (including synchronous DRAM (SDRAM)), magnetoresistive RAM (MRAM), resistive RAM (RRAM), or other types of memory devices. The video data memory 230 and DPB 218 can be provided by the same memory device or separate memory devices. In various examples, the video data memory 230 can be on-chip (as illustrated) with other components of the video encoder 200, or off-chip relative to those components.

[0191] In this disclosure, references to video data memory 230 should not be construed as limited to memory inside video encoder 200 (unless specifically described) or memory outside video encoder 200 (unless specifically described). Rather, references to video data memory 230 should be understood as a reference memory that stores video data received by video encoder 200 for encoding (e.g., video data for the current block to be encoded). Figure 1 The memory 106 can also provide temporary storage for the outputs from various units of the video encoder 200.

[0192] Examples Figure 23Various units help understand the operations performed by the video encoder 200. Units can be implemented as fixed-function circuits, programmable circuits, or combinations thereof. Fixed-function circuits are circuits that provide specific functionality and are pre-defined for the operations that can be performed. Programmable circuits are circuits that can be programmed to perform various tasks and provide flexible functionality for the operations that can be performed. For example, a programmable circuit can execute software or firmware that causes the programmable circuit to operate in a manner defined by the instructions of the software or firmware. Fixed-function circuits can execute software instructions (e.g., to receive or output parameters), but the type of operation performed by a fixed-function circuit is generally immutable. In some examples, one or more units in the unit may be different circuit blocks (fixed-function or programmable), and in some examples, one or more units in the unit may be integrated circuits.

[0193] The video encoder 200 may include an arithmetic logic unit (ALU), an essential function unit (EFU), digital circuitry, analog circuitry, and / or a programmable core, all formed by programmable circuitry. In an example where the operation of the video encoder 200 is performed using software executed by programmable circuitry, memory 106 ( Figure 1 The video encoder 200 may store instructions (e.g., target code) of the software received and executed by the video encoder 200, or another memory (not shown) within the video encoder 200 may store such instructions.

[0194] The video data storage unit 230 is configured to store received video data. The video encoder 200 can retrieve images of the video data from the video data storage unit 230 and provide the video data to the residual generation unit 204 and the mode selection unit 202. The video data in the video data storage unit 230 can be raw video data to be encoded.

[0195] The mode selection unit 202 includes a motion estimation unit 222, a motion compensation unit 224, and an intra-frame prediction unit 226. The mode selection unit 202 may include additional functional units for performing video prediction based on other prediction modes. As an example, the mode selection unit 202 may include a palette unit, an intra-frame block copying unit (which may be part of the motion estimation unit 222 and / or the motion compensation unit 224), an affine unit, a linear model (LM) unit, etc.

[0196] Mode selection unit 202 typically coordinates multiple coding channels to test combinations of coding parameters and the resulting rate-distortion values ​​for such combinations. Coding parameters may include the CTU-CU partitioning, the prediction mode for the CU, the transformation type of the residual data for the CU, the quantization parameters of the residual data for the CU, etc. Mode selection unit 202 can ultimately select a combination of coding parameters that has a better rate-distortion value compared to other tested combinations.

[0197] The video encoder 200 can divide images retrieved from the video data storage 230 into a series of CTUs, and encapsulate one or more CTUs within slices. The mode selection unit 202 can divide the image's CTUs according to the tree structure described above (such as an MTT structure, a QTBT structure, a superblock structure, or the quadtree structure described above). As described above, the video encoder 200 can form one or more CUs by dividing CTUs according to a tree structure. Such CUs are also commonly referred to as "video blocks" or "blocks".

[0198] Typically, mode selection unit 202 also controls its components (e.g., motion estimation unit 222, motion compensation unit 224, and intra-prediction unit 226) to generate prediction blocks for the current block (e.g., the current CU, or, in HEVC, the overlapping portion of PU and TU). To perform inter-frame prediction for the current block, motion estimation unit 222 may perform a motion search to identify one or more closely matching reference blocks in one or more reference pictures (e.g., one or more previously decoded pictures stored in DPB 218). Specifically, motion estimation unit 222 may calculate values ​​representing the similarity between a potential reference block and the current block, for example, based on sum of absolute differences (SAD), sum of squared differences (SSD), mean absolute difference (MAD), mean squared difference (MSD), etc. Motion estimation unit 222 may typically perform these calculations using sample-by-sample differences between the current block and the reference blocks under consideration. Motion estimation unit 222 may identify reference blocks with the lowest values ​​produced by these calculations to indicate the reference block that best matches the current block.

[0199] Motion estimation unit 222 can generate one or more motion vectors (MVs) that define the location of a reference block in a reference image relative to the location of the current block in the current image. Motion estimation unit 222 can then provide the motion vectors to motion compensation unit 224. For example, for unidirectional inter-frame prediction, motion estimation unit 222 can provide a single motion vector, while for bidirectional inter-frame prediction, motion estimation unit 222 can provide two motion vectors. Motion compensation unit 224 can then use the motion vectors to generate a prediction block. For example, motion compensation unit 224 can use the motion vectors to retrieve data for the reference block. As another example, where the motion vectors have fractional sample precision, motion compensation unit 224 can interpolate the values ​​of the prediction block according to one or more interpolation filters. Furthermore, for bidirectional inter-frame prediction, motion compensation unit 224 can retrieve data for two reference blocks identified by corresponding motion vectors and combine the retrieved data, for example, by per-sample averaging or weighted averaging.

[0200] When operating according to the AV1 video decoding format, the motion estimation unit 222 and the motion compensation unit 224 can be configured to encode the decoded blocks of video data (e.g., both luma and chroma decoded blocks) using translational motion compensation, affine motion compensation, overlap block motion compensation (OBMC), and / or composite inter-intra-frame prediction.

