DEVICES AND METHODS

Description

1 / 133 “DEVICES AND METHODS I. Cross-Reference to Related Deposit Requests

[0001] The present application claims the benefit of priority of non-provisional US patent application No. 18 / 183,867, in common ownership, filed on March 14, 2023, the content of which is expressly incorporated herein by reference in its entirety. II. Field

[0002] This disclosure refers generally to performing predictions using a compression network. III. Description of the Related Technique

[0003] Advances in technology have resulted in increasingly smaller and more powerful computing devices. For example, there is currently a variety of portable personal computing devices, including cordless phones such as mobile phones and smartphones, tablets, and laptop-type computers, which are small, lightweight, and easily carried by users. These devices can communicate voice and data packets over wireless networks. Additionally, many of these devices incorporate additional functionalities such as a digital camera, a digital video camera, a digital recorder, and an audio file player. Furthermore, such devices can process executable instructions, including software applications such as a web browser application, which can be used to access the internet. In this way, these devices can include significant computing capabilities.

[0004] Such computing devices often incorporate functionality for processing large Petition 870250078561, dated 03 / 09 / 2025, page 7 / 313 2 / 133 Data quantities. Compressing data before storage or transmission can conserve resources such as memory and bandwidth. For example, a computing device can generate an encoded version of an image frame that uses fewer bits than the original image frame. Techniques that reduce the size of compressed data can further conserve resources. IV. Summary

[0005] According to one implementation of the present disclosure, a device includes one or more processors configured to obtain encoded data associated with one or more motion values. The one or more processors are also configured to obtain conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values. The one or more processors are further configured to process, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values.

[0006] According to another implementation of the present disclosure, a method includes obtaining, on a device, encoded data associated with one or more motion values. The method also includes obtaining, on the device, conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values. The method also includes processing, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values. Petition 870250078561, dated 03 / 09 / 2025, page 8 / 313 3 / 133

[0007] According to another implementation of the present disclosure, a non-transient, computer-readable medium includes instructions that, when executed by one or more processors, cause the one or more processors to obtain encoded data associated with one or more motion values. The instructions, when executed by the one or more processors, also cause the one or more processors to obtain conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values. The instructions, when executed by the one or more processors, further cause the one or more processors to process, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values.

[0008] According to another implementation of the present disclosure, an apparatus includes means for obtaining encoded data associated with one or more motion values. The apparatus also includes means for obtaining conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values. The apparatus further includes means for processing, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values.

[0009] According to another implementation of the present disclosure, a device includes one or more processors configured to obtain conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values. The one Petition 870250078561, dated 03 / 09 / 2025, page 9 / 313 4 / 133 or more processors are additionally configured to process, using the compression network, the conditional input and one or more motion values ​​to generate encoded data associated with the one or more motion values.

[0010] According to another implementation of the present disclosure, a method includes obtaining, on a device, conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values. The method also includes processing, using the compression network, the conditional input and one or more motion values ​​to generate encoded data associated with the one or more motion values.

[0011] According to another implementation of the present disclosure, a non-transient, computer-readable medium includes instructions that, when executed by one or more processors, cause the one or more processors to obtain a conditional input from a compression network, wherein the conditional input is based on one or more predicted first movement values. The instructions, when executed by the one or more processors, also cause the one or more processors to process, using the compression network, the conditional input and one or more movement values ​​to generate encoded data associated with the one or more movement values.

[0012] According to another implementation of the present disclosure, an apparatus includes means for obtaining conditional input from a compression network, wherein the conditional input is based on one or more predicted first movement values. The apparatus also includes means for processing, using the compression network, the conditional input. Petition 870250078561, dated 03 / 09 / 2025, page 10 / 313 5 / 133 and one or more movement values ​​to generate encoded data associated with the one or more movement values.

[0013] Other aspects, advantages and attributes of this disclosure will become apparent after reviewing the entire application, including the following sections: Brief description of the drawings, Detailed description and Claims. V. Brief Description of the Drawings

[0014] Figure 1 is a block diagram of a particular illustrative aspect of an operable compression network for performing prediction, according to some examples in this disclosure.

[0015] Figure 2 is a diagram of an illustrative aspect of an operable system for performing prediction using a compression network, according to some examples in this disclosure.

[0016] Figure 3 is a diagram of an illustrative aspect of an operable system for performing prediction using a compression network, according to some examples in this disclosure.

[0017] Figure 4 is a diagram of an illustrative aspect of an example of an encoder portion of a compression network that is operable to perform prediction, according to some examples in this disclosure.

[0018] Figure 5 is a diagram of an illustrative aspect of an example of a decoder portion of the compression network that is operable to perform prediction, according to some examples in this disclosure.

[0019] Figure 6 is a diagram of an illustrative aspect of an example of an encoder portion of a compression network that is operable to perform prediction. Petition 870250078561, dated 03 / 09 / 2025, page 11 / 313 6 / 133 corresponding to reconstruction, according to some examples in this publication.

