Entropy adaptation for depth feature compression using flexible networks

Abstract

A flexible neural architecture for end-to-end compression is used for image or video compression. In one embodiment, the first device is an encoder and the second device is a decoder. Transmission of a compression feature is performed between a first device and a second device. Mapping of the probability table is performed in the first and second devices. In one embodiment, a mapping is transmitted between the first and second devices by updating a bitstream for encoding features. In another embodiment, segmentation between the user equipment and the network is inferred.

Description

[0001] Cross - Reference to Related Applications This application claims the benefit of European Application Serial No. 22306666.3, filed on November 4, 2022, which is incorporated herein by reference in its entirety. Technical Field

[0002] At least one of the present embodiments generally relates to a method or apparatus for compressing images and videos using a neural network-based tool. Background Art

[0003] Compared with classical methods, machine learning (ML) has become a new tool for disruption compression. The main idea is to learn the entire compression chain, including content description, quantization, entropy coding, and descriptor decompression. Deep neural networks (DNNs) can be used for such compression. Summary of the Invention

[0004] At least one of the present embodiments generally relates to a method or apparatus in the context of compressing images and videos using a flexible neural architecture. In particular, an object of the described embodiments is to give the device the flexibility to select the probability distribution for its entropy encoder when using a flexible network. This allows each device to optimize the trade-off between coding complexity and coding efficiency based on its specific constraints.

[0005] According to a first aspect, a method is provided. The method includes the following steps: determining a probability table mapping by a first neural device based on the constraints of an image; transmitting the probability table mapping to a second neural device; retrieving an associated probability table based on the configuration of the first neural device; encoding the features of the first neural device based on the associated probability table; sending the encoded features to the second neural device; and re-evaluating the configuration of the first neural device.

[0006] According to a second aspect, a method is provided. The method includes the following steps: receiving a probability table mapping from a first neural device; receiving encoded features from the first neural device; and decoding the encoded features by a second device using the received probability table mapping to generate an image.

[0007] According to another aspect, an apparatus is provided. The apparatus includes a processor. The processor can be configured to implement the general aspects by executing any of the described methods.

[0008] In another general aspect according to at least one embodiment, there is provided a device including means according to any decoding embodiment; and at least one of the following: (i) an antenna configured to receive a signal including a video block, (ii) a band limiter configured to limit the received signal including the video block to a band, or (iii) a display configured to display an output representative of the video block.

[0009] In another general aspect according to at least one embodiment, there is provided a non-transitory computer-readable medium including data content generated according to any of the described encoding embodiments or variations.

[0010] In another general aspect according to at least one embodiment, there is provided a signal including video data generated according to any of the described encoding embodiments or variations.

[0011] In another general aspect according to at least one embodiment, a bitstream is formatted to include data content generated according to any of the described encoding embodiments or variations.

[0012] In another general aspect according to at least one embodiment, there is provided a computer program product including instructions that, when executed by a computer, cause the computer to perform any of the described decoding embodiments or variations.

[0013] These and other aspects, features, and advantages of the general aspects will become apparent from the following detailed description of the exemplary embodiments read in conjunction with the accompanying drawings.

[0014] In another general aspect according to at least one embodiment, there is provided a non-transitory computer-readable medium including data content including instructions for performing any encoding or decoding method. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A deep neural network (DNN) is shown.

[0016] Figure 2 An example of split inference of a neural network is shown.

[0017] Figure 3 A flexible neural network with two layers is shown.

[0018] Figure 4 An example of mapping selection in a flexible network is shown.

[0019] Figure 5 An example of a mapping for selecting a probability table is shown.

[0020] Figure 6Shows an example of an embodiment using the generally described aspects.

[0021] Figure 7 Shows an example of communication mapped to a device.

[0022] Figure 8 Shows an embodiment of the described aspect where the data source comes from a UE.

[0023] Figure 9 Shows another example of an embodiment of the described aspect, where the inference is split between a UE (User Equipment) and the network.

[0024] Figure 10 Shows an embodiment of a method for encoding video using the described embodiment.

[0025] Figure 11 Shows an embodiment of a method for decoding video using the described embodiment.

[0026] Figure 12 Shows an embodiment of an apparatus for encoding or decoding using the described embodiment.

[0027] Figure 13 Shows a standard general video compression scheme.

[0028] Figure 14 Shows a standard general video decompression scheme.

[0029] Figure 15 Shows a processor - based system for encoding / decoding under the generally described aspects. Detailed Description

[0030] The background of the described embodiments is the compression of image / video content. One goal of the aspects is to develop new methods for end - to - end neural compression. Although not yet standardized, MPEG is exploring these technologies.

[0031] Figure 1 Shows a deep neural network. The deep neural network (100) includes multiple neural layers, such as convolutional layers. In the figure, (130) is the first layer of the neural network. Each neural layer can be described as a function that first multiplies the input by a tensor, adds a value called a bias, and then applies a non - linear function to the resulting value. The shape (and other properties) of the tensor and the type of non - linear function are referred to as the architecture of the network. We use the term "weights" to represent the values of the tensor and the bias. The weights and parameters (if applicable) of the non - linear function are called parameters. The architecture and parameters define the "model".

[0032] Typically, a model M is trained on a database D of images to learn its weights. In supervised learning, the database contains input / output pairs (x, y), and the model M is a function that attempts to predict an output (120) from an input (110): Typically, the weights are optimized to minimize the training loss where d measures the difference between the actual and predicted outputs. For example, d can be the squared error or the Euclidean distance. The loss function can also include additional terms, such as regularization terms. Using the trained model is called inference.

[0033] Some neural networks are too large or computationally expensive to perform inference on a mobile device. To alleviate this problem, split inference can be performed. This is as Figure 2 shown. The neural network (100) is split into two parts. The first part runs on the mobile device (210), and the second part is run by another more powerful device (220). Thus, the intermediate features (250) of the neural network, i.e., the values computed at the split point of the network, must be transmitted from the first device to the second device.