[0201] As another example, for intra-prediction or intra-prediction decoding, intra-prediction unit 226 may generate a prediction block from samples adjacent to the current block. For example, for directional mode, intra-prediction unit 226 may typically mathematically combine the values ​​of adjacent samples and fill these calculated values ​​across the current block in a defined direction to produce a prediction block. As another example, for DC mode, intra-prediction unit 226 may calculate the average of the adjacent samples of the current block and generate a prediction block to include the resulting average for each sample of the prediction block.

[0202] When operating according to the AV1 video decoding format, the intra-frame prediction unit 226 can be configured to encode decoded blocks of video data (e.g., both luma and chroma decoded blocks) using directional intra-frame prediction, non-directional intra-frame prediction, recursive filter intra-frame prediction, luma-chroma (CFL) prediction, intra-block copying (IBC), and / or palette modes. The mode selection unit 202 may include additional functional units for performing video prediction based on other prediction modes.

[0203] Mode selection unit 202 provides a prediction block to residual generation unit 204. Residual generation unit 204 receives an uncoded raw version of the current block from video data memory 230 and a prediction block from mode selection unit 202. Residual generation unit 204 calculates the sample-by-sample difference between the current block and the prediction block. The resulting sample-by-sample difference defines the residual block for the current block. In some examples, residual generation unit 204 may also determine the differences between sample values ​​in the residual block to generate the residual block using residual differential pulse decoding modulation (RDPCM). In some examples, residual generation unit 204 may be formed using one or more subtractor circuits performing binary subtraction.

[0204] In the example where mode selection unit 202 divides a CU into PUs, each PU can be associated with a luma prediction unit and a corresponding chroma prediction unit. Video encoder 200 and video decoder 300 can support PUs of various sizes. As noted above, the size of a CU can refer to the size of the luma decoding block of the CU, while the size of a PU can refer to the size of the luma prediction unit of the PU. Assuming a particular CU size is 2Nx2N, video encoder 200 can support PU sizes of 2Nx2N or NxN for intra-frame prediction, and symmetric PU sizes of 2Nx2N, 2NxN, Nx2N, NxN, or similar for inter-frame prediction. Video encoder 200 and video decoder 300 can also support asymmetric partitioning for PU sizes of 2NxnU, 2NxnD, nLx2N, and nRx2N for inter-frame prediction.

[0205] In an example where mode selection unit 202 does not further divide the CU into PUs, each CU can be associated with a luminance decoding block and a corresponding chrominance decoding block. As mentioned above, the size of the CU can refer to the size of the luminance decoding block of the CU. The video encoder 200 and the video decoder 300 can support CU sizes of 2Nx2N, 2NxN, or Nx2N.

[0206] For other video decoding techniques, such as intra-block copy mode decoding, affine mode decoding, and linear model (LM) mode decoding, as some examples, mode selection unit 202 generates a prediction block for the current block being encoded via a corresponding unit associated with the decoding technique. In some examples (such as palette mode decoding), mode selection unit 202 may not generate a prediction block, but instead generate syntax elements indicating how the block is reconstructed based on a selected palette. In such modes, mode selection unit 202 may provide these syntax elements to entropy coding unit 220 for encoding.

[0207] As described above, the residual generation unit 204 receives video data for the current block and the corresponding prediction block. Then, the residual generation unit 204 generates a residual block for the current block. To generate the residual block, the residual generation unit 204 calculates the sample-by-sample difference between the prediction block and the current block.

[0208] Transform processing unit 206 applies one or more transformations to the residual block to generate a block of transform coefficients (referred to herein as a "transform coefficient block"). Transform processing unit 206 may apply various transformations to the residual block to form the transform coefficient block. For example, transform processing unit 206 may apply a discrete cosine transform (DCT), direction transform, Karhunen-Loeve transform (KLT), or conceptually similar transformations to the residual block. In some examples, transform processing unit 206 may perform multiple transformations on the residual block, such as primary and secondary transformations (e.g., rotation transformations). In some examples, transform processing unit 206 does not apply any transformations to the residual block.

[0209] When operating according to AV1, transform processing unit 206 may apply one or more transforms to the residual block to generate a block of transform coefficients (referred to herein as a "transform coefficient block"). Transform processing unit 206 may apply various transforms to the residual block to form the transform coefficient block. For example, transform processing unit 206 may apply a combination of horizontal / vertical transforms, which may include the Discrete Cosine Transform (DCT), the Asymmetric Discrete Sine Transform (ADST), the Reversed ADST (e.g., ADST in reverse order), and the Identity Transform (IDTX). When using the Identity Transform, the transform is skipped in either the vertical or horizontal direction. In some examples, transform processing may be skipped entirely.

[0210] Quantization unit 208 quantizes the transform coefficients in a transform coefficient block to produce a quantized transform coefficient block. Quantization unit 208 quantizes the transform coefficients of the transform coefficient block according to the quantization parameter (QP) value associated with the current block. Video encoder 200 (e.g., via mode selection unit 202) can adjust the degree of quantization applied to the transform coefficient block associated with the current block by adjusting the QP value associated with the CU. Quantization may cause information loss, and therefore, the quantized transform coefficients may have lower accuracy compared to the original transform coefficients produced by transform processing unit 206.