[0020] Figure 7 is a diagram of an illustrative aspect of an example of a decoder portion of the compression network that is operable to perform prediction matching reconstruction, according to some examples in this disclosure.

[0021] Figure 8 is a diagram of an illustrative aspect of an example of an encoding portion of a compression network that is operable to perform prediction matching reconstruction based on auxiliary prediction, according to some examples in this disclosure.

[0022] Figure 9 is a diagram of an illustrative aspect of an example of a decoder portion of a compression network that is operable to perform prediction matching reconstruction based on auxiliary prediction, according to some examples in this disclosure.

[0023] Figure 10 is a diagram of an illustrative aspect of an example of an encoder portion of a compression network that is operable to perform low-latency reconstruction matching prediction, according to some examples in this disclosure.

[0024] Figure 11 is a diagram of an illustrative aspect of an example of a decoder portion of the compression network that is operable to perform prediction matching low-latency reconstruction, according to some examples in this disclosure.

[0025] Figure 12 is a diagram of an illustrative aspect of an example of an encoder portion of a compression network that is operable to perform prediction. Petition 870250078561, dated 03 / 09 / 2025, page 12 / 313 7 / 133 corresponds to the future prediction, according to some examples from this disclosure.

[0026] Figure 13 is a diagram of an illustrative aspect of an example of a decoder portion of the compression network that is operable to perform matching prediction of future prediction, according to some examples in this disclosure.

[0027] Figure 14 is a diagram of an illustrative aspect of an example of a motion estimation encoder portion of a compression network that is operable to perform image reconstruction matching prediction, according to some examples in this disclosure.

[0028] Figure 15 is a diagram of an illustrative aspect of an example of a motion estimation decoder portion of the compression network that is operable to perform image reconstruction matching prediction, according to some examples in this disclosure.

[0029] Figure 16 is a diagram of an illustrative aspect of an example of an image estimator of a compression network that is operable to perform prediction corresponding to image reconstruction, according to some examples in this disclosure.

[0030] Figure 17 is a diagram of an illustrative aspect of an example of an image encoder portion of the compression network that is operable to perform prediction corresponding to image reconstruction, according to some examples in this disclosure.

[0031] Figure 18 is a diagram of an illustrative aspect of an example of a decoder portion of Petition 870250078561, dated 03 / 09 / 2025, page 13 / 313 8 / 133 image of the compression network that is operable to perform prediction corresponding to image reconstruction, according to some examples in this disclosure.

[0032] Figure 19 is a diagram of an illustrative aspect of an example of a motion estimation encoder portion of a compression network that is operable to perform matching prediction for low-latency image reconstruction, according to some examples in this disclosure.

[0033] Figure 20 is a diagram of an illustrative aspect of an example of a motion estimation decoder portion of the compression network that is operable to perform matching prediction for low-latency image reconstruction, according to some examples in this disclosure.

[0034] Figure 21 is a diagram of an illustrative aspect of an example of an image estimator of a compression network that is operable to perform prediction corresponding to low-latency image reconstruction, according to some examples in this disclosure.

[0035] Figure 22 is a diagram of an illustrative aspect of an example of an image encoder portion of the compression network that is operable to perform matching prediction for low-latency image reconstruction, according to some examples in this disclosure.

[0036] Figure 23 is a diagram of an illustrative aspect of an example of an image decoder portion of the compression network that is operable to perform prediction matching low-resolution image reconstruction. Petition 870250078561, dated 03 / 09 / 2025, page 14 / 313 9 / 133 latency, according to some examples in this disclosure.

[0037] Figure 24 illustrates an example of an operable integrated circuit for performing prediction using a compression network, according to some examples in this disclosure.

[0038] Figure 25 is a diagram of an operable mobile device for performing prediction using a compression network, according to some examples in this disclosure.

[0039] Figure 26 is a diagram of an operable headset for performing prediction using a compression network, according to some examples in this disclosure.

[0040] Figure 27 is a diagram of an operable wearable electronic device for performing prediction using a compression network, according to some examples in this disclosure.

[0041] Figure 28 is a diagram of a voice-controlled loudspeaker system operable to perform prediction using a compression network, according to some examples in this disclosure.

[0042] Figure 29 is a diagram of an operable camera for performing prediction using a compression network, according to some examples in this disclosure.

[0043] Figure 30 is a diagram of a headset, such as a virtual reality, mixed reality, or augmented reality headset, operable to perform prediction using a compression network, according to some examples in this disclosure.

[0044] Figure 31 is a diagram of a first Petition 870250078561, dated 03 / 09 / 2025, page 15 / 313 10 / 133 Example of an operable vehicle for performing prediction using a compression network, according to some examples in this disclosure.

[0045] Figure 32 is a diagram of a second example of an operable vehicle for performing prediction using a compression network, according to some examples in this disclosure.

[0046] Figure 33 is a diagram of a particular implementation of a method for generating encoded data using the compression network of Figure 1, according to some examples in this disclosure.