[0034] The size of these intermediate features can be huge. To transmit them, they can be compressed to reduce their size. Several compression methods have been developed to compress these features. These features can be quantized or compressed, for example, by neural network layers similar to autoencoders. The latter is as Figure 2 shown. The first part (241) of the autoencoder lossily compresses the output of the first part of the layer into features. Then, the compressed features are losslessly compressed, typically using an entropy coding method, such as an arithmetic encoder (261). Such an encoder uses a probability distribution over the values to be encoded. On the second device, a corresponding decoder (262) and the second part (242) of the autoencoder decompress the features. Since the bandwidth of the transmission channel may vary over time, it may be necessary to adapt the compression ratio during operation. This can be done at the autoencoder level by using a so-called flexible autoencoder. Flexible models are specifically designed such that their computational characteristics can be modified on the fly. There are many designs for flexible networks, including designs that change the size of the intermediate features or the quantization level.

[0035] Figure 3A flexible neural network with two layers is shown from left to right. This flexible neural network is called a "slimmable" network. The flexibility is achieved by modifying the number of neurons in each layer - thus modifying the input / output size. The circles represent values: the input value (301), a set of intermediate values (302), and the output value (303). The lines represent the dependencies between consecutive layers. Specifically, the line connecting two values indicates that the rightmost value is calculated using the leftmost value. In other words, the line between two values indicates that the corresponding weight is non-zero. In this figure, the flexible network has two operating modes. In the first mode, only the elements with full lines are used: the network uses only the first half of the input, calculates only half of the output, and uses only 1 / 4 of the weights. The corresponding weight matrix (310) can thus be regarded as a matrix in which the elements of 3 / 4 are set to zero (white) and 1 / 4 are kept (the dashed part of the matrix). In the second mode, all annotations (solid and dashed lines) are used: all inputs are used, all outputs are calculated, and all weights are used. The corresponding matrix uses all the weights (320). Thus, this network provides a trade-off between computation and the information computed from the input.

[0036] When lossy compressing the transmitted features by changing the size of the feature vector, even in the remaining neurons, some weights are set to zero, resulting in a change in the value of the feature vector. Therefore, the distribution of the values of this vector also changes. Encoding the feature vector using an entropy encoder based on a single probability distribution over the values of the vector results in sub-optimal encoding. Therefore, different entropy distributions are used for each configuration of the flexible network [1]. For example, a probability table (or a set of probability tables) t k ∈{t 1 ,..., t K} is used for each configuration k ∈ {1,..., K} of the flexible network.

[0037] When the device does not have enough processing power for the entire network, split inference is usually used. Repeatedly changing the distribution used by the entropy encoder can be a burden for such a device. This problem is more severe when the entropy distribution is not encoded in a probability table but is dynamically computed by another neural network.

[0038] In the general aspect described, it is proposed that: 1. When using a flexible network, give the device the flexibility to choose the probability distribution it uses for the entropy encoder. This allows each device to optimize the trade-off between encoding complexity and encoding efficiency based on its specific constraints.

[0039] 2. An embodiment in which a device directly encodes its choice in the bitstream.

[0040] 3. Another embodiment allows two devices to synchronize the probability distributions they use to encode feature vectors when each of these devices performs a part of the neural network's computation and features are transferred from one device to another.

[0041] The key idea of these aspects is to give a device the possibility to choose the probability distribution it uses for entropy encoding at any time when using a flexible network in a neural network system. For clarity, unless otherwise stated, we will limit the description to the case of a prunable autoencoder network. However, it should be understood that the described embodiments apply to any other type of flexible neural network used in split inference.

[0042] Rather than using the associated probability table t k ∈{t 1 ,..., t K} for each configuration k ∈ {1,..., K} of the flexible network, we propose that the first device selects a mapping f(k): {1,..., K} → S, where S is a subset of {t 1 ,..., t K}. This is as Figure 4 shown. Each configuration of the network (410) is associated with a table index i(420). According to this example mapping, the table t i is used for configuration k, e.g., t 3 is used for configurations 1, 2, and 3. This mapping is freely chosen by the device based on its constraints (such as limited computational power, memory, or energy). Figure 5 Shows how this mapping is used to select the probability table for the encoder during operation. Figure 5 Shows the changes during device operation. At any point in time (503), the device uses a configuration (501) of the flexible network and a probability table (502) defined by the Figure 4 mapping. Here, the device has changed the configuration used by the flexible network six times during operation. However, due to the general aspect described, only 3 changes of the probability table have been made. This illustrates how the general aspect reduces the number of changes in the encoder, thus simplifying the encoder. The trade-off is a higher bit rate because the encoding is less efficient. For a device with low computational power, such as a device where split inference may be used, this trade-off may be interesting.

[0043] Not all mappings are obviously possible. The table mapped to the configuration must define at least the probability distribution in the output domain of the configuration. For example, for the case of a prunable autoencoder, the probability distribution t k defines the probability distribution over the first N k elements of the feature vector, where N kis the number of elements of the feature vector generated by configuration k. Thus, this probability table can be used to generate any configuration k′ with fewer elements of the feature vector. If the probability table contains independent probability distributions for each element of the feature vector, it can be simply adapted by discarding the N k -N k′ elements not used. In other cases, any probability table can be adapted by marginalizing it [1] to the N k′ elements of the feature vector.

[0044] On the other hand, for k′ < k, mapping the table t k′ to configuration k is less straightforward because t k′ does not contain any information about the probability distribution of the N k -N k′ elements of the feature vector not generated in configuration k′. To allow such a mapping, a mechanism must be defined to generate the probability distribution over these N k -N k′ elements. Possible examples include using a given possible distribution, such as a uniform probability distribution or marginalizing t k . The latter may involve a large computational cost. Both devices should know this mechanism. Although it may be less efficient in terms of coding efficiency, especially when using a uniform probability distribution, this mapping has the advantage of limiting the memory required for the maximum configuration of the flexible network.

