Approximating implicit neural representation through learnt dictionary atoms

Abstract

In one implementation, we model the information content in the signal as a composition of the global and local information. The global information is common and shared for all the natural signals, and this can be learned from a large collection of data. However, the local information is specific to each signal. By leveraging the compositional properties of neural networks, we propose to decompose the INR network into head and tail layers, where the head layers are responsible for the global information and tail for the local information. The weights of the head layers are approximated through the dictionary basis, and this dictionary is learnt from a large collection of the data and known to both encoder and decoder. Thus, we need to transmit only the weights of the tail layers plus some extra information about the head layers.

Description

APPROXIMATING IMPLICIT NEURAL REPRESENTATION THROUGH LEARNT DICTIONARY ATOMSTECHNICAL FIELD[1] The present embodiments generally relate to a method and an apparatus for neural compression.BACKGROUND[2] Neural compression or learning-based compression is the application of neural networks and other machine learning methods to data compression. Those techniques are currently being investigated by MPEG, and there is a new ad-hoc group which focuses on the Implicit Neural Representation-based compression (INR-based) within Working Group 4. Typically, INR-based compression techniques have a far lower computational complexity than end-to-end neural compression approaches.SUMMARY[3] According to an embodiment, a method of decoding video data representative of an image or a 3D scene is presented, comprising: decoding a set of sparse coefficients; obtaining a dictionary corresponding to a first set of layers of an INR (Implicit Neural Representation) network, said INR network including said first set of layers and a second set of layers; decoding parameters of said first set of layers of said INR network, based on said set of sparse coefficients and said dictionary; decoding parameters of said second set of layers of said INR network; and performing inference to decode said image or 3D scene based on said INR network.[4] According to another embodiment, a method for encoding video data representative of an image or a 3D scene is presented, comprising: obtaining parameters representing an INR (Implicit Neural Representation) network used to encode said image or 3D scene; decomposing said INR network into a first set of layers and a second set of layers; obtaining a dictionary corresponding to said first set of layers; obtaining a set of sparse coefficients to represent said first set of layers based on said parameters for said first set of layers and said dictionary; encoding said set of sparse coefficients; and encoding parameters of said second set of layers of said INR network.[5] According to another embodiment, an apparatus for decoding video data representative of an image or a 3D scene, comprising one or more processors and at least one memory, wherein said one or more processors are configured to: decode a set of sparse coefficients; obtain a dictionary corresponding to a first set of layers of an INR (Implicit Neural Representation) network, said INR network including said first set of layers and a second set of layers; decode parameters of said first set of layers of said INR network, based on said set of sparse coefficients and said dictionary; decode parameters of said second set of layers of said INR network; and perform inference to decode said image or 3D scene based on said INR network.[6] According to another embodiment, an apparatus for encoding video data representative of an image or a 3D scene is presented, comprising one or more processors and at least one memory, wherein said one or more processors are configured to: obtain parameters representing an INR (Implicit Neural Representation) network used to encode said image or 3D scene; decompose said INR network into a first set of layers and a second set of layers; obtain a dictionary corresponding to said first set of layers; obtain a set of sparse coefficients to represent said first set of layers based on said parameters for said first set of layers and said dictionary; encode said set of sparse coefficients; and encode parameters of said second set of layers of said INR network.[7] One or more embodiments also provide a computer program comprising instructions which when executed by one or more processors cause the one or more processors to perform the encoding method or decoding method according to any of the embodiments described herein. One or more of the present embodiments also provide a computer readable storage medium having stored thereon instructions for video encoding or decoding according to the methods described herein.[8] One or more embodiments also provide a computer readable storage medium having stored thereon video data generated according to the methods described above. One or more embodiments also provide a method and apparatus for transmitting or receiving the video data generated according to the methods described herein.BRIEF DESCRIPTION OF THE DRAWINGS[9] FIG. 1 illustrates a block diagram of a system within which aspects of the present embodiments may be implemented.

[0010] FIG. 2 illustrates a simple neural network used for Implicit Neural Representation(INR).

[0011] FIG. 3 illustrates a typical process to encode a signal using INR.