[0211] The inverse quantization unit 210 and the inverse transform processing unit 212 can apply inverse quantization and inverse transform, respectively, to the quantized transform coefficient block to reconstruct the residual block based on the transform coefficient block. The reconstruction unit 214 can generate a reconstructed block corresponding to the current block (although potentially with some degree of distortion) based on the reconstructed residual block and the prediction block generated by the mode selection unit 202. For example, the reconstruction unit 214 can add samples of the reconstructed residual block to corresponding samples of the prediction block generated by the mode selection unit 202 to generate the reconstructed block.

[0212] Filter unit 216 may perform one or more filtering operations on the reconstructed block. For example, filter unit 216 may perform a deblocking operation to reduce block artifacts along the edges of the CU. In some examples, the operation of filter unit 216 may be skipped.

[0213] When operating according to AV1, filter unit 216 may perform one or more filtering operations on the reconstructed block. For example, filter unit 216 may perform a deblocking operation to reduce block artifacts along the edges of the CU. In other examples, filter unit 216 may apply a constrained direction enhancement filter (CDEF), which may be applied after deblocking, and may include the application of a non-separable, nonlinear, low-pass directional filter based on the estimated edge direction. Filter unit 216 may also include a loop recovery filter applied after CDEF, and may include a separable symmetric normalized Wiener filter or a dual-guided filter.

[0214] The video encoder 200 stores reconstructed blocks in the DPB 218. For example, in an example where the filter unit 216 is not operated, the reconstruction unit 214 may store the reconstructed blocks in the DPB 218. In an example where the filter unit 216 is operated, the filter unit 216 may store the filtered reconstructed blocks in the DPB 218. The motion estimation unit 222 and the motion compensation unit 224 may retrieve a reference image formed by the reconstructed (and potentially filtered) blocks from the DPB 218 to perform inter-frame prediction for blocks of subsequent encoded images. Additionally, the intra-frame prediction unit 226 may use the reconstructed blocks of the current image in the DPB 218 to perform intra-frame prediction for other blocks in the current image.

[0215] Typically, entropy coding unit 220 can entropy-encode syntax elements received from other functional components of video encoder 200. For example, entropy coding unit 220 can entropy-encode quantized transform coefficient blocks from quantization unit 208. As another example, entropy coding unit 220 can entropy-encode predictive syntax elements (e.g., motion information for inter-frame prediction or intra-frame mode information for intra-frame prediction) from mode selection unit 202. Entropy coding unit 220 can perform one or more entropy coding operations on syntax elements (another example of video data) to generate entropy-coded data. For example, entropy coding unit 220 can perform context-adaptive variable-length decoding (CAVLC), CABAC, variable-to-variable (V2V) length decoding, syntax-based context-adaptive binary arithmetic decoding (SBAC), probability interval partitioning entropy (PIPE) decoding, exponential Golomb coding, or another type of entropy coding operation on the data. In some examples, entropy coding unit 220 can operate in a bypass mode where syntax elements are not entropy-encoded.

[0216] The video encoder 200 can output a bitstream containing entropy-encoded syntax elements required to reconstruct slices or blocks of images. Specifically, the entropy coding unit 220 can output a bitstream.

[0217] According to AV1, entropy coding unit 220 can be configured as a symbol-to-symbol adaptive multi-symbol arithmetic decoder. The syntax elements in AV1 consist of an N-element alphabet, and the context (e.g., a probability model) consists of a set of N probabilities. Entropy coding unit 220 can store the probabilities as an n-bit (e.g., 15-bit) cumulative distribution function (CDF). Entropy coding unit 220 can perform recursive scaling to update the context using an update factor based on the alphabet size.

[0218] The operations described above are block-related. This description should be understood as operations applied to luma decoding blocks and / or chroma decoding blocks. As described above, in some examples, the luma decoding block and chroma decoding block are the luma and chroma components of the CU. In some examples, the luma decoding block and chroma decoding block are the luma and chroma components of the PU.

[0219] In some examples, it is not necessary to repeat the operations performed for the luma decoding block for the chroma decoding block. As an example, the operations for identifying the motion vector (MV) and reference image of the luma decoding block do not need to repeat the MV and reference image used to identify the chroma block. Instead, the MV used for the luma decoding block can be scaled to determine the MV used for the chroma block, and the reference image can be the same. As another example, the intra-frame prediction process can be the same for both the luma and chroma decoding blocks.

[0220] Video encoder 200 represents an example of a device configured to encode video data, the device including: one or more memories configured to store video data; and one or more processors configured to receive images of the video data, reconstruct images of the video data, and apply neural network (NN)-based filters to the reconstructed images of the video data, wherein the NN-based filters include a unified filter comprising a head block, a transform block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, transform block, one or more backbone blocks, or tail block comprises a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0221] Figure 24 This is a block diagram illustrating an example video decoder 300 that can perform the techniques of this disclosure. Figure 24This disclosure is provided for illustrative purposes and not for limiting the techniques extensively illustrated and described herein. For illustrative purposes, this disclosure describes the video decoder 300 based on VVC and HEVC techniques. However, the techniques of this disclosure can be implemented by video decoding devices configured for other video decoding standards.

[0222] exist Figure 24 In the example, the video decoder 300 includes a decoded picture buffer (CPB) memory 320, an entropy decoding unit 302, a prediction processing unit 304, an inverse quantization unit 306, an inverse transform processing unit 308, a reconstruction unit 310, a filter unit 312, and a DPB 314. Any or all of the CPB memory 320, entropy decoding unit 302, prediction processing unit 304, inverse quantization unit 306, inverse transform processing unit 308, reconstruction unit 310, filter unit 312, and DPB 314 can be implemented in one or more processors or in processing circuitry. For example, the units of the video decoder 300 can be implemented as one or more circuit or logic elements as part of hardware circuitry, or as part of a processor, ASIC, or FPGA. Furthermore, the video decoder 300 may include additional or alternative processors or processing circuitry to perform these and other functions.