[0045] As another example, consider the case where the flexibility of an autoencoder results in different quantization levels for the features of different configurations, and configuration k results in using more bits than k′. Then, the table t k can be easily used for configuration k′, for example, by marginalizing the values that are no longer used, or by allocating the probability mass of all values in the less quantized space (configuration k) that would be rounded to v to a given value v in the more quantized space (configuration k′). Both devices should know this mechanism.

[0046] For some neural networks, the values of the probability table used to encode some or all of the elements of the features are not static, but are dynamically computed by the neural network or a part of the neural network. This network can also be flexible. Without a generally described aspect, such a flexible network is controlled by the same index k as the other parts of the network. The described embodiments can also be used in this context and include using a different configuration for this table generation network than for the other parts of the network. For simplicity, when using configuration k, this network will be denoted by N k . A dynamically generated table can be marginalized in the same way as a static table. It should be noted that if k′ < k, where k′ represents the configuration of the main network, the N kCertain inputs. Several solutions are possible, such as filling in missing values with an agreed value, filling in each missing value with its average value, its last value, a weighted average of some last values, or an interpolation of these last values.

[0047] Figure 6 Illustrates how the described embodiments can be used in practice. First, the first device selects a mapping (610) by optimizing the mapping for its constraints and transmits it to the second device (620). The first device then selects a configuration k (630) and retrieves the associated probability table (640). The table is then used in an entropy encoder to encode the features (650), and the features are sent to the second device (660). Then, if the conditions change, for example, the available computing resources change, the device can re-evaluate its configuration. This communication can be done in various ways. Next, we describe two possible embodiments and list some variations.

[0048] Bitstream embodiment In this first embodiment, the mapping is transmitted to the second device by updating the bitstream used to encode the features. There are several ways to do this. Figure 7 Shows several examples. The figure is based on Figure 4 the mapping in. The first possibility (710) explicitly lists the mapping in the bitstream in the first part (711) of the bitstream by including an index i associated with a table for each possible configuration k associated with the flexible network, sorted in increasing k. Any other order is possible. In this example and all other examples, the mapping is included in the bitstream (719) before encoding the features and any other information, but any position is possible. If the table is only mapped to configurations with an index lower than that of the table and the table is not mapped to configurations with an index lower than that of another mapped table, the resulting bitstream (720) can include only the index of the tables used (721). Under the same conditions, another implementation can include the number of times the table is used (731) in the bitstream (730). This information can also be directly included in the bitstream or encoded differently, for example, by an entropy encoder.

[0049] Another possible embodiment is to transmit the mapping on the fly during execution. In other words, when a given configuration k is first used, the first device transmits the table used in that configuration k. Another possibility is to allow the first device to transmit the tables it uses at any time. This can be done with or without an initial mapping. If a mapping has been defined, different conventions are possible. For example, the table can be used until device 1 selects and transmits another table, until the configuration changes, or for a given number of transmitted feature vectors.

[0050] Device embodiment Figure 8 and 9 shows an embodiment according to the described aspect, where the data source comes from the UE (Device 1) and the inference is split between the UE and the network (Device 2). Figure 8 Describes a configuration phase, where the model data and additional model metadata are configured and sent before the runtime phase. Figure 9 Shows the runtime phase, where the UE generates intermediate data towards the network side.

[0051] In other embodiments, the network can be the data source, so the intermediate data will flow from the network side to the UE side. During Figure 8 the configuration phase of, the network sends a mapping table (t 1 ,..., t K ) together with different flexible model subsets or configurations (1,..., K).

[0052] During the runtime phase, the UE 1. Estimates the network and UE conditions, 2. Selects an adapted model subset among different model configurations 3. Selects the corresponding entropy distribution table, 4. Performs UE inference according to the selected model 5. Processes the input content and generates intermediate features 6. Encodes the intermediate features into a bitstream using the corresponding entropy distribution table 7. Segments the intermediate bitstream into different intermediate data delivery segments 8. Distributes the data segments and metadata indicating the model identifier and / or entropy distribution table identifier applied to the segments to the network.

[0053] Conversely, the network: 1. Receives the data segments and metadata indicating the model identifier and entropy distribution table identifier to be applied to the received segments. 2. Reconstructs the bitstream from the data segments. 3. Selects the network model according to the model identifier. 4. Selects the entropy distribution table according to the entropy distribution table identifier. 5. Decodes the bitstream using the corresponding entropy distribution table. 6. Performs network inference according to the selected network subset for the recovered intermediate features 7. Generates an inference output and sends it back to the UE.

[0054] In one embodiment, the UE can send the entropy distribution table identifier when the entropy distribution table identifier changes.

[0055] In another embodiment, the UE can send the entropy distribution table for the entropy encoder instead of the index.$.

[0056] In another embodiment, the UE sends the selected mapping to the network instead of the table identifier.

[0057] Variant: It is straightforward for the general aspects to be applied to other or all elements of the neural network that are flexible, in addition to the autoencoder.

[0058] A special case of these aspects is when the neural network is used for image or video compression. This type of compression is typically referred to as "end-to-end" neural compression. In this case, the input is typically an image or video, the transmitted feature is the encoding of the image or video, and the output of the network is the decoded image / video. The first device is the encoder, and the second device is the decoder.

[0059] The entropy distribution can also be encoded by the neural network.

[0060] As described here, the general aspects can be applied to end-to-end compression. In this case, flexible models have been studied in the academic literature. Several working groups in MPEG are currently considering end-to-end compression techniques. Different research groups in several companies are also researching end-to-end compression techniques. The current technology is developing rapidly, and in terms of the rate-distortion trade-off, these techniques are starting to compete with traditional compression techniques. Neural network-based compression is also used to enhance traditional compression. The described embodiments are applicable to any flexible neural architecture. Therefore, it has the potential to be used both for end-to-end video compression and for improving the neural network models used in traditional compression, even when there are changes in the prior art architectures.