[0047] Figure 34 is a diagram of a particular implementation of a method for generating predicted data from encoded data using the compression network of Figure 1, according to some examples in this disclosure.

[0048] Figure 35 is a block diagram of a particular illustrative example of a device that is operable to perform prediction using a compression network, according to some examples in this disclosure. VI. Detailed Description

[0049] Computing devices often incorporate functionalities for processing large amounts of data. Compressing data before storage or transmission can conserve resources such as memory and bandwidth. For example, a computing device can generate an encoded version of an image frame that uses fewer bits than the original image frame. Techniques that reduce the size of compressed data can further conserve resources. Compressed data Petition 870250078561, dated 03 / 09 / 2025, page 16 / 313 11 / 133 can be processed to generate predicted data. For example, the predicted data may correspond to a reconstructed version of the image frame, a predicted future image frame in a sequence of images that includes the image frame, a classification of the image frame, other types of data associated with the image frame, or a combination thereof.

[0050] Systems and methods for performing prediction using a compression network are disclosed. For example, a compression network includes an encoder portion and a decoder portion. The encoder portion is configured to process an input value from a sequence of input values ​​and conditional encoder input to generate encoded data. The encoded data corresponds to a compressed version of (has a smaller number of bits than) the input value. The decoder portion is configured to process the encoded data and conditional decoder input to generate a predicted value associated with the input value.

[0051] The encoder conditional input is an estimate of the decoder conditional input. For example, the decoder conditional input is based on a previously predicted value generated in the decoder portion, a local decoder portion in the encoder portion is used to generate an estimate of the previously predicted value, and the encoder conditional input is based on the estimate of the previously predicted value. Generating the encoded data based on an estimate of information (e.g., the previously predicted value) that is available in the decoder portion can reduce the size of the information (e.g., the encoded data) that must be provided to Petition 870250078561, dated 03 / 09 / 2025, page 17 / 313 12 / 133 portion of the decoder to generate the predicted value.

[0052] Particular aspects of the present disclosure are described below with reference to the drawings. In the description, common attributes are referred to as common reference numbers. As used in the present invention, various terminologies are used for the purpose of describing only particular implementations, and are not intended to be limiting of the implementations. For example, the singular forms a, an, and are intended to include plural forms as well, unless the context clearly indicates otherwise. Furthermore, some attributes described in the present invention are singular in some implementations and plural in other implementations. To illustrate, Figure 2 depicts a device 202 that includes one or more processors (processor(s) 290 in Figure 2), indicating that in some implementations, the device 202 includes a single processor 290 and in other implementations, the device 202 includes multiple processors 290.For ease of reference in the present invention, such attributes are generally introduced as one or more attributes, and are subsequently referred to in the singular, unless aspects related to several of the attributes are being described.

[0053] In some designs, multiple instances of a particular type of attribute are used. Although these attributes are physically and / or logically distinct, the same reference number is used for each, and the different instances are distinguished by adding a letter to the reference number. When the attributes are referred to in the present invention as a group or a type (for example, when none of the attributes are being specifically referred to), the number Petition 870250078561, dated 03 / 09 / 2025, p. 18 / 313 The reference number 13 / 133 is used without a distinguishing letter. However, when reference is made in the present invention to a particular attribute of multiple attributes of the same type, the reference number is used with the distinguishing letter. For example, with reference to Figure 1, multiple input values ​​are illustrated and associated with the reference numbers 105A, 105B, and 105C. When reference is made to a particular input value of these input values, such as input value 105A, the distinguishing letter A is used. However, when reference is made arbitrarily to any of these input values ​​or to these input values ​​as a group, the reference number 105 is used without a distinguishing letter.

[0054] As used in the present invention, the terms comprise, includes, and comprising can be used interchangeably with include, includes, or including. Additionally, the term in which can be used interchangeably with where. As used in the present invention, exemplifier indicates an example, an implementation, and / or an aspect, and should not be interpreted as limiting or as indicating a preference or a preferred implementation. As used in the present invention, an ordinal term (e.g., first, second, third, etc.) used to modify an element, such as a structure, a component, an operation, etc., does not, by itself, indicate any priority or order of the element in relation to another element, but simply distinguishes the element from another element that has the same name (but for use of the ordinal term).As used in the present invention, the term "set" refers to one or more of a particular element, and the term "plurality" refers to multiples (e.g., two). Petition 870250078561, dated 03 / 09 / 2025, p. 19 / 313 14 / 133 or more) of a particular element.