[0012] FIG. 4 illustrates a process to encode a signal using INR through learnt dictionary, according to an embodiment.

[0013] FIG. 5 illustrates a process to decode a signal using INR through leamt dictionary, according to an embodiment.DETAILED DESCRIPTION

[0014] FIG. 1 illustrates a block diagram of an example of a system in which various aspects and embodiments can be implemented. System 100 may be embodied as a device including the various components described below and is configured to perform one or more of the aspects described in this application. 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 home appliances, and servers. Elements of system 100, singly or in combination, may be embodied in a single integrated circuit, multiple ICs, and / or discrete components. For example, in at least one embodiment, the processing and encoder / decoder elements of system 100 are distributed across multiple ICs and / or discrete components. In various embodiments, the system 100 is communicatively coupled to other systems, or to other electronic devices, via, for example, a communications bus or through dedicated input and / or output ports. In various embodiments, the system 100 is configured to implement one or more of the aspects described in this application.

[0015] The system 100 includes at least one processor 110 configured to execute instructions loaded therein for implementing, for example, the various aspects described in this application. Processor 110 may include embedded memory, input output interface, and various other circuitries as known in the art. The system 100 includes at least one memory 120 (e.g., a volatile memory device, and / or a non-volatile memory device). System 100 includes a storage device 140, which may include non-volatile memory and / or volatile memory, including, but not limited to, EEPROM, ROM, PROM, RAM, DRAM, SRAM, flash, magnetic disk drive, and / or optical disk drive. The storage device 140 may include an internal storage device, an attached storage device, and / or a network accessible storage device, as non-limiting examples.

[0016] System 100 includes an encoder / decoder module 130 configured, for example, toprocess data to provide an encoded video or decoded video, and the encoder / decoder module 130 may include its own processor and memory. The encoder / decoder module 130 represents module(s) that may be included in a device to perform the encoding and / or decoding functions. As is known, a device may include one or both of the encoding and decoding modules. Additionally, encoder / decoder module 130 may be implemented as a separate element of system 100 or may be incorporated within processor 110 as a combination of hardware and software as known to those skilled in the art.

[0017] Program code to be loaded onto processor 110 or encoder / decoder 130 to perform the various aspects described in this application may be stored in storage device 140 and subsequently loaded onto memory 120 for execution by processor 110. In accordance with various embodiments, one or more of processor 110, memory 120, storage device 140, and encoder / decoder module 130 may store one or more of various items during the performance of the processes described in this application. Such stored items may include, but are not limited to, the input video, the decoded video or portions of the decoded video, the bitstream, matrices, variables, and intermediate or final results from the processing of equations, formulas, operations, and operational logic.

[0018] In several embodiments, memory inside of the processor 110 and / or the encoder / decoder module 130 is used to store instructions and to provide working memory for processing that is needed during encoding or decoding. In other embodiments, however, a memory external to the processing device (for example, the processing device may be either the processor 110 or the encoder / decoder module 130) is used for one or more of these functions. The external memory may be the memory 120 and / or the storage device 140, for example, a dynamic volatile memory and / or a non-volatile flash memory. In several embodiments, an external non-volatile flash memory is used to store the operating system of a television. In at least one embodiment, a fast external dynamic volatile memory such as a RAM is used as working memory for video coding and decoding operations, such as for MPEG-2, JPEG Pleno, MPEG-I, HEVC, VVC or MPEG VCM.

[0019] The input to the elements of system 100 may be provided through various input devices as indicated in block 105. Such input devices include, but are not limited to, (i) an RF portion that receives an RF signal transmitted, for example, over the air by a broadcaster, (ii) a Composite input terminal, (iii) a USB input terminal, and / or (iv) an HDMI input terminal.