[0061] The described aspects can also be applied to segmentation inference, which is currently being standardized by 3GPP in document S4-221160.

[0062] Segmentation inference is also studied in the context of the MPEG video coding group for machines.

[0063] Figure 10An embodiment of a method 1000 for encoding / decoding video data is shown. The method starts at start block 1001 and proceeds to block 1010 for determining a probability table mapping by a first neural device based on constraints on an image. Control proceeds from block 1010 to block 1020 for transmitting the probability table mapping to a second neural device. Control proceeds from block 1020 to block 1030 for retrieving an associated probability table based on the configuration of the first neural device. Control proceeds from block 1030 to block 1040 for encoding features of the first neural device based on the associated probability table. Control proceeds from block 1040 to block 1050 for sending the encoded features to the second neural device. Control proceeds from block 1050 to block 1060 for re-evaluating the configuration of the first neural device.

[0064] Figure 11 An embodiment of a method 1100 for decoding video data is shown. The method begins at start block 1101 and proceeds to block 1110 for receiving a probability table mapping from a first neural device. Control proceeds from block 1110 to block 1120 for receiving encoded features from the first neural device. Control proceeds from block 1120 to block 1130 for a second device to decode the encoded features using the received probability table mapping to generate an image.

[0065] Figure 12 An embodiment of an apparatus 1200 for compressing, encoding, or decoding video using the above method is shown. The apparatus includes a processor 1210 and may be interconnected with a memory 1220 through at least one port. Both the processor 1210 and the memory 1220 may also have one or more externally connected additional interconnections.

[0066] The processor 1210 is further configured to insert or receive information in a bitstream and perform compression, encoding, or decoding using the foregoing method.

[0067] The embodiments described herein include multiple aspects, including tools, features, embodiments, models, methods, etc. Many of these aspects are specifically described and are typically described in a way that may sound restrictive, at least to show the various characteristics. However, this is for clarity of description and does not limit the application or scope of these aspects. In fact, all the different aspects can be combined and interchanged to provide more aspects. Additionally, the described aspects can also be combined and interchanged with aspects described in earlier submissions.

[0068] The aspects described and contemplated in this application can be implemented in many different forms. Figure 13 、 14 and 15 provide some embodiments, but other embodiments are also expected, and Figure 13 、 14The discussion of 15 does not limit the scope of implementation. At least one aspect generally relates to video encoding and decoding, and at least one other aspect generally relates to transmitting the generated or encoded bitstream. These and other aspects can be implemented as methods, apparatuses, computer-readable storage media storing instructions for encoding or decoding video data according to any of the described methods, and / or computer-readable storage media storing a bitstream generated according to any of the described methods.

[0069] In this application, the terms "reconstruction" and "decoding" may be used interchangeably, the terms "pixel" and "sample" may be used interchangeably, and the terms "image", "picture", and "frame" may be used interchangeably. Generally, but not necessarily, the term "reconstruction" is used on the encoder side, while "decoding" or "reconstruction" is used on the decoder side.

[0070] Various methods are described herein, and each method includes one or more steps or actions for implementing the method. Unless the correct operation of the method requires a specific order of steps or actions, the order and / or use of specific steps and / or actions may be modified or combined. Additionally, terms such as "first", "second", etc. may be used in various embodiments to modify elements, components, steps, operations, etc., such as for example "first decoding" and "second decoding". Unless specifically required, the use of these terms does not imply an ordering of the modified operations. Thus, in this example, the first decoding does not need to be performed before the second decoding, and may occur, for example, before, during, or in a time period overlapping with the second decoding.

[0071] The various methods and other aspects described in this application can be used to modify the modules of the video encoder 100 and decoder 200 as shown in Figure 13 and Figure 14 For example, the intra prediction, entropy encoding, and / or decoding modules (160, 360, 145, 330). Additionally, this aspect is not limited to VVC or HEVC, and can be applied to, for example, other standards and recommendations, whether pre-existing or future-developed, and extensions of any such standards and recommendations (including VVC and HEVC). Unless otherwise specified or technically excluded, the aspects described in this application can be used alone or in combination.

[0072] Various numerical values are used in this application. The specific values are for illustrative purposes, and the described aspects are not limited to these specific values.

[0073] Figure 13 The encoder 100 is shown. Variants of the encoder 100 are envisioned, but for clarity, the encoder 100 is described below without describing all the expected variants.

[0074] Before being encoded, the video sequence may undergo pre-encoding processing (101). For example, a color transformation (e.g., conversion from RGB 4:4:4 to YCbCr 4:2:0) may be applied to the input color picture, or remapping of the input picture components may be performed to obtain a signal distribution that is more resilient to compression (e.g., histogram equalization using one of the color components). Metadata may be associated with the preprocessing and appended to the bitstream.

[0075] In the encoder 100, the picture is encoded by the encoder elements as described below. The picture to be encoded is partitioned (102) and processed in units such as CUs. Each unit is encoded using, for example, an intra-frame or inter-frame mode. When a unit is encoded in the intra-frame mode, it performs intra-frame prediction (160). In the inter-frame mode, motion estimation (175) and compensation (170) are performed. The encoder decides (105) which of the intra-frame mode or inter-frame mode to use to encode the unit and indicates the intra / inter-frame decision, for example, by a prediction mode flag. For example, the prediction residual is calculated by subtracting (110) the predicted block from the original image block.

[0076] The prediction residual is then transformed (125) and quantized (130). The quantized transform coefficients, along with the motion vectors and other syntax elements, are entropy encoded (145) to output the bitstream. The encoder may skip the transformation and directly apply quantization to the untransformed residual signal. The encoder may bypass the transformation and quantization, i.e., the residual is directly encoded without applying the transformation or quantization process.