[0055] As used in the present invention, coupled may include communicatively coupled, electrically coupled, or physically coupled, and may also (or alternatively) include any combination thereof. Two devices (or components) may be coupled (e.g., communicatively coupled, electrically coupled, or physically coupled), directly or indirectly, via one or more other devices, components, wires, buses, networks (e.g., a wired network, a wireless network, or a combination thereof), etc. Two devices (or components) that are electrically coupled may be included in the same device or in different devices, and may be connected via electronic circuits, one or more connectors, or inductive coupling, as illustrative and non-limiting examples.In some implementations, two devices (or components) that are communicatively coupled, as in an electrical communication, can send and receive signals (e.g., digital signals or analog signals) directly or indirectly, via one or more wires, buses, networks, etc. As used in the present invention, directly coupled may include two devices that are coupled (e.g., communicatively coupled, electrically coupled, or physically coupled) without intervening components.

[0056] In this disclosure, terms such as determine, calculate, estimate, move, adjust, etc., may be used to describe how one or more operations are performed. It should be noted that such terms should not be interpreted as limiting, and other Petition 870250078561, dated 03 / 09 / 2025, page 20 / 313 15 / 133 techniques can be used to perform similar operations. Additionally, as referred to in the present invention, generating, calculating, estimating, using, selecting, accessing, and determining can be used interchangeably. For example, generating, calculating, estimating, or determining a parameter (or signal) can refer to actively generating, estimating, calculating, or determining the parameter (or signal), or it can refer to using, selecting, or accessing the parameter (or signal) that has already been generated, such as by another component or device.

[0057] With reference to Figure 1, a particular illustrative aspect of a compression network configured to perform prediction is disclosed and generically designated 140. The compression network 140 includes an encoder portion 160 that is configured to be coupled to a decoder portion 180.

[0058] In some respects, the 160 encoder portion is included in a first device that is different from a second device that includes the 180 decoder portion, as further described with reference to Figure 2. To illustrate, in these respects, the 140 compression network can be used to reduce the resource usage (e.g., memory, bandwidth, transmission time, etc.) associated with data transmission from the first device to the second device. In other respects, the 160 encoder portion and the 180 decoder portion are included in a single device, as further described with reference to Figure 3. To illustrate, in these respects, the 140 compression network can be used to reduce resource usage (e.g., Petition 870250078561, dated 03 / 09 / 2025, page 21 / 313 16 / 133 memory) associated with data storage for access by the device.

[0059] The encoder portion 160 is configured to process one or more input values ​​105 to generate one or more sets of encoded data 165. In one example, the one or more input values ​​105 include an input value 105A, an input value 105B, and an input value 105C, one or more additional input values, or a combination thereof. The encoder portion 160 is configured to process the input value 105A to generate encoded data 165A, process the input value 105B to generate encoded data 165B, process the input value 105C to generate encoded data 165C, and so on.

[0060] The encoder portion 160 includes a conditional input generator 162 coupled via an attribute generator 164 to an encoder 166. In some examples, the encoder 166 is configured to process some of the one or more input values ​​105 (e.g., key values) to generate corresponding encoded data independently of other input values ​​105. In one example, input value 105A corresponds to an image keyframe (e.g., an intraframe or I-frame) and the encoder 166 processes input value 105A to generate encoded data 165A independently of other input values ​​105. In some examples, the encoder 166 is configured to process some of the one or more input values ​​105 based on at least one other input value 105 (e.g., non-key values) to generate corresponding encoded data. In one example, the input value is 105B. Petition 870250078561, dated 03 / 09 / 2025, page 22 / 313 17 / 133 corresponds to a non-key picture frame, such as a predicted frame (P-frame) or a bidirectional frame (B-frame), and encoder 166 processes input value 105B based on encoded data 165A (corresponding to input value 105A) to generate encoded data 165B. To illustrate, conditional input generator 162 is configured to process encoded data 165A to generate conditional input 167B, attribute generator 164 is configured to process conditional input 167B to generate attribute data 163B, and encoder 166 is configured to process input value 105B and attribute data 163B to generate encoded data 165B.

[0061] The decoder portion 180 is configured to process one or more sets of encoded data 165 to generate one or more predicted values ​​195. In one example, the decoder portion 180 is configured to process encoded data 165A to generate a predicted value 195A, process encoded data 165B to generate a predicted value 195B, process encoded data 165C to generate a predicted value 195C, and so on.

[0062] The decoder portion 180 includes a conditional input generator 182 coupled via an attribute generator 184 to a decoder 186. In some examples, the decoder 186 is configured to process some of the encoded data sets 165 to generate corresponding predicted values ​​independently of other encoded data sets 165. In one example, the decoder 186 processes the encoded data 165A to generate the predicted value 195A independently of other encoded data sets 165. In some examples, the decoder 186 is Petition 870250078561, dated 03 / 09 / 2025, page 23 / 313 18 / 133 is configured to process some of the 165-encoded datasets based on at least one other 165-encoded dataset to generate corresponding predicted values. For example, decoder 186 processes 165B-encoded data based on a predicted value 195A (corresponding to 165A-encoded data) to generate the predicted value 195B. To illustrate, conditional input generator 182 is configured to process predicted value 195A to generate conditional input 187B, attribute generator 184 is configured to process conditional input 187B to generate attribute data 183B, and decoder 186 is configured to process both 165B-encoded data and attribute data 183B to generate the predicted value 195B.