[0020] In various embodiments, the input devices of block 105 have associated respective inputprocessing elements as known in the art. For example, the RF portion may be associated with elements suitable for (i) selecting a desired frequency (also referred to as selecting a signal, or band-limiting a signal to a band of frequencies), (ii) down converting the selected signal, (iii) band-limiting again to a narrower band of frequencies to select (for example) a signal frequency band which may be referred to as a channel in certain embodiments, (iv) demodulating the down converted and band-limited signal, (v) performing error correction, and (vi) demultiplexing to select the desired stream of data packets. The RF portion of various embodiments includes one or more elements to perform these functions, for example, frequency selectors, signal selectors, band-limiters, channel selectors, filters, downconverters, demodulators, error correctors, and demultiplexers. The RF portion may include a tuner that performs various of these functions, including, for example, down converting the received signal to a lower frequency (for example, an intermediate frequency or a near-baseband frequency) or to baseband. In one set-top box embodiment, the RF portion and its associated input processing element receives an RF signal transmitted over a wired (for example, cable) medium, and performs frequency selection by filtering, down converting, and filtering again to a desired frequency band. Various embodiments rearrange the order of the above-described (and other) elements, remove some of these elements, and / or add other elements performing similar or different functions. Adding elements may include inserting elements in between existing elements, for example, inserting amplifiers and an analog-to-digital converter. In various embodiments, the RF portion includes an antenna.

[0021] Additionally, the USB and / or HDMI terminals may include respective interface processors for connecting system 100 to other electronic devices across USB and / or HDMI connections. It is to be understood that various aspects of input processing, for example, Reed- Solomon error correction, may be implemented, for example, within a separate input processing IC or within processor 110 as necessary. Similarly, aspects of USB or HDMI interface processing may be implemented within separate interface ICs or within processor 110 as necessary. The demodulated, error corrected, and demultiplexed stream is provided to various processing elements, including, for example, processor 110, and encoder / decoder 130 operating in combination with the memory and storage elements to process the datastream as necessary for presentation on an output device.

[0022] Various elements of system 100 may be provided within an integrated housing. Within the integrated housing, the various elements may be interconnected and transmit data therebetween using suitable connection arrangement 115, for example, an internal bus asknown in the art, including the I2C bus, wiring, and printed circuit boards.

[0023] The system 100 includes communication interface 150 that enables communication with other devices via communication channel 190. The communication interface 150 may include, but is not limited to, a transceiver configured to transmit and to receive data over communication channel 190. The communication interface 150 may include, but is not limited to, a modem or network card and the communication channel 190 may be implemented, for example, within a wired and / or a wireless medium.

[0024] Data is streamed to the system 100, in various embodiments, using a Wi-Fi network such as IEEE 802. 11. The Wi-Fi signal of these embodiments is received over the communications channel 190 and the communications interface 150 which are adapted for WiFi communications. The communications channel 190 of these embodiments is typically connected to an access point or router that provides access to outside networks including the Internet for allowing streaming applications and other over-the-top communications. Other embodiments provide streamed data to the system 100 using a set-top box that delivers the data over the HDMI connection of the input block 105. Still other embodiments provide streamed data to the system 100 using the RF connection of the input block 105.

[0025] The system 100 may provide an output signal to various output devices, including a display 165, speakers 175, and other peripheral devices 185. The other peripheral devices 185 include, in various examples of embodiments, one or more of a stand-alone DVR, a disk player, a stereo system, a lighting system, and other devices that provide a function based on the output of the system 100. In various embodiments, control signals are communicated between the system 100 and the display 165, speakers 175, or other peripheral devices 185 using signaling such as AV. Link, CEC, or other communications protocols that enable device-to-device control with or without user intervention. The output devices may be communicatively coupled to system 100 via dedicated connections through respective interfaces 160, 170, and 180. Alternatively, the output devices may be connected to system 100 using the communications channel 190 via the communications interface 150. The display 165 and speakers 175 may be integrated in a single unit with the other components of system 100 in an electronic device, for example, a television. In various embodiments, the display interface 160 includes a display driver, for example, a timing controller (T Con) chip.

[0026] The display 165 and speaker 175 may alternatively be separate from one or more of the other components, for example, if the RF portion of input 105 is part of a separate set-top box.In various embodiments in which the display 165 and speakers 175 are external components, the output signal may be provided via dedicated output connections, including, for example, HDMI ports, USB ports, or COMP outputs.