[0077] The encoder decodes the encoded blocks for use as a reference for further prediction. The quantized transform coefficients are dequantized (140) and inverse-transformed (150) to decode the prediction residual. The decoded prediction residual and the predicted block are combined (155) to reconstruct the image block. An in-loop filter (165) is applied to the reconstructed picture to perform, for example, deblocking / SAO (sample adaptive offset) filtering to reduce encoding artifacts. The filtered image is stored in the reference picture buffer (180).

[0078] Figure 14 A block diagram of the video decoder 200 is shown. In the decoder 200, the bitstream is decoded by the decoder elements as described below. The video decoder 200 generally performs a decoding process that is the reverse of the encoding process as shown Figure 13 The encoder 100 generally also performs video decoding as part of encoding the video data.

[0079] Specifically, the input to the decoder includes a video bitstream, which may be generated by the video encoder 100. The bitstream is first entropy decoded (230) to obtain transform coefficients, motion vectors, and other coding information. The picture partitioning information indicates how the picture is partitioned. Thus, the decoder can partition (235) the picture according to the decoded picture partitioning information. The transform coefficients are dequantized (240) and inverse transformed (250) to decode the prediction residuals. The decoded prediction residuals and the prediction blocks are combined (255) to reconstruct the image blocks. The prediction blocks can be obtained (270) from intra prediction (260) or motion compensated prediction (i.e., inter prediction) (275). An in-loop filter (265) is applied to the reconstructed image. The filtered image is stored in the reference image buffer (280).

[0080] The decoded picture may further undergo post-decoding processing (285), such as an inverse color transformation (e.g., conversion from YcbCr 4:2:0 to RGB 4:4:4) or an inverse remapping, which performs the inverse of the remapping process performed in the precoding process (101). The post-decoding processing may use metadata derived in the precoding process and signaled in the bitstream.

[0081] Figure 15 A block diagram illustrating an example of a system implementing various aspects and embodiments is shown. The system 1000 may be implemented as a device including various components described below and configured to perform one or more aspects described in this document. Examples of such devices include, but are not limited to, various electronic devices such as personal computers, laptop computers, smartphones, tablet computers, digital multimedia set-top boxes, digital television receivers, personal video recording systems, connected household appliances, and servers. The elements of the system 1000 may be embodied singly or in combination in a single integrated circuit (IC), multiple ICs, and / or discrete components. For example, in at least one embodiment, the processing and encoder / decoder elements of the system 1000 are distributed over multiple ICs and / or discrete components. In various embodiments, the system 1000 is communicatively coupled to one or more other systems or other electronic devices via, for example, a communication bus or through dedicated input and / or output ports. In various embodiments, the system 1000 is configured to implement one or more aspects described herein.

[0082] System 1000 includes at least one processor 1010 configured to execute instructions loaded therein for implementing various aspects described herein, for example. The processor 1010 may include embedded memory, input / output interfaces, and various other circuits known in the art. System 1000 includes at least one memory 1020 (e.g., volatile memory devices and / or non-volatile memory devices). System 1000 includes a storage device 1040, which may include non-volatile memory and / or volatile memory, including but not limited to electrically erasable programmable read-only memory (EEPROM), read-only memory (ROM), programmable read-only memory (PROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, disk drives, and / or optical disk drives. As a non-limiting example, the storage device 1040 may include internal storage devices, attached storage devices (including removable and non-removable storage devices), and / or network-accessible storage devices.

[0083] System 1000 includes an encoder / decoder module 1030 configured to, for example, process data to provide encoded video or decoded video, and the encoder / decoder module 1030 may include its own processor and memory. The encoder / decoder module 1030 represents (one or more) modules that may be included in a device to perform encoding and / or decoding functions. As is well known, a device may include one or both of an encoding and a decoding module. Additionally, the encoder / decoder module 1030 may be implemented as a stand-alone element of System 1000, or may be incorporated within the processor 1010 as a combination of hardware and software known to those skilled in the art.

[0084] Program code to be loaded onto the processor 1010 or encoder / decoder 1030 to execute the various aspects described herein may be stored in the storage device 1040 and subsequently loaded onto the memory 1020 for execution by the processor 1010. According to various embodiments, one or more of the processor 1010, memory 1020, storage device 1040, and encoder / decoder module 1030 may store one or more of various items during the execution of the processes described herein. Such stored items may include but are not limited to input video, decoded video or portions of decoded video, bitstreams, matrices, variables, and intermediate or final results from processing of equations, formulas, operations, and operation logic.

[0085] In some embodiments, the memory internal to the processor 1010 and / or the encoder / decoder module 1030 is used to store instructions and provide working memory for the processing required during encoding or decoding. However, in other embodiments, memory external to the processing device (e.g., the processing device may be the processor 1010 or the encoder / decoder module 1030) is used for one or more of these functions. The external memory may be the memory 1020 and / or the storage device 1040, e.g., dynamic volatile memory and / or non-volatile flash memory. In several embodiments, the external non-volatile flash memory is used to store, for example, the operating system of a television. In at least one embodiment, fast external dynamic volatile memory such as RAM is used as the working memory for video encoding and decoding operations, such as for MPEG-2 (MPEG refers to the Moving Picture Experts Group, MPEG-2 is also known as ISO / IEC 13818, and 13818-1 is also known as H.222, 13818-2 is also known as H.262), HEVC (HEVC refers to High Efficiency Video Coding, also known as H.265 and MPEG-H Part 2), or VVC (Versatile Video Coding, a new standard developed by JVET, the Joint Video Experts Team).

[0086] As shown in block 1130, input can be provided to the elements of the system 1000 via various input devices. Such input devices include, but are not limited to, (i) an RF (radio frequency) section that receives, for example, RF signals transmitted over the air by a broadcaster, (ii) component (COMP) input terminals (or a set of COMP input terminals), (iii) universal serial bus (USB) input terminals, and / or (iv) high-definition multimedia interface (HDMI) input terminals. Figure 15 Other examples not shown include composite video.