[0063] The conditional input 167B in the encoder portion 160 corresponds to an estimate of the conditional input 187B available in the decoder portion 180. A technical advantage of generating the encoded data 165B based on an estimate of information (e.g., the conditional input 187B) that is available in the decoder portion 180 may include reducing the size of the information (e.g., the encoded data 165B) that must be provided to the decoder portion 180 to generate the predicted value 195B.

[0064] In some implementations, a compression network component (for example, the encoder portion 160, the decoder portion 180, or both) of the compression network 140 corresponds to, or is included in, one of several device types. In one illustrative example, the compression network component is integrated into a headset device, as further described with reference to Figure 26. In other examples, the component of Petition 870250078561, dated 03 / 09 / 2025, page 24 / 313 19 / 133 The compression network is integrated into at least one of the following: a mobile phone or tablet computer, as described with reference to Figure 25; a wearable electronic device, as described with reference to Figure 27; a voice-controlled speaker system, as described with reference to Figure 28; a camera device, as described with reference to Figure 29; or a virtual reality, mixed reality, or augmented reality headset, as described with reference to Figure 30. In another illustrative example, the compression network component is integrated into a vehicle, as further described with reference to Figure 31 and Figure 32.

[0065] During operation, the encoder portion 160 obtains one or more input values ​​105. In some implementations, the one or more input values ​​105 are based on the output of one or more sensors, as further described with reference to Figure 2. As illustrative, non-limiting examples, the one or more sensors may include an image sensor, an inertial measurement unit (IMU), a motion sensor, a temperature sensor, another type of sensor, or a combination thereof. In some implementations, the encoder portion 160 obtains the one or more input values ​​105 from a storage device, a processor, another device, or a combination thereof.

[0066] In some implementations, one or more 105 input values ​​indicate movement, as further described with reference to Figure 14. For example, the 105A input value includes a first image frame captured by a vehicle camera and the value Petition 870250078561, dated 03 / 09 / 2025, page 25 / 313 Input 105B includes a second image frame captured by the camera. A difference between the first image frame and the second image frame may indicate the speed and direction of movement of the vehicle. The one or more input 105 values ​​that include image frames indicating movement are provided as an illustrative example; in other examples, the one or more input 105 values ​​may include other types of data indicating movement. The one or more input 105 values ​​indicating movement are provided as an illustrative example; in other examples, the one or more input 105 values ​​may correspond to data indicating other types of information.

[0067] The encoder portion 160 generates encoded data sets 165 corresponding to one or more input values ​​105. For example, encoder 166 processes input value 105A to generate encoded data 165A. In some implementations, encoder 166, based on the determination that input value 105A satisfies an independent encoding criterion, processes (i.e., encodes) input value 105A to generate encoded data 165A independently of other one or more input values ​​105.In one example, encoder 166 determines that the independent encoding criterion is satisfied based on the determination that the input value 105A is an initial value of one or more input values ​​105, that the input value 105A corresponds to a key value (e.g., an I-frame), that at least one input value count threshold has been encoded since a more recent independently encoded input value, and that there is a difference between the input value 105A and a previous input value of one. Petition 870250078561, dated 03 / 09 / 2025, page 26 / 313 21 / 133 or more input values ​​of 105 is greater than a difference threshold (for example, because of a scene change in image frames), or a combination thereof.

[0068] As another example, encoder 166 processes input value 105B to generate encoded data 165B. In some implementations, encoder 166, in response to the determination that input value 105B does not satisfy the independent encoding criterion, processes input value 105B based, at least in part, on encoded data 165A from an input value 105A. In some respects, encoder 166 selects encoded data 165A to process input value 105B based on the determination that input value 105A corresponds to a value (e.g., a key value) of one or more input values ​​105 that is most recently independently encoded to generate encoded data 165A.In some respects, the 166 encoder selects the 165A encoded data to process the 105B input value based on determining whether the 105B input value matches a nearby value among the one or more 105 input values ​​following the 105A input value. In these respects, the 105A input value may match a key value that is independently encoded to generate the 165A encoded data or a non-key value that is encoded based on at least one other of the one or more 105 input values.

[0069] Encoder portion 160, in response to the determination that input value 105B does not satisfy the independent encoding criterion, provides encoded data 165A to conditional input generator 162 to obtain conditional input 167B from the compression network. Petition 870250078561, dated 03 / 09 / 2025, page 27 / 313 22 / 133 140. The conditional input generator 162 includes a local decoder portion 168, an estimator 170, or both. In some respects, the local decoder portion 168 is configured to perform operations similar to those of the decoder portion 180. For example, the local decoder portion 168 performs one or more operations described in the present invention with reference to the decoder portion 180 to process the encoded data 165A to generate a predicted value 169A. The predicted value 169A corresponds to an estimate of a predicted value 195A that can be generated in the decoder portion 180 by processing the encoded data 165A.