[0223] The prediction processing unit 304 includes a motion compensation unit 316 and an intra-prediction unit 318. The prediction processing unit 304 may include additional units that perform predictions based on other prediction modes. As an example, the prediction processing unit 304 may include a palette unit, an intra-block copying unit (which may form part of the motion compensation unit 316), an affine unit, a linear model (LM) unit, etc. In other examples, the video decoder 300 may include more, fewer, or different functional components.

[0224] When operating according to AV1, motion compensation unit 316 can be configured to decode video data blocks (e.g., both luma and chroma blocks) using translational motion compensation, affine motion compensation, OBMC, and / or composite inter-intra-frame prediction, as described above. Intra-frame prediction unit 318 can be configured to decode video data blocks (e.g., both luma and chroma blocks) using directional intra-frame prediction, non-directional intra-frame prediction, recursive filter intra-frame prediction, CFL, IBC, and / or palette mode, as described above.

[0225] CPB memory 320 can store video data to be decoded by components of video decoder 300, such as encoded video bitstreams. For example, it can be retrieved from computer-readable medium 110 ( Figure 1The video data stored in the CPB memory 320 is obtained. The CPB memory 320 may include a CPB that stores encoded video data (e.g., syntax elements) from the encoded video bitstream. Furthermore, the CPB memory 320 may store video data other than the syntax elements of the decoded picture, such as temporary data representing the output from various units of the video decoder 300. The DPB 314 typically stores a decoded picture that the video decoder 300 may output, and / or uses the decoded picture as reference video data when decoding subsequent data or pictures of the encoded video bitstream. The CPB memory 320 and the DPB 314 may be formed from any of a variety of memory devices, such as DRAM (including SDRAM), MRAM, RRAM, or other types of memory devices. The CPB memory 320 and the DPB 314 may be provided by the same memory device or separate memory devices. In various examples, the CPB memory 320 may be on-chip with other components of the video decoder 300, or off-chip relative to those components.

[0226] Additionally or alternatively, in some examples, the video decoder 300 may be from the memory 120 ( Figure 1 The decoded video data can be retrieved from the memory. In other words, memory 120 can utilize CPB memory 320 to store data as discussed above. Similarly, when some or all of the functionality of video decoder 300 is implemented in software to be executed by the processing circuitry of video decoder 300, memory 120 can store instructions to be executed by video decoder 300.

[0227] Examples Figure 24 The various units shown help to understand the operations performed by the video decoder 300. These units can be implemented as fixed-function circuits, programmable circuits, or a combination thereof. Similar to... Figure 23 Fixed-function circuits are circuits that provide specific functionality and are predefined for the operations they can perform. Programmable circuits are circuits that can be programmed to perform various tasks and provide flexible functionality for the operations they can perform. For example, a programmable circuit can execute software or firmware that causes the programmable circuit to operate in a manner defined by the instructions of the software or firmware. Fixed-function circuits can execute software instructions (e.g., to receive or output parameters), but the type of operation performed by a fixed-function circuit is usually immutable. In some examples, one or more units in a cell can be different circuit blocks (fixed-function or programmable), and in some examples, one or more units in a cell can be integrated circuits.

[0228] The video decoder 300 may include an ALU, EFU, digital circuitry, analog circuitry, and / or a programmable core formed by programmable circuitry. In an example where the operation of the video decoder 300 is performed by software executed on programmable circuitry, on-chip or off-chip memory may store instructions (e.g., object code) of the software received and executed by the video decoder 300.

[0229] The entropy decoding unit 302 can receive encoded video data from the CPB and perform entropy decoding on the video data to reproduce the syntax elements. The prediction processing unit 304, the inverse quantization unit 306, the inverse transform processing unit 308, the reconstruction unit 310, and the filter unit 312 can generate decoded video data based on the syntax elements extracted from the bitstream.

[0230] Typically, the video decoder 300 reconstructs the image block by block. The video decoder 300 can perform the reconstruction operation on each block individually (where the block currently being reconstructed (i.e., decoded) can be referred to as the "current block").

[0231] Entropy decoding unit 302 can entropy decode the syntax elements of the quantized transform coefficients defining the quantized transform coefficient block, as well as transform information (such as quantization parameters (QP) and / or transform mode indications). Inverse quantization unit 306 can use the QP associated with the quantized transform coefficient block to determine the degree of quantization, and similarly, determine the degree of inverse quantization to be applied by inverse quantization unit 306. Inverse quantization unit 306 can, for example, perform a bit-by-bit left shift operation to inverse quantize the quantized transform coefficients. Inverse quantization unit 306 can thereby form a transform coefficient block including the transform coefficients.

[0232] After the inverse quantization unit 306 forms the transform coefficient block, the inverse transform processing unit 308 may apply one or more inverse transforms to the transform coefficient block to generate a residual block associated with the current block. For example, the inverse transform processing unit 308 may apply an inverse DCT, an inverse integer transform, an inverse Karhunen-Loeve transform (KLT), an inverse rotation transform, an inverse direction transform, or another inverse transform to the transform coefficient block.

[0233] Furthermore, the prediction processing unit 304 generates a prediction block based on the prediction information syntax elements entropy decoded by the entropy decoding unit 302. For example, if the prediction information syntax elements indicate that the current block is an inter-frame prediction, the motion compensation unit 316 can generate the prediction block. In this case, the prediction information syntax elements may indicate the reference picture from which the reference block is to be retrieved in the DPB 314, and a motion vector identifying the position of the reference block in the reference picture relative to the position of the current block in the current picture. The motion compensation unit 316 can generally be configured according to the parameters of the motion compensation unit 224 ( Figure 23 The method described is essentially the same as the method used to perform the inter-frame prediction process.