[0027] FIG. 2 illustrates a simple neural network used for implicit neural representation (INR). Such a neural network used for INR can be referred to as an INR network. For clarity, we use for illustration a 2D signal such as an image, but INR can be used for signals of any dimension. INR parameterizes a signal as a function (200), which takes coordinates (210) as input and outputs potentially approximated values (220) of a signal at these coordinates. INR has recently been applied to images, 2D videos or 3D objects among other applications. In the image case, the inputs (210) can be pixel coordinates (x, y) and the INR outputs (220) the color values (r, g, b) of the input pixels. In the video case, the output is similar, but the input can include the frame index t in addition to pixel coordinates. The INR can be used to reconstruct a signal by computing the signal values for every necessary coordinate inputs.

[0028] An INR network (200) is typically a neural network composed of multiple neural layers, such as fully connected layers. In FIG. 2, the network has four neural layers. Intermediate outputs are represented by circles. Each neural layer can be described as a function that first multiplies the input by a tensor, adds a vector called the bias and then applies a nonlinear function on the resulting values. In this document, we may also refer to “neural layer” simply as “layer.” The shape (and other characteristics) of the tensor and the type of non-linear functions are called the architecture of the network. We will denote the values of the tensor and the bias by the term “weights”. The weights and, if applicable, the parameters of the nonlinear functions, are called the parameters 0 of the network. The architecture and the parameters define a “model”. We will use feto denote an INR function parameterized by 0.

[0029] FIG. 3 illustrates a typical process to encode a signal using INR. This is done by optimizing the parameters 0 (or a subset of them) of the INR network to reconstruct (310) the signal and encoding (320) parameters to create the output bitstream. For an image I of size (Af x IV), the parameters 0 or (the chosen subset) can for example be optimized by minimizing the following loss function:where D is a distortion which quantifies the difference between the image predicted (reconstructed) by fdand the original image 7, R is the bitrate of the encoded parameters andA a trade-off parameter between D and R. D could be any differentiable distortion measure, such as mean squared error as shown in Eq. (2). M and N are the width and height of an image. Other metrics such as LPIPS (Learned Perceptual Image Patch Similarity) can also be used in this case. The optimization of the weights 0 is typically performed by a machine learning approach such as a batch / stochastic gradient descent method.

[0030] To decompress the signal, fdis evaluated at all relevant coordinates. These coordinates can be selected at decoding. A typical choice would be all pixel coordinates for an image or video. As an example, for a 256x256 pixel image, these coordinates could be all pairs (x, y) for all x 6 {0,1, ... ,255} and y 6 {0,1, ... ,255}. Other choices are possible, for example to upsample, downsample or extend the original image.

[0031] There are some issues in compressing a signal with INR. For example, using the limited capacity of an INR network to overfit the signal makes the optimization difficult due to a limited number of parameters. As the capacity of INR increases, the bitrate also increases, thus increasing the capacity of INR is not an option in the compression. In another example, using one INR network globally for the whole signal makes learning difficult, as all parameters contribute to all values and lead to a large network as it must encode all details of the signal.

[0032] There is not much direct work in the literature to overcome the first issue, apart from the solutions like having the high-capacity model and then pruning / quantizing of the INR network to decrease the bitrate. On the second issue, hybrid INR is used to work around this issue, where the input coordinates are first mapped to a vector of features, then these features are then used as input for an INR network that outputs the value of the signal for the input coordinates. Using such an architecture rather than a plain INR network helps to handle local features of the input signal.

[0033] Using one INR network forces the network to encode everything, meaning that all parameters contribute to the whole signal. This makes training difficult in the case of small network and limits representation capacity of neural network or may lead to a large network and increases the bitrate. To decrease the bitrate of high-capacity INR network, model compression techniques like quantization / pruning can be applied but might not decrease the bitrate drastically. Furthermore, the redundant information contained in the image is not handled very well. In this context, we propose solution that a part of the INR network is amortized for all the signals, and the remaining part of the INR network is specialized for the underlying signal to be encoded. Here, “amortized” means that the dictionary encodesknowledge that is reused (either spatially or temporally), such that common / shared knowledge is transmitted only once over the network and its cost is amortized by the number of frames for instance, in case of video encoding. By this solution, the amortized part of INR captures the redundant or global information, and this can be large enough. Thus, it is possible to increase the capacity of the INR network and better model the redundant information.