[0087] In various embodiments, the input devices of block 1130 have respective associated input processing elements, as known in the art. For example, the RF section may be associated with elements suitable for: (i) selecting a desired frequency (also referred to as selecting a signal, or band-limiting a signal band to a band), (ii) down-converting the selected signal, (iii) again band-limiting the narrower band to select a signal band that may be referred to as a channel in some embodiments, (iv) demodulating the down-converted and band-limited signal, (v) performing error correction, and (vi) demultiplexing to select a desired data packet stream. The RF section of various embodiments includes one or more elements that perform these functions, such as, for example, a frequency selector, a signal selector, a band limiter, a channel selector, a filter, a down-converter, a demodulator, an error corrector, and a demultiplexer. The RF section may include a tuner that performs various of these functions, including, for example, down-converting a received signal to a lower frequency (e.g., an intermediate frequency or near-baseband frequency) or to baseband. In one set-top box embodiment, the RF section and its associated input processing elements receive an RF signal transmitted through a wired (e.g., cable) medium and perform frequency selection by filtering, down-converting, and again filtering to a desired band. Various embodiments rearrange the order of the above (and other) elements, remove some of these elements, and / or add other elements that perform similar or different functions. Adding elements may include inserting elements between existing elements, such as, for example, inserting an amplifier and an analog-to-digital converter. In various embodiments, the RF section includes an antenna.

[0088] In addition, the USB and / or HDMI terminals may include respective interface processors for connecting the system 1000 to other electronic devices via the USB and / or HDMI connections. It should be understood that various aspects of input processing, such as Reed-Solomon error correction, may be implemented, as needed, within, for example, a separate input processing IC or within the processor 1010. Similarly, as needed, various aspects of USB or HDMI interface processing may be implemented within a separate interface IC or within the processor 1010. The demodulated, error-corrected, and demultiplexed stream is provided to various processing elements, including, for example, the processor 1010 and the encoder / decoder 1030, which operate in conjunction with memory and storage elements to process the data stream as needed for presentation on an output device.

[0089] The various elements of the system 1000 may be provided within an integrated housing. Within the integrated housing, the various elements may be interconnected using suitable connection means and data may be transmitted between them, such as internal buses known in the art, including the inter-integrated circuit (I2C) bus, wiring, and printed circuit boards.

[0090] System 1000 includes a communication interface 1050 that is capable of communicating with other devices via a communication channel 1060. The communication interface 1050 may include, but is not limited to, a transceiver configured to send and receive data via the communication channel 1060. The communication interface 1050 may include, but is not limited to, a modem or a network card, and the communication channel 1060 may be implemented in, for example, wired and / or wireless media.

[0091] In various embodiments, a wireless network such as a Wi-Fi network, e.g., IEEE 802.11 (IEEE refers to the Institute of Electrical and Electronics Engineers), is used to stream or otherwise provide data to system 1000. The Wi-Fi signals of these embodiments are received via a communication channel 1060 and a communication interface 1050 suitable for Wi-Fi communication. The communication channel 1060 of these embodiments is typically connected to an access point or a router that provides access to an external network (including the Internet) to allow streaming applications and other over-the-top communications. Other embodiments use a set-top box to provide streaming data to system 1000, and the set-top box delivers data via an HDMI connection of an input block 1130. Other embodiments use an RF connection of the input block 1130 to provide streaming data to system 1000. As described above, various embodiments provide data in a non-streaming manner. In addition, various embodiments use wireless networks other than Wi-Fi, such as cellular networks or Bluetooth networks.

[0092] System 1000 may provide output signals to various output devices, including a display 1100, speakers 1110, and other peripheral devices 1120. The display 1100 of various embodiments includes, for example, one or more of a touchscreen display, an organic light-emitting diode (OLED) display, a curved display, and / or a foldable display. The display 1100 may be used in a television, a tablet, a laptop, a mobile phone (cellular phone), or other devices. The display 1100 may also be integrated with other components (e.g., in a smart phone) or be standalone (e.g., an external monitor for a laptop). In various embodiments, other peripheral devices 1120 include one or more of a standalone digital video disc (or digital versatile disc) (DVR, both terms), a disc player, a stereo system, and / or a lighting system. Various embodiments use one or more peripheral devices 1120 that perform functions based on the output of system 1000. For example, a disc player performs the function of playing the output of system 1000.

[0093] In various embodiments, control signals are transmitted between system 1000 and display 1100, speaker 1110, or other peripheral devices 1120 using signaling such as AV.Link, Consumer Electronics Control (CEC), or other communication protocols capable of enabling device-to-device control with or without user intervention. Output devices can be communicatively coupled to system 1000 via dedicated connections through respective interfaces 1070, 1080, and 1090. Alternatively, output devices can be connected to system 1000 using communication channel 1060 through communication interface 1050. Display 1100 and speaker 1110 can be integrated into a single unit with other components of system 1000 in an electronic device such as, for example, a television. In various embodiments, display interface 1070 includes a display driver, such as a timing controller (T Con) chip.

[0094] For example, if the RF portion of input 1130 is part of a standalone set-top box, display 1100 and speaker 1110 can alternatively be separated from one or more other components. In various embodiments where display 1100 and speaker 1110 are external components, output signals can be provided via dedicated output connections including, for example, HDMI ports, USB ports, or COMP outputs.

[0095] Embodiments can be implemented by computer software implemented by processor 1010, or by hardware, or by a combination of hardware and software. As a non-limiting example, embodiments can be implemented by one or more integrated circuits. Memory 1020 can be of any type suitable for the technical environment and can be implemented using any appropriate data storage technology, as non-limiting examples, such as optical memory devices, magnetic memory devices, semiconductor-based memory devices, fixed memory, and removable memory. As a non-limiting example, processor 1010 can be of any type suitable for the technical environment and can include one or more of a microprocessor, a general-purpose computer, a special-purpose computer, and a processor based on a multi-core architecture.