[0070] Optionally, in some implementations, estimator 170 processes predicted value 169A to generate an estimated value 171B. In some respects, the estimated value 171B corresponds to an estimate of the input value 105B that can be generated in the decoder portion 180 based on predicted value 195A. In some respects, the closer the estimated value 171B is to the input value 105B, the less information needs to be provided to the decoder portion 180, such as the encoded data 165B.

[0071] Conditional input 167B is based on predicted value 169A, estimated value 171B, or both. In some implementations, conditional input 167B may be based on one or more additional predicted values, one or more additional estimated values, or a combination thereof, associated with one or more other input values ​​105.

[0072] The encoder portion 160, in response to the determination that the input value 105B does not satisfy Petition 870250078561, dated 03 / 09 / 2025, page 28 / 313 23 / 133 The independent encoding criterion uses compression network 140 to process input value 105B and conditional input 167B to generate encoded data 165B associated with input value 105B. For example, attribute generator 164 processes conditional input 167B to generate attribute data 163B, as further described with reference to Figure 4. In some implementations, attribute data 163B corresponds to multiscale attribute data with different resolutions (e.g., different spatial resolutions). In some implementations, attribute data 163B includes multiscale wavelet transform data. Encoder 166 processes input value 105B and attribute data 163B to generate encoded data 165B, as further described with reference to Figure 4.

[0073] The decoder portion 180 obtains one or more of the encoded data sets 165 associated with one or more input values ​​105. In some implementations, the encoder portion 160 in a first device provides the encoded data 165 via a bitstream to the decoder portion 180 in a second device, as further described with reference to Figure 2. In other implementations, the encoder portion 160 stores the encoded data 165 in a storage device, and the decoder portion 180 retrieves the encoded data 165 from the storage device, as further described with reference to Figure 3.

[0074] The decoder portion 180 generates one or more predicted values ​​195 corresponding to the encoded datasets 165. For example, decoder 186 Petition 870250078561, dated 03 / 09 / 2025, page 29 / 313 24 / 133 processes the 165A encoded data to generate a predicted 195A value. In some implementations, the 186 decoder, based on the determination that the 165A encoded data satisfies an independent decoding criterion, processes (e.g., decodes) the 165A encoded data to generate the predicted 195A value independently of one or more other predicted 195 values. In one example, the 186 decoder determines that the independent encoding criterion is satisfied based on the determination that the metadata associated with the 165 encoded data indicates that the 165 encoded data corresponds to a key value (e.g., an I-frame), that the 165 encoded data should be decoded independently, or both.

[0075] As another example, the 186 decoder processes the 165B encoded data to generate the predicted value 195B. In some implementations, the 186 decoder, in response to the determination that the 165B encoded data does not satisfy the independent decoding criterion, processes the 165B encoded data based, at least in part, on a predicted value 195A corresponding to the 165A encoded data that is associated with an input value 105A. In some respects, the 186 decoder selects the predicted value 195A to process the 165B encoded data based on the determination that the predicted value 195A corresponds to a value (e.g., a key value) that is most recently independently decoded. In some respects, the 186 decoder selects the predicted value 195A to process the encoded data 165B based on the determination that the predicted value 195B corresponds to a nearby value in the one or more predicted values ​​195 following the predicted value 195A. Petition 870250078561, dated 03 / 09 / 2025, page 30 / 313 25 / 133 In these aspects, the predicted value 195A may correspond to a key value that is independently decoded or a non-key value that is decoded based on at least one other of the one or more predicted values ​​195.

[0076] The decoder portion 180, in response to the determination that the encoded data 165B does not satisfy the independent decoding criterion, provides the predicted value 195A to the conditional input generator 182 to obtain the conditional input 187B from the compression network 140. In some implementations, the conditional input generator 182 outputs the predicted value 195A as the conditional input 187B. Optionally, in some implementations, the conditional input generator 182 includes an estimator 188 that processes the predicted value 195A to generate an estimated value 193B. In some respects, the estimated value 193B corresponds to an estimate of the input value 105B based on the predicted value 195A. In some respects, the closer the estimated value 193B is to the input value 105B, the less information needs to be processed by the decoder portion 180 as the encoded data 165B to generate the predicted value 195B.

[0077] Conditional input 187B is based on predicted value 195A, estimated value 193B, or both. In some implementations, conditional input 187B may be based on at least one additional predicted value 195, one or more additional estimated values, or a combination thereof.

[0078] The decoder portion 180, in response to the determination that the encoded data 165B does not satisfy the independent decoding criterion, uses the compression network 140 to process the encoded data. Petition 870250078561, dated 03 / 09 / 2025, page 31 / 313 26 / 133 165B and conditional input 187B to generate the predicted value 195B associated with the input value 105B. For example, attribute generator 184 processes conditional input 187B to generate attribute data 183b, as further described with reference to Figure 5. In some implementations, attribute data 183b corresponds to multiscale attribute data with different resolutions (e.g., different spatial resolutions). In some implementations, attribute data 183b includes multiscale wavelet transform data. Decoder 186 processes encoded data 165B and attribute data 183B to generate the predicted value 195B, as further described with reference to Figure 5.