[0034] In this document, we propose solutions to increase the capacity of the INR network and to better handle local contents in a signal compressed by an INR. We model the information content in the signal as a composition of the global and local information. The global information is common and shared for all the natural signals, and this can be learned from a large collection of data. However, the local information is specific to each signal. For example, suppose an image composed of N objects, each object having a consistent texture, but not exactly the same for all the blocks composing the object. Hence, the global information here would refer to the knowledge or representation bound to each object, while the local information refers to the fine-tuning / adaptation of this global knowledge for each block / part / superpixel / image portion in the object.

[0035] By leveraging the compositional properties of neural networks, we propose to decompose the INR network into head and tail layers, where the head layers are responsible for the global information and tail for the local information. In one embodiment, the decomposition into head and tail layers is up to the user. Exploiting the compositional property of the neural network, a function is decomposed into a combination / composition of head (global information) and tail (local) layers. Typically, the head would need to have a larger capacity to be general (e.g., from 6 to 10 layers) while the tail is an adaptation with limited layers (1 to 3 layers). The user chooses the setting, what is head and what is tail, that could depend on the complexity of the signal to be encoded.

[0036] The weights of the head layers are approximated through the dictionary basis, and this dictionary is leamt from a large collection of the data, and it is known to both encoder and decoder. Thus, we need to transmit only the weights of the tail layers plus some extra information about the head layers.

[0037] To better model the local redundancy, we also consider the case where the input domain is partitioned, for example, using a coding tree unit (CTU), superpixels or other kind of partition method, such as uniform size patch. We also describe the full pipeline to generate a bitstream for the signal encoding.

[0038] Dictionary approximation: To approximate the redundant contents in a signal, we propose to decompose the parameters of the INR network fQinto at least two parts and to approximate the parameters of at least one part of the INR network using a dictionary approximation. By doing so, the INR network can be encoded by the weights of the nonapproximated parts and some additional information that describes the approximated parts. There are multiple possible approaches to decompose the INR network into parts. For example, a part may consist of the bias and / or the weights and / or the parameters of the non-linear functions and / or any subset of these elements. Such a subset may for example be defined as a subset of the layers, such as the last k layers, or the bias of the last k layers, or a subset of the neurons. In the remaining of the disclosure, we will take as an example an INR divided into head and tail layers, as f0= feg.feh, and we transmit only the weights of the tail layers with some additional information about the head layers. This is an example of a decomposition into a composited representation, and other types of decompositions could be used without any loss of generality.

[0039] First, we describe the case where one INR is used to encode the entire image, later we describe the case where the input is partitioned, and different INRs are used for different partitions.

[0040] To mutualize the redundant information across images or to increase the capacity of the INR network, we propose to leam a dictionary D = [dlrd2, ... dK] with k atoms so that each head layer is approximated (represented) using a sparse linear combination of the atoms of the dictionary. That is,The sparse coefficients y are for example approximated by optimizing the following loss function, argwhere Qhis the weights of the head layers, D is the leamt dictionary, and y is the sparse coefficients to be optimized. To enforce the sparsity in the coefficient vector, the LI norm is used and a is the trade-off between the two terms in the equation.

[0041] Once the sparse coefficients are estimated, it is transmitted to the decoder side and using this, the approximated head layers are computed from the dictionary D in the decoder. It is noted that the dictionary is known to both encoder and decoder. In this case, the size of the head layers can be large enough to increase the representation capacity of the INR, as it won’t iobe included in the bitstream.

[0042] Dictionary learning: Now, we describe how the dictionary D can be leamt. Let 0h=Q2h, ... , 0phbe the collection of head layers, where P is the number of images / data (collection of data). Learning the dictionary D to summarize or approximate the representation of head layers is formulated as the following optimization problem, argT,where D is the dictionary to be estimated, and T is the sparse coefficient matrix, || . ||xis the LI norm to enforce the sparsity in the coefficient matrix T and a is the trade-off parameter to control the level of the sparsity: the higher the value, the greater the level of sparsity.