[0234] As another example, when the prediction information syntax element indicates that the current block is intra-predictive, the intra-predictive unit 318 can generate a prediction block according to the intra-predictive mode indicated by the prediction information syntax element. Similarly, the intra-predictive unit 318 can generally be associated with the intra-predictive unit 226 ( Figure 23 The intra-prediction process is performed in a manner substantially similar to that described above. The intra-prediction unit 318 can retrieve data of neighboring samples of the current block from the DPB 314.

[0235] Reconstruction unit 310 can use prediction blocks and residual blocks to reconstruct the current block. For example, reconstruction unit 310 can add samples from the residual block to the corresponding samples from the prediction block to reconstruct the current block.

[0236] Filter unit 312 can perform one or more filtering operations on the reconstructed block. For example, filter unit 312 can perform a deblocking operation to reduce block artifacts along the edges of the reconstructed block. The operation of filter unit 312 is not necessarily performed in all examples.

[0237] The video decoder 300 can store reconstructed blocks in the DPB 314. For example, in an example where the operation of the filter unit 312 is not performed, the reconstruction unit 310 can store the reconstructed blocks in the DPB 314. In an example where the operation of the filter unit 312 is performed, the filter unit 312 can store the filtered reconstructed blocks in the DPB 314. As discussed above, the DPB 314 can provide reference information (such as samples of the current image for intra-frame prediction and samples of previously decoded images for subsequent motion compensation) to the prediction processing unit 304. Furthermore, the video decoder 300 can output decoded images (e.g., decoded video) from the DPB 314 for use in applications such as... Figure 1 The subsequent presentation on display devices such as display device 118.

[0238] In this manner, video decoder 300 represents an example of a video decoding device, which includes: one or more memories configured to store video data; and one or more processors configured to receive images of the video data, reconstruct images of the video data, and apply neural network (NN)-based filters to the reconstructed images of the video data, wherein the NN-based filters include a unified filter comprising a head block, a transform block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, transform block, one or more backbone blocks, or tail block comprises a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0239] Figure 25This is a flowchart illustrating an example method for encoding the current block according to the technology of this disclosure. The current block may be or may include the current CU. Although this relates to video encoder 200 ( Figure 1 and Figure 23 This is a description, but it should be understood that other devices can be configured to perform the same actions. Figure 25 Similar to the method.

[0240] In this example, the video encoder 200 initially predicts the current block (400). For example, the video encoder 200 may form a prediction block for the current block. The video encoder 200 may then compute a residual block for the current block (402). To compute the residual block, the video encoder 200 may compute the difference between the unencoded original block for the current block and the prediction block. The video encoder 200 may then transform the residual block and quantize the transform coefficients of the residual block (404). Next, the video encoder 200 may scan the quantized transform coefficients of the residual block (406). During or after the scan, the video encoder 200 may entropy encode the transform coefficients (408). For example, the video encoder 200 may use CAVLC or CABAC to encode the transform coefficients. The video encoder 200 may then output the entropy-encoded data of the block (410).

[0241] Figure 26 This is a flowchart illustrating an example method for decoding a current block of video data according to the technology of this disclosure. The current block may be or may include the current CU. Although this relates to video decoder 300 ( Figure 1 and Figure 24 This is a description, but it should be understood that other devices can be configured to perform the same actions. Figure 26 Similar to the method.

[0242] The video decoder 300 can receive entropy-coded data for the current block, such as entropy-coded prediction information and entropy-coded data for the transform coefficients of the residual block corresponding to the current block (420). The video decoder 300 can entropy decode the entropy-coded data to determine the prediction information for the current block and reproduce the transform coefficients of the residual block (422). The video decoder 300 can predict the current block, for example, using an intra-frame prediction mode or an inter-frame prediction mode indicated by the prediction information for the current block (424), to compute a prediction block for the current block. The video decoder 300 can then perform an inverse scan on the reproduced transform coefficients (426) to create a block of quantized transform coefficients. The video decoder 300 can then inverse quantize the transform coefficients and apply the inverse transform to the transform coefficients to produce a residual block (428). The video decoder 300 can finally decode the current block by combining the prediction block and the residual block (430).

[0243] The following numbered clauses illustrate one or more aspects of the devices and technologies described in this disclosure.

[0244] Clause 1A. A method for decoding video data, the method comprising:

[0245] The method includes receiving images from video data; reconstructing the images from the video data; and performing a neural network (NN)-based filtering process on the reconstructed images from the video data, wherein the NN-based filtering process includes performing positive multivariate (CP) decomposition using separable convolutions.

[0246] Clause 2A. The method according to Clause 1A, the method further comprising avoiding multi-scale feature extraction in residual blocks of the images in the video data.

[0247] Clause 3A. The method described in accordance with Clause 1A or Clause 2A, wherein the method further removes one or more activation layers in the header block.

[0248] Clause 4A. The method according to any one of Clauses 1A to 3A, wherein the input group does not include activating PReLU.

[0249] Clause 5A. The method described in Clause 4A, wherein the input group includes ReLU but excludes action PReLU.

[0250] Clause 6A. The method according to any one of Clauses 1A to 4A, wherein the input group does not include any activation for the header block.

[0251] Clause 7A. The method according to any one of Clauses 1A to 6A, the method further comprising avoiding the application of multi-scale branching to the backbone residual block.

[0252] Clause 8A. The method according to Clause 7A further includes applying the multi-scale branch to at least one head block.