[0096] Various implementations involve decoding. "Decoding" as used in this application can include, for example, performing all or part of a process on a received coded sequence to produce a final output suitable for display. In various embodiments, these processes include one or more processes typically performed by a decoder, such as entropy decoding, inverse quantization, inverse transform, and differential decoding. In various embodiments, the processes also or alternatively include processes performed by the decoders of the various implementations described in this application.

[0097] As a further example, in one embodiment, "decoding" refers only to entropy decoding, in another embodiment, "decoding" refers only to differential decoding, and in another embodiment, "decoding" refers to a combination of entropy decoding and differential decoding. Based on the context of the specific description, it will be clear whether the phrase "decoding process" is intended to refer specifically to a subset of operations or to a more general decoding process, and this is considered to be well understood by those skilled in the art.

[0098] Various implementations involve encoding. In a manner similar to the discussion above regarding "decoding", "encoding" as used in this application can include, for example, all or part of the processes performed on an input video sequence to produce an encoded bitstream. In various embodiments, these processes include one or more processes typically performed by an encoder, such as partitioning, differential encoding, transformation, quantization, and entropy encoding. In various embodiments, such processes also or alternatively include processes performed by the encoders of the various implementations described in this application.

[0099] As a further example, in one embodiment, "encoding" refers only to entropy encoding, in another embodiment, "encoding" refers only to differential encoding, and in another embodiment, "encoding" refers to a combination of differential encoding and entropy encoding. Based on the context of the specific description, it will be clear whether the phrase "encoding process" is intended to refer specifically to a subset of operations or to a more general encoding process, and this is considered to be well understood by those skilled in the art.

[0100] Note that the grammatical elements used here are descriptive terms. Thus, they do not exclude the use of other grammatical element names.

[0101] When a figure is represented as a flowchart, it should be understood that it also provides a block diagram of the corresponding apparatus. Similarly, when a figure is presented as a block diagram, it should be understood that it also provides a flowchart of the corresponding method / process.

[0102] Various embodiments may relate to parametric models or rate - distortion optimization. In particular, during the encoding process, a balance or trade - off between rate and distortion is typically considered, usually given a constraint on computational complexity. It can be measured by a rate - distortion optimization (RDO) metric, or by least mean square (LMS), mean absolute error (MAE), or other such metrics. Rate - distortion optimization is typically formulated as minimizing a rate - distortion function, which is a weighted sum of rate and distortion. There are different ways to solve the rate - distortion optimization problem. For example, these methods can be based on an extensive test of all encoding options, including all considered modes or encoding parameter values, and a complete evaluation of their encoding costs and the associated distortion of the reconstructed signal after encoding and decoding. Faster methods can also be used to save encoding complexity, especially by calculating an approximate distortion based on the predicted or prediction - residual signal rather than the reconstructed signal. A hybrid of these two methods can also be used, for example, by using approximate distortion for only some of the possible encoding options and full distortion for others. Other methods only evaluate a subset of the possible encoding options. More generally, many methods employ any of a variety of techniques to perform the optimization, but the optimization is not necessarily a complete evaluation of encoding cost and associated distortion.

[0103] The implementations and aspects described herein can be implemented in, for example, a method or process, an apparatus, a software program, a data stream, or a signal. Even if discussed only in the context of a single form of implementation (e.g., only as a method), the implementation of the features discussed can be implemented in other forms (e.g., an apparatus or a program). The apparatus can be implemented in, for example, appropriate hardware, software, and firmware. These methods can be implemented in, for example, a processor, which generally refers to a processing device, including, for example, a computer, a microprocessor, an integrated circuit, or a programmable logic device. The processor also includes communication devices, such as, for example, a computer, a cellular phone, a portable / personal digital assistant (“PDA”), and other devices that facilitate information communication between end - users.

[0104] References to “one embodiment” or “an embodiment” or “one implementation” or “an implementation” and other variations thereof mean that the particular features, structures, characteristics, etc. described in connection with that embodiment are included in at least one embodiment. Thus, the appearances of the phrases “in one embodiment” or “in an embodiment” or “in one implementation” or “in an implementation” and any other variations throughout the application do not necessarily all refer to the same embodiment.

[0105] In addition, this application may refer to “determining” various pieces of information. Determining information can include, for example, one or more of estimating information, calculating information, predicting information, or retrieving information from a memory.

[0106] In addition, the present application may refer to "accessing" various pieces of information. Accessing information may include, for example, receiving information, retrieving information (e.g., from a memory), storing information, moving information, copying information, computing information, determining information, predicting information, or estimating information, or one or more of these.

[0107] In addition, the present application may refer to "receiving" various pieces of information. Like "accessing", receiving is a broad term. Receiving information may include, for example, one or more of accessing information or retrieving information (e.g., from a memory). Further, during operations such as, for example, storing information, processing information, sending information, moving information, copying information, erasing information, computing information, determining information, predicting information, or estimating information, "receiving" is generally involved in one way or another.

[0108] It should be understood that the use of any of the following, " / ", "and / or", and "at least one of...", e.g., in the case of "A / B", "A and / or B", and "at least one of A and B", is intended to include only the first-listed option (A), or only the second-listed option (B), or both options (A and B). As a further example, in the case of "A, B, and / or C" and "at least one of A, B, and C", such wording is intended to include only the first-listed option (A), or only the second-listed option (B), or only the third-listed option (C), or only the first and second-listed options (A and B), or only the first and third-listed options (A and C), or only the second and third-listed options (B and C), or all three options (A and B and C). As will be clear to those of ordinary skill in the art and related fields, this can be extended to any number of items listed.