[0043] The solution to problem (5) can be found using alternative minimization by fixing one variable and optimizing another alternatively. More specifically, given the dictionary D, the sparse coefficient matrix T is optimized using any sparse solvers, such as Lasso. In the next step, the sparse coefficient matrix T is fixed, and optimized for the dictionary D, for example, using (stochastic) gradient descent or co-ordinate descent method. These alternative steps are continued until convergence or a certain number of iterations.

[0044] The leamt dictionary D should be able to represent diverse information contained in the head layers. For this, the collection of head layers is generated from a large dataset of images representing the diversity of structures found in natural images.

[0045] Dictionary used at encoding. Once the dictionary D is leamt, it is accessible for both encoder and decoder. In one example, the encoding of a given image or video is performed as illustrated in FIG. 4.• At step 410, an INR composed of head and tail layers are overfitted for the underlying image.• At step 420, the sparse coefficients (y's) are computed to approximate head layers from the off-the-shelf dictionary D using sparse coding, e.g., by solving equation (4).• At step 430, the sparse coefficients (y's) are entropy coded directly, or they can be quantized using fixed bit quantization and then encoded in the bit-stream, for example as follows: o the maximal value is found: am= maxy^. o The values are normalized as: y = i / amo Fixed bit quantization using 2qbits is performed to obtain the symbols to be transmitted Yj = Q(y ) = round(y * {(2} — 1)) o The quantization bit q is included in the bit-stream unless it is pre-defined (known to both side). The maximum values amand the list of symbols are also encoded into the bit-stream.• At step 440, the weights of the tail layers are quantized (fixed bit quantization) and encoded in the bit-stream using any entropy coder or codec, for example NNC.

[0046] In the case where we use the CTU or patch-based partition to handle local content, the head layers and tail layers are overfitted for each specific CTU or patch. The dictionary may therefore be leamt from a collection of head layers generated from the patches or CTU from the large collection of data. For encoding an image / video with partition, we repeat the above listed steps for each partition, and we also signal the partition in the bitstream to the decoder.

[0047] It is possible to obtain different bitrates by varying the level of the sparsity in the sparse coding, the higher the number of non-zero coefficients, the better the approximation of the head layers is.

[0048] From the bit-stream, in one example, decoding is performed as illustrated in FIG. 5.• At step 510, quantization value q is decoded. The maximum value amand the list of symbolsare decoded. Then the sparse coefficients are dequantized:=*am- Note that this quantization / dequantization method is used as an example and any other methods can be used without loss of generality.• At step 520, the weights of the tail layers are decoded from the bit-stream.

[0049] Once all the required information is decoded from the bit-stream, at step 530, the decoding of the image or video is performed as follows:• Decode / recompute the signal partition if partition is used.• For each partition: o Select the decoded sparse coefficients of the partition. o Compute the head layers weights using the sparse linear combination of dictionary atoms as Qh~ DQ-1(Y), and use them in the head layers of the INR network. o Use (540) the decoded tail layers weights in the tail layers of the INR network, o Perform (550) inference using all the coordinates in the corresponding partition.

[0050] As described above, the proposed methods are related to learning based compression.In particular, representing a major part of the INR network parameters using dictionary-based approximation paves the way towards better modeling of the spatial redundancy of the signal and also increases the capacity of the INR, hence leading to better compression performance. Furthermore, these approaches have a far lower computational complexity than end-to-end neural compression approaches.

[0051] In the above, the proposed methods are mainly described with respects to 2D video compression. However, INR is investigated for many other signals, in particular 3D scenes or objects, and the proposed methods can be used for any signal.

[0052] In the above, the unmodified coordinates of the image are used as input and rgb values are output. More generally, the input / output can be in other representations. For example, the input can be normalized coordinates, or the output can be in the YUV format. In addition, the image itself could also be encoded differently, for example, the output could be Fourier or Cosine transforms coefficients.

[0053] Various numeric values are used in the present application. The specific values are for example purposes and the aspects described are not limited to these specific values.