[0253] Clause 9A. The method according to any one of Clauses 1A to 8A, wherein the CP decomposition comprises: at least one 1x1 convolution; and 3x1 separable convolutions in the horizontal and vertical dimensions.

[0254] Clause 10A. The method according to Clause 9A, wherein the at least one 1x1 convolution comprises: a first 1x1 convolution occurring before the 3x1 separable convolution; and a second 1x1 convolution occurring after the 3x1 separable convolution.

[0255] Clause 11A. The method according to Clause 10A, wherein the second 1x1 convolution combination is derived from the extracted features of the previous layer.

[0256] Clause 12A. The method according to any one of Clauses 1A to 11A, wherein performing the CP decomposition using separable convolution includes performing the CP decomposition on the backbone block and the tail block.

[0257] Clause 13A. The method according to any one of Clauses 1A to 12A, wherein performing the CP decomposition using separable convolutions comprises performing the CP decomposition on all blocks.

[0258] Clause 14A. The method according to any one of Clauses 1A to 13A, wherein decoding includes decoding.

[0259] Clause 15A. The method according to any one of Clauses 1A to 14A, wherein decoding includes encoding.

[0260] Clause 16A. An apparatus for decoding video data, the apparatus comprising one or more components for performing the method according to any one of Clauses 1A to 15A.

[0261] Clause 17A. The device pursuant to Clause 16A, wherein the one or more components include one or more processors implemented in a circuit.

[0262] Clause 18A. The device according to any one of Clauses 16A and 17A, the device further comprising: a memory for storing the video data.

[0263] Clause 19A. The device according to any one of Clauses 16A to 18A, the device further includes a display configured to display decoded video data.

[0264] Clause 20A. The device pursuant to any one of Clauses 16A to 19A, wherein the device includes one or more of a camera, computer, mobile device, broadcast receiver device or set-top box.

[0265] Clause 21A. The device pursuant to any one of Clauses 16A to 20A, wherein the device includes a video decoder.

[0266] Clause 22A. The device pursuant to any one of Clauses 16A to 21A, wherein the device includes a video encoder.

[0267] Clause 23A. A computer-readable storage medium having instructions stored thereon, which, when executed, cause one or more processors to perform the method according to any one of Clauses 1A to 15A.

[0268] Clause 1B. A method for encoding or decoding video data, the method comprising: receiving data for reconstructing images of the video data; reconstructing images of the video data; and applying a neural network (NN)-based filter to the reconstructed images of the video data, wherein the NN-based filter includes a unified filter comprising a head block, a transition block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, the transition block, one or more backbone blocks, or the tail block comprises a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0269] Clause 2B. The method according to Clause 1B, wherein the CP decomposition comprises: at least one 1x1 convolution; convolution in the horizontal dimension; and convolution in the vertical dimension.

[0270] Clause 3B. The method according to Clause 2B, wherein the convolution in the horizontal dimension comprises one of a 1x3 convolution or a 3x1 convolution, and wherein the convolution in the vertical dimension comprises the other of the 1x3 convolution or the 3x1 convolution.

[0271] Clause 4B. The method according to Clause 3B, wherein the at least one 1x1 convolution comprises: a first 1x1 convolution occurring before the convolution in the horizontal dimension and the convolution in the vertical dimension; and a second 1x1 convolution occurring after the convolution in the horizontal dimension and the convolution in the vertical dimension.

[0272] Clause 5B. The method according to Clause 4B, wherein the second 1x1 convolution combination is derived from the extracted features of the previous layer.

[0273] Clause 6B. The method according to any one of Clauses 1B to 5B, wherein the one or more backbone blocks and the tail block each comprise a corresponding CP decomposition utilizing separable convolution.

[0274] Clause 7B. The method according to any one of Clauses 1B to 6B, wherein the head block, the transition block, the one or more backbone blocks and the tail block each comprise a corresponding CP decomposition utilizing separable convolution.

[0275] Clause 8B. The method according to any one of Clauses 1B to 7B, wherein the one or more backbone blocks do not include multi-scale branches.

[0276] Clause 9B. The method according to any one of Clauses 1B to 8B, wherein the header block does not include corresponding parameter modified linear unit activation (PReLU) to at least one input of the header block.

[0277] Clause 10B. The method according to Clause 9B, wherein the header block does not include a corresponding activation layer to at least one input of the header block.

[0278] Clause 11B. The method described in Clause 10B, wherein the header block does not include any activation layer.

[0279] Clause 12B. The method according to any one of Clauses 1B to 10B, wherein the head block includes at least one corresponding modified linear unit activation (ReLU) to at least one input of the head block.

[0280] Clause 13B. The method according to any one of Clauses 1B to 12B, wherein the head block comprises at least one multi-scale branch.

[0281] Clause 14B. The method according to any one of Clauses 1B to 13B, wherein each of the one or more backbone blocks comprises 24 channels.

[0282] Clause 15B. An apparatus for encoding or decoding at least one type of video data, the apparatus comprising: one or more memories configured to store the video data; and one or more processors configured to: receive data for reconstructing an image of the video data; reconstruct the image of the video data; and apply a neural network (NN)-based filter to the reconstructed image of the video data, wherein the NN-based filter includes a unified filter comprising a head block, a transform block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, transform block, one or more backbone blocks, or tail block comprises a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0283] Clause 16B. The device according to Clause 15B, wherein the CP decomposition comprises: at least one 1x1 convolution; convolution in the horizontal dimension; and convolution in the vertical dimension.

[0284] Clause 17B. The device according to Clause 16B, wherein the convolution in the horizontal dimension comprises one of a 1x3 convolution or a 3x1 convolution, and wherein the convolution in the vertical dimension comprises the other of the 1x3 convolution or the 3x1 convolution.