[0079] In some examples, the predicted value 195B corresponds to a reconstructed version of the input value 105B. In some examples, the predicted value 195B corresponds to a predicted future value, such as a prediction of a future value of one or more input values ​​105. To illustrate, the future value might be an input value 105C that is not yet available in the encoder portion 160 at the time the encoded data 165B is generated. In some examples, the predicted value 195B corresponds to a classification associated with the input value 105B. To illustrate, the predicted value 195B might indicate whether the input value 105B corresponds to an alert condition.

[0080] In some examples, the predicted value 195B corresponds to a detection result associated with the input value 105B. To illustrate, the predicted value 195B indicates whether a face is detected in an image frame associated with the input value 105B. In some examples, the predicted value Petition 870250078561, dated 03 / 09 / 2025, page 32 / 313 27 / 133 195B corresponds to a collision avoidance output. For example, the input value 105B indicates a vehicle's first position relative to an object's second position. In some respects, the predicted value 195B indicates a predicted future first position of the vehicle relative to a predicted future second position of the object. In some respects, the predicted value 195B indicates whether the predicted future first position of the vehicle is within a collision threshold of the object's future second position. In some examples, one or more predicted values ​​195 are processed by one or more downstream applications, and the compression network 140 is trained (e.g., configured) based on a performance metric associated with a downstream application. For example, the compression network 140 is trained to reduce a loss metric associated with the downstream application.

[0081] A technical advantage of using the compression network 140 to generate the one or more predicted values ​​195 may include reduced resource usage. For example, generating the encoded data 165B based on the conditional input 167B as an estimate of information (e.g., the conditional input 187B) available in the decoder portion 180 reduces the information that must be provided as the encoded data 165B to the decoder portion 180 to maintain the accuracy of the predicted value 195B.

[0082] With reference to Figure 2, an illustrative aspect of an operable system for performing prediction using a compression network is shown and generally designated as 200. System 200 includes a device 202 configured to be coupled to, or include, one or more sensors 240. Device 202 is configured to be Petition 870250078561, dated 03 / 09 / 2025, page 33 / 313 28 / 133 communicate with a device 260. For example, device 202 is configured to be coupled via a network (e.g., a wired network, a wireless network, or both) to device 260.

[0083] The encoder portion 160 is included in one or more processors 290 of device 202, and the decoder portion 180 is included in one or more processors 292 of device 260. The one or more processors 290 are coupled to one or more sensors 240 and to the modem 270. The one or more processors 292 are coupled to the modem 280. Optionally, in some implementations, the one or more processors 292 include one or more applications 262. Optionally, in some implementations, device 260 is configured to be coupled to a device 264.

[0084] During operation, the encoder portion 160 receives sensor data 226, such as one or more input values ​​105, from one or more sensors 240. As illustrative, non-limiting examples, the one or more sensors 240 may include an image sensor, an IMU, a motion sensor, an accelerometer, a speedometer, a gyroscope, a radar, a temperature sensor, a microphone, another type of sensor, or a combination thereof. Encoder portion 160 processes one or more input values ​​105 (e.g., sensor data 226) to generate encoded data sets 165, as described with reference to Figure 1. Encoder portion 160 provides the encoded data sets 165 to modem 270 for transmission as a bitstream 235 to device 260. Modem 280 receives the bitstream 235 and provides the encoded data sets 165 to decoder portion 180. Petition 870250078561, dated 03 / 09 / 2025, page 34 / 313 29 / 133

[0085] The decoder portion 180 processes the encoded datasets 165 to generate one or more predicted values ​​195, as described with reference to Figure 1. One or more processors 292 generate an output 295 based on the one or more predicted values ​​195. Optionally, in some implementations, the decoder portion 180 provides the one or more predicted values ​​195 to one or more applications 262, and one or more applications 262 generate an output 263 based on the one or more predicted values ​​195. For example, the one or more predicted values ​​195 may correspond to image frames, and an application 262 processes the predicted value 195B corresponding to an image frame to generate the output 263 indicating a classification associated with the image frame. Output 295 is based on output 263, on one or more predicted values ​​of 195, or on a combination thereof.

[0086] In some implementations, the encoder portion 160, the decoder portion 180, or both are trained (e.g., configured) based on a performance metric associated with one or more applications 262. For example, a network trainer trains (e.g., configures network weights and biases of) the encoder portion 160, the decoder portion 180, or both, based on a loss metric associated with the classification output generated by an application 262. In some implementations, one or more processors 292 provide the output 295 to the device 264. The device 264 may include a display device, a network device, a storage device, a user device, or a combination thereof.