[0054] Various methods are described herein, and each of the methods comprises one or more steps or actions for achieving the described method. Unless a specific order of steps or actions is required for proper operation of the method, 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 an element, component, step, operation, etc., such as, for example, a “first decoding” and a “second decoding”. Use of such terms does not imply an ordering to the modified operations unless specifically required. So, in this example, the first decoding need not be performed before the second decoding, and may occur, for example, before, during, or in an overlapping time period with the second decoding.

[0055] The implementations and aspects described herein may be implemented in, for example, a method or a process, an apparatus, a software program, a data stream, or a signal. Even if only discussed in the context of a single form of implementation (for example, discussed only as a method), the implementation of features discussed may also be implemented in other forms (for example, an apparatus or program). An apparatus may be implemented in, for example, appropriate hardware, software, and firmware. The methods may be implemented in, for example, an apparatus, for example, a processor, which refers to processing devices in general, including, for example, a computer, a microprocessor, an integrated circuit, or a programmablelogic device. Processors also include communication devices, for example, computers, cell phones, portable / personal digital assistants (“PDAs”), and other devices that facilitate communication of information between end-users.

[0056] Reference to “one embodiment” or “an embodiment” or “one implementation” or “an implementation”, as well as other variations thereof, means that a particular feature, structure, characteristic, and so forth described in connection with the embodiment is included in at least one embodiment. Thus, the appearances of the phrase “in one embodiment” or “in an embodiment” or “in one implementation” or “in an implementation”, as well any other variations, appearing in various places throughout this application are not necessarily all referring to the same embodiment.

[0057] Additionally, this application may refer to “determining” various pieces of information. Determining the information may include one or more of, for example, estimating the information, calculating the information, predicting the information, or retrieving the information from memory.

[0058] Further, this application may refer to “accessing” various pieces of information. Accessing the information may include one or more of, for example, receiving the information, retrieving the information (for example, from memory), storing the information, moving the information, copying the information, calculating the information, determining the information, predicting the information, or estimating the information.

[0059] Additionally, this application may refer to “receiving” various pieces of information. Receiving is, as with “accessing”, intended to be a broad term. Receiving the information may include one or more of, for example, accessing the information, or retrieving the information (for example, from memory). Further, “receiving” is typically involved, in one way or another, during operations, for example, storing the information, processing the information, transmitting the information, moving the information, copying the information, erasing the information, calculating the information, determining the information, predicting the information, or estimating the information.

[0060] It is to be appreciated that the use of any of the following “ / ”, “and / or”, and “at least one of’, for example, in the cases of “A / B”, “A and / or B” and “at least one of A and B”, is intended to encompass the selection of the first listed option (A) only, or the selection of the second listed option (B) only, or the selection of both options (A and B). As a further example, in the cases of “A, B, and / or C” and “at least one of A, B, and C”, such phrasing is intended toencompass the selection of the first listed option (A) only, or the selection of the second listed option (B) only, or the selection of the third listed option (C) only, or the selection of the first and the second listed options (A and B) only, or the selection of the first and third listed options (A and C) only, or the selection of the second and third listed options (B and C) only, or the selection of all three options (A and B and C). This may be extended, as is clear to one of ordinary skill in this and related arts, for as many items as are listed.

[0061] As will be evident to one of ordinary skill in the art, implementations may produce a variety of signals formatted to carry information that may be, for example, stored or transmitted. The information may include, for example, instructions for performing a method, or data produced by one of the described implementations. For example, a signal may be formatted to carry the bitstream of a described embodiment. Such a signal may be formatted, for example, as an electromagnetic wave (for example, using a radio frequency portion of spectrum) or as a baseband signal. The formatting may include, for example, encoding a data stream and modulating a carrier with the encoded data stream. The information that the signal carries may be, for example, analog or digital information. The signal may be transmitted over a variety of different wired or wireless links, as is known. The signal may be stored on a processor-readable medium.

Claims

CLAIMS1. A method of decoding video data representative of an image or a 3D scene, comprising: decoding a set of sparse coefficients; obtaining a dictionary corresponding to a first set of layers of an INR (Implicit Neural Representation) network, said INR network including said first set of layers and a second set of layers; decoding parameters of said first set of layers of said INR network, based on said set of sparse coefficients and said dictionary; decoding parameters of said second set of layers of said INR network; and performing inference to decode said image or 3D scene based on said INR network.