[0285] Clause 18B. The device according to Clause 17B, wherein the at least one 1x1 convolution comprises: a first 1x1 convolution occurring before the convolution in the horizontal dimension and the convolution in the vertical dimension; and a second 1x1 convolution occurring after the convolution in the horizontal dimension and the convolution in the vertical dimension.

[0286] Clause 19B. The apparatus as described in Clause 18B, wherein the second 1x1 convolution combination is derived from extracted features from the previous layer.

[0287] Clause 20B. The device according to any one of Clauses 15B to 19B, wherein the one or more backbone blocks and the tail block each comprise a corresponding CP decomposition utilizing separable convolution.

[0288] Clause 21B. The device according to any one of Clauses 15B to 20B, wherein the head block, the transition block, the one or more backbone blocks and the tail block each comprise a corresponding CP decomposition utilizing separable convolution.

[0289] Clause 22B. The device pursuant to any one of Clauses 15B to 21B, wherein the one or more backbone blocks do not include multi-scale branches.

[0290] Clause 23B. The device according to any one of Clauses 15B to 22B, wherein the head block does not include a corresponding parameter modified linear unit activation (PReLU) to at least one input of the head block.

[0291] Clause 24B. The device according to Clause 23B, wherein the header block does not include a corresponding activation layer to at least one input of the header block.

[0292] Clause 25B. The device as described in Clause 24B, wherein the head block does not include any activation layer.

[0293] Clause 26B. The device according to any one of Clauses 15B to 24B, wherein the head block includes at least one corresponding modified linear unit activation (ReLU) to at least one input of the head block.

[0294] Clause 27B. The device according to any one of Clauses 15B to 26B, wherein the head block comprises at least one multi-scale branch.

[0295] Clause 28B. The device pursuant to any one of Clauses 15B to 27B, wherein each of the one or more backbone blocks comprises 24 channels.

[0296] Clause 29B. The device according to any one of Clauses 15B to 28B, the device further comprising a display configured to display decoded video data.

[0297] Clause 30B. The device pursuant to any one of Clauses 15B to 29B, wherein the device includes one or more of a camera, computer, mobile device, broadcast receiver device or set-top box.

[0298] Clause 31B. An apparatus for encoding or decoding at least one of video data, the apparatus comprising: means for receiving data for reconstructing an image of the video data; means for reconstructing the image of the video data; and means for applying a neural network (NN) based filter to the reconstructed image of the video data, wherein the NN based filter includes a unified filter comprising a head block, a transform block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, transform block, one or more backbone blocks, or tail block comprises positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0299] Clause 32B. A non-transitory storage medium comprising instructions that, when executed, cause one or more processors to: receive data for reconstructing images of video data; reconstruct the images of the video data; and apply a neural network (NN)-based filter to the reconstructed images of the video data, wherein the NN-based filter includes a unified filter comprising a head block, a transition block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, the transition block, one or more backbone blocks, or the tail block comprises a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

[0300] It should be recognized that, based on the examples, certain actions or events of any technique described herein may be performed in a different order, and may be added, combined, or omitted entirely (e.g., not all actions or events described are necessary for implementing the technique). Furthermore, in some examples, actions or events may be performed concurrently (e.g., through multithreading, interrupt handling, or multiple processors) rather than sequentially.

[0301] In one or more examples, the described functionality may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functionality may be stored as one or more instructions or code on a computer-readable medium or transmitted via a computer-readable medium and executed by a hardware-based processing unit. A computer-readable medium may include a computer-readable storage medium (which corresponds to a tangible medium such as a data storage medium) or a communication medium, including, for example, any medium that facilitates the transfer of a computer program from one place to another according to a communication protocol. In this way, a computer-readable medium may generally correspond to (1) a non-transitory tangible computer-readable storage medium or (2) a communication medium, such as a signal or carrier wave. A data storage medium may be any available medium that can be accessed by one or more computers or one or more processors to retrieve instructions, code, and / or data structures for implementing the techniques described in this disclosure. Computer program products may include computer-readable media.

[0302] By way of example, and not limitation, such computer-readable storage media may include one or more of RAM, ROM, EEPROM, CD-ROM or other optical disc storage devices, magnetic disk storage devices or other magnetic storage devices, flash memory, or any other medium capable of storing desired program code in the form of instructions or data structures and accessible by a computer. Furthermore, any connection is appropriately referred to as a computer-readable medium. For example, if instructions are transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technology (such as infrared, radio, and microwave), then coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technology (such as infrared, radio, and microwave) is included in the definition of medium. However, it should be understood that computer-readable storage media and data storage media do not include connections, carrier waves, signals, or other transient media, but instead refer to non-transient tangible storage media. As used herein, disks and optical discs include compact optical discs (CDs), laser optical discs, optical discs, digital versatile optical discs (DVDs), floppy disks, and Blu-ray discs, where disks typically reproduce data magnetically, while optical discs use lasers to reproduce data optically. Combinations of these should also be included within the scope of computer-readable media.

[0303] Instructions can be executed by one or more processors, such as one or more DSPs, general-purpose microprocessors, ASICs, FPGAs, or other equivalent integrated or discrete logic circuits. Therefore, the terms "processor" and "processing circuit" as used herein can refer to any of the foregoing structures or any other structure suitable for implementing the techniques described herein. Additionally, in some aspects, the functionality described herein can be provided within dedicated hardware and / or software modules configured for encoding and decoding, or incorporated into combined codecs. Furthermore, these techniques can be fully implemented in one or more circuit or logic elements.