[0109] In addition, as used herein, the word "signal" particularly refers to indicating something to a corresponding decoder. For example, in some embodiments, the encoder signals a particular one of a plurality of transforms, coding modes, or flags. In this way, in one embodiment, the same transform, parameter, or mode may be used on both the encoder side and the decoder side. Thus, for example, the encoder may send (explicit signaling) a particular parameter to the decoder such that the decoder may use the same particular parameter. Conversely, if the decoder already has a particular parameter as well as other parameters, signaling may be used without transmission (implicit signaling) to simply allow the decoder to know and select the particular parameter. By avoiding the transmission of any actual functionality, bit savings are achieved in various embodiments. It should be understood that signaling may be implemented in a variety of ways. For example, in various embodiments, one or more syntax elements, flags, etc. are used to signal information to the corresponding decoder. Although the previous discussion related to the verb form of the word "signal", the word "signal" may also be used as a noun herein.

[0110] It will be apparent to those of ordinary skill in the art that implementations can generate multiple signals that are formatted to carry information such as can be stored or transmitted. The information can include, for example, instructions for performing a method or data generated by one of the described implementations. For example, a signal can be formatted to carry a bitstream of the described embodiments. Such a signal can be formatted as, for example, an electromagnetic wave (e.g., using the radio frequency portion of the spectrum) or a baseband signal. Formatting can include, for example, encoding a data stream and modulating a carrier with the encoded data stream. The information carried by the signal can be, for example, analog or digital information. The signal can be transmitted over various different known wired or wireless links. The signal can be stored on a processor-readable medium.

[0111] The previous sections have described multiple embodiments across various claim categories and types. The features of these embodiments can be provided individually or in any combination. Additionally, embodiments can include one or more of the following features, devices, or aspects, either individually or in any combination across various claim categories and types.

[0112] We have described multiple embodiments across various claim categories and types. The features of these embodiments can be provided individually or in any combination. Additionally, embodiments can include one or more of the following features, devices, or aspects, either individually or in any combination across various claim categories and types, including a first device that includes a user equipment and a second device that includes a network.

[0113] In one embodiment, mapping information for image compression is transmitted from a first device to a second device using bitstream data between the first and second devices.

[0114] In one embodiment, inference is split between a user equipment and a network.

[0115] The present disclosure contemplates creating and / or transmitting and / or receiving and / or decoding a bitstream or signal that includes one or more of the described syntax elements or variants thereof.

[0116] In one embodiment, a TV, set-top box, mobile phone, tablet, or other electronic device performs the (multiple) transformation methods according to any of the described embodiments.

[0117] In one embodiment, a TV, set-top box, mobile phone, tablet, or other electronic device performs the (multiple) transformation methods determined according to any of the described embodiments and displays (e.g., using a monitor, screen, or other type of display) the resulting image.

[0118] In one embodiment, a TV, set-top box, cellular phone, tablet, or other electronic device selects, band-limits, or tunes (e.g., using a tuner) a channel to receive a signal including an encoded image and performs a transform method according to any of the described embodiments(s).

[0119] In one embodiment, a TV, set-top box, cellular phone, tablet, or other electronic device receives (e.g., using an antenna) a signal including an encoded image over the air and performs a transform method(s).

[0120] [1]https: / / en.wikipedia.org / wiki / Marginal_distribution。

Claims

1. A method, comprising: determining a probability table mapping by a first neural device based on image constraints; transmitting the probability table mapping to a second neural device; retrieving an associated probability table based on the configuration of the first neural device; encoding features of the first neural device based on the associated probability table; sending the encoded features to the second neural device; and reevaluating the configuration of the first neural device.

2. An apparatus, comprising: a memory, and a processor configured to execute: determining a probability table mapping by a first neural device based on image constraints; transmitting the probability table mapping to a second neural device; retrieving an associated probability table based on the configuration of the first neural device; encoding features of the first neural device based on the associated probability table; sending the encoded features to the second neural device; and reevaluating the configuration of the first neural device.

3. A method, comprising: receiving a probability table mapping from a first neural device; receiving encoded features from the first neural device; and decoding, by a second device, the encoded features using the received probability table mapping to generate an image.

4. An apparatus, comprising: a memory, and a processor configured to execute: receiving a probability table mapping from a first neural device; receiving encoded features from the first neural device; and decoding, by a second device, the encoded features using the received probability table mapping to generate an image.

5. The method according to any one of claims 1 or 3, or the apparatus according to any one of claims 2 or 4, wherein the probability table mapping is transmitted from a first device within an encoded bitstream.

6. The method or apparatus according to claim 5, wherein the probability table mapping includes an index associated with possible configurations, or the number of times a corresponding table is used.

7. The method or apparatus according to claim 5, wherein the probability table mapping is transmitted on-the-fly during execution.

8. The method according to any one of claims 1 or 3, or the apparatus according to any one of claims 2 or 4, wherein the first device is a user device and the second device is a network.

9. The method or apparatus according to claim 8, wherein the network is a data source of the user device.

10. The method or apparatus according to claim 8, wherein the user device sends information including an entropy distribution table identifier, entropy distribution table usage, or a selected mapping to the network.

11. A device, comprising: the apparatus according to any one of claims 4 to 10; and at least one of the following: (i) an antenna configured to receive a signal including a video block, (ii) a band limiter configured to limit a received signal including a video block to a frequency band, and (iii) a display configured to display an output representative of the video block.

12. A non-transitory computer-readable medium containing data content, the data content being generated according to the method of any one of claims 1 and 5 to 10 or by the apparatus of any one of claims 2 and 5 to 10 for playback using a processor.

13. A signal, comprising video data, the video data being generated according to the method of any one of claims 1 and 5 to 10 or by the apparatus of any one of claims 2 and 5 to 10 for playback using a processor.

14. A computer program product, comprising instructions which, when executed by a computer, cause the computer to perform the method of any one of claims 1, 3 or 4 to 10.

15. A non-transitory computer-readable medium containing data content, the data content including instructions for performing the method of any one of claims 1 or 3 and 5 to 10.

Images