2. The method of claim 1, wherein said image or 3D scene includes a plurality of partitions, and said decoding a set of sparse coefficients, said decoding parameters of said first set of layers and said performing inference are performed for each of said plurality of partitions.

3. A method for encoding video data representative of an image or a 3D scene, comprising: obtaining parameters representing an INR (Implicit Neural Representation) network used to encode said image or 3D scene; decomposing said INR network into a first set of layers and a second set of layers; obtaining a dictionary corresponding to said first set of layers; obtaining a set of sparse coefficients to represent said first set of layers based on said parameters for said first set of layers and said dictionary; encoding said set of sparse coefficients; and encoding parameters of said second set of layers of said INR network.

4. The method of claim 3, wherein said image or 3D scene includes a plurality of partitions, and said obtaining a set of sparse coefficients, said encoding said set of sparse coefficients, and said encoding parameters of said second set of layers are performed for each of said plurality of partitions.

5. The method of any one of claims 1-4, wherein parameters for said dictionary are not transmitted from an encoder to a corresponding decoder.

6. The method of any one of claims 1-5, wherein one or more parameters for said first set of layers are transmitted from an encoder to a corresponding decoder.

7. The method of any one of claims 1-6, wherein said first set of layers corresponds to global information of said image or 3D scene and said second set of layers corresponds to local information of said image or 3D scene.

8. The method of any one of claims 1-7, wherein said dictionary includes a set of atoms, wherein each layer in said first set of layers is represented using a sparse linear combination of said set of atoms of said dictionary.

9. An apparatus for decoding video data representative of an image or a 3D scene, comprising one or more processors and at least one memory, wherein said one or more processors are configured to: decode a set of sparse coefficients; obtain a dictionary corresponding to a first set of layers of an INR (Implicit Neural Representation) network, said INR network including said first set of layers and a second set of layers; decode parameters of said first set of layers of said INR network, based on said set of sparse coefficients and said dictionary; decode parameters of said second set of layers of said INR network; and perform inference to decode said image or 3D scene based on said INR network.

10. The apparatus of claim 9, wherein said image or 3D scene includes a plurality of partitions, and said one or more processors are configured to decode a set of sparse coefficients, to decode parameters of said first set of layers and to perform inference for each of said plurality of partitions.

11. An apparatus for encoding video data representative of an image or a 3D scene, comprising one or more processors and at least one memory, wherein said one or more processors are configured to: obtain parameters representing an INR (Implicit Neural Representation) network used to encode said image or 3D scene; decompose said INR network into a first set of layers and a second set of layers;obtain a dictionary corresponding to said first set of layers; obtain a set of sparse coefficients to represent said first set of layers based on said parameters for said first set of layers and said dictionary; encode said set of sparse coefficients; and encode parameters of said second set of layers of said INR network.

12. The apparatus of claim 11, wherein said image or 3D scene includes a plurality of partitions, and wherein said one or more processors are configured to obtain a set of sparse coefficients, to encode said set of sparse coefficients, and to encode parameters of said second set of layers for each of said plurality of partitions.

13. The apparatus of any one of claims 9-12, wherein parameters for said dictionary are not transmitted from an encoder to a corresponding decoder.

14. The apparatus of any one of claims 9-13, wherein one or more parameters for said first set of layers are transmitted from an encoder to a corresponding decoder.

15. The apparatus of any one of claims 9-14, wherein said first set of layers corresponds to global information of said image or 3D scene and said second set of layers corresponds to local information of said image or 3D scene.

16. The apparatus of any one of claims 9-15, wherein said dictionary includes a set of atoms, wherein each layer in said first set of layers is represented using a sparse linear combination of said set of atoms of said dictionary.

17. A signal comprising video data representative of an image or a 3D scene, formed by performing the method of any one of claims 3-8.

18. A computer readable storage medium having stored thereon instructions for encoding or decoding video data according to the method of any one of claims 1-8.

Images