[0087] In some respects, output 295 initiates one or more operations on device 264. For example Petition 870250078561, dated 03 / 09 / 2025, page 35 / 313 30 / 133 illustrative, device 264 and device 202 are the same device (e.g., a vehicle), and the one or more predicted values ​​195 correspond to collision avoidance outputs. Device 260, in response to the determination that predicted value 195B indicates that a first predicted future location of device 260 is expected to be within a threshold distance of a second predicted future location of an object, sends output 295 to initiate one or more collision avoidance operations (e.g., braking) on ​​device 202. In some implementations, device 260 is the same as or included within device 202 (e.g., the vehicle). In other implementations, device 260 is external to device 202 and generates output 295 based on one or more predicted values ​​195, based on one or more additional predicted values ​​(for example, associated with the object, another vehicle, or both) or a combination thereof.

[0088] A technical advantage of using the encoder portion 160 to generate the encoded data 165 based on an estimate of information that is available in the decoder portion 180 may include reduced resource usage (e.g., memory, bandwidth, and transmission time) associated with transmitting bitstream 235 to device 260.

[0089] With reference to Figure 3, an illustrative aspect of an operable system for performing prediction using a compression network is shown and generally designated as 300. System 300 includes a device 302 that is configured to couple or include one or more sensors 240. Device 302 is configured to couple or include a storage device 392. Petition 870250078561, dated 03 / 09 / 2025, page 36 / 313 31 / 133

[0090] The encoder portion 160 and the decoder portion 180 are included in one or more processors 390 of the device 302. The encoder portion 160 is configured to store one or more of the encoded data sets 165 in the storage device 392. The decoder portion 180 is configured to retrieve one or more of the encoded data sets 165 from the storage device 392. Storing one or more of the encoded data sets 165 in the storage device 392 may use less memory than storing the corresponding values ​​of the sensor data 226 in the storage device 392.

[0091] The decoder portion 180 processes the encoded data sets 165 to generate one or more predicted values ​​195, as described with reference to Figure 1. The one or more processors 390 generate the output 295 based on the one or more predicted values ​​195. Optionally, in some implementations, the decoder portion 180 provides the one or more predicted values ​​195 to the one or more applications 262, and the one or more applications 262 generate an output 263 based on the one or more predicted values ​​195. The output 295 is based on the output 263, the one or more predicted values ​​195, or a combination thereof. In some implementations, the one or more processors 390 provide the output 295 to the device 264.

[0092] In an illustrative example, device 264 and device 302 are the same device (e.g., a vehicle), and the one or more predicted values ​​195 correspond to collision avoidance outputs. Device 302, in response to the determination that the predicted value 195B indicates that a first predicted future location of the is expected Petition 870250078561, dated 03 / 09 / 2025, page 37 / 313 32 / 133 device 302 is within a threshold distance of a second predicted future location of an object, generates output 295 to initiate one or more collision avoidance operations (e.g., braking) on ​​device 302.

[0093] With reference to Figure 4, an example 400 of an encoder portion 160 of a compression network 140 is shown that is operable to perform prediction. For example, the encoder portion 160 is configured to generate encoded data 165B which can be processed in a decoder portion 180 of the compression network 140 to generate a predicted value 195B, as further described with reference to Figure 5.

[0094] The 140 compression network includes a multi-layered neural network. For example, the 164 attribute generator includes one or more 404 attribute layers coupled to one or more 402 encoder layers of the 166 encoder. The one or more 404 attribute layers include a 404A attribute layer, a 404B attribute layer, a 404C attribute layer, one or more additional attribute layers including a 404N attribute layer, or a combination thereof. The one or more 402 encoder layers include a 402A encoder layer, a 402B encoder layer, a 402C encoder layer, one or more additional encoder layers including a 402N encoder layer, or a combination thereof. An output from each 404 attribute layer is coupled to an input from a corresponding 402 encoder layer.For example, an output from attribute layer 404A is coupled to an input from encoder layer 402A, and an output from attribute layer 404B is coupled to an input from layer 402A. Petition 870250078561, dated 03 / 09 / 2025, page 38 / 313 33 / 133 of encoder 402B, and so on.

[0095] An output from each preceding attribute layer 404 is coupled to an input of a subsequent attribute layer 404. For example, an output from attribute layer 404A is coupled to an input of attribute layer 404B, an output from attribute layer 404B is coupled to an input of attribute layer 404C, and so on. An output from each preceding encoder layer 402 is coupled to an input of a subsequent encoder layer 402. For example, an output from encoder layer 402A is coupled to an input of encoder layer 402B, an output from encoder layer 402B is coupled to an input of encoder layer 402C, and so on.

[0096] In some implementations, the 166 encoder is configured to encode multiple resolution orders of an input value 105 to generate 165-encoded data. In an illustrative example, the 166 encoder corresponds to a video encoder that is configured to encode multiple spatial resolution orders of the input value 105 (e.g., an image frame). The one or more 404 attribute layers and the one or more 402 encoder layers correspond to network layers that are associated with multiple resolutions. For example, attribute