[0304] The techniques disclosed herein can be implemented in a wide variety of devices or apparatuses, including wireless mobile phones, integrated circuits (ICs), or a set of ICs (e.g., chipsets). Various components, modules, or units are described in this disclosure to emphasize functional aspects of a device configured to perform the disclosed techniques, but implementation by different hardware units is not necessarily required. Specifically, as described above, various units may be combined in a codec hardware unit, or various units may be provided by a collection of interoperable hardware units (including one or more processors as described above) combined with appropriate software and / or firmware.

[0305] Various examples have been described. These and other examples are within the scope of the following claims.

Claims

1. A method for encoding or decoding video data, the method comprising: Receive image data used to reconstruct the video data; The image is reconstructed from the video data; as well as A neural network (NN)-based filter is applied to the reconstructed image from the video data, wherein the NN-based filter includes a unified filter that includes a head block, a transition block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, the transition block, the one or more backbone blocks, or the tail block includes a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

2. The method according to claim 1, wherein the CP decomposition comprises: At least one 1x1 convolution; Convolution in the horizontal dimension; and Convolution in the vertical dimension.

3. The method of claim 2, wherein the convolution in the horizontal dimension comprises one of a 1x3 convolution or a 3x1 convolution, and wherein the convolution in the vertical dimension comprises the other of the 1x3 convolution or the 3x1 convolution.

4. The method of claim 3, wherein the at least one 1x1 convolution comprises: The first 1x1 convolution that occurs before the convolution in the horizontal dimension and the convolution in the vertical dimension; and A second 1x1 convolution occurs after the convolution in the horizontal dimension and the convolution in the vertical dimension.

5. The method of claim 4, wherein the second 1x1 convolution combination is derived from the extracted features of the previous layer.

6. The method of claim 1, wherein the one or more backbone blocks and the tail block each comprise a corresponding CP decomposition utilizing separable convolution.

7. The method of claim 1, wherein the head block, the transition block, the one or more backbone blocks and the tail block each comprise a corresponding CP decomposition utilizing separable convolution.

8. The method of claim 1, wherein the one or more skeleton blocks do not include multi-scale branches.

9. The method of claim 1, wherein the head block does not include corresponding parameter modified linear unit activation (PReLU) to at least one input of the head block.

10. The method of claim 9, wherein the header block does not include a corresponding activation layer to at least one input of the header block.

11. The method of claim 10, wherein the head block does not include any activation layer.

12. The method of claim 1, wherein the head block includes at least one corresponding modified linear unit activation (ReLU) to at least one input of the head block.

13. The method of claim 1, wherein the head block comprises at least one multi-scale branch.

14. The method of claim 1, wherein each of the one or more backbone blocks comprises 24 channels.

15. An apparatus for encoding or decoding video data, the apparatus comprising: One or more memories, the one or more memories being configured to store the video data; and One or more processors, said one or more processors being configured to: Receive image data used to reconstruct the video data; The image reconstructed from the video data; and A neural network (NN)-based filter is applied to the reconstructed image from the video data, wherein the NN-based filter includes a unified filter that includes a head block, a transition block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, the transition block, the one or more backbone blocks, or the tail block includes a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

16. The apparatus of claim 15, wherein the CP decomposition comprises: At least one 1x1 convolution; Convolution in the horizontal dimension; and Convolution in the vertical dimension.

17. The device of claim 16, wherein the convolution in the horizontal dimension comprises one of a 1x3 convolution or a 3x1 convolution, and wherein the convolution in the vertical dimension comprises the other of the 1x3 convolution or the 3x1 convolution.

18. The device of claim 17, wherein the at least one 1x1 convolution comprises: The first 1x1 convolution that occurs before the convolution in the horizontal dimension and the convolution in the vertical dimension; and A second 1x1 convolution occurs after the convolution in the horizontal dimension and the convolution in the vertical dimension.

19. The apparatus of claim 18, wherein the second 1x1 convolution combination is derived from extracted features from a previous layer.

20. The device of claim 15, wherein the one or more backbone blocks and the tail block each comprise a corresponding CP decomposition utilizing separable convolution.

21. The device of claim 15, wherein the head block, the transition block, the one or more backbone blocks and the tail block each comprise a corresponding CP decomposition utilizing separable convolution.

22. The device of claim 15, wherein the one or more backbone blocks do not include multi-scale branches.

23. The device of claim 15, wherein the head block does not include corresponding parameter modified linear unit activation (PReLU) to at least one input of the head block.

24. The device of claim 23, wherein the head block does not include a corresponding activation layer to at least one input of the head block.

25. The device of claim 24, wherein the head block does not include any activation layer.

26. The device of claim 15, wherein the head block includes at least one corresponding modified linear unit activation (ReLU) to at least one input of the head block.

27. The device of claim 15, wherein the head block comprises at least one multi-scale branch.

28. The device of claim 15, wherein each of the one or more backbone blocks comprises 24 channels.

29. The device of claim 15, further comprising a display configured to display decoded video data.

30. The device of claim 15, wherein the device comprises one or more of a camera, a computer, a mobile device, a broadcast receiver device, or a set-top box.

31. An apparatus for encoding or decoding video data, the apparatus comprising: A component for receiving data from images used to reconstruct the video data; The component used to reconstruct the image from the video data; and Components for applying neural network (NN) based filters to reconstructed images from video data, wherein the NN based filters include a unified filter, the unified filter including a head block, a transition block, one or more backbone blocks, and a tail block, and wherein at least one of the head block, the transition block, the one or more backbone blocks, or the tail block includes a positive quasi-multivariate (CP) decomposition utilizing separable convolutions.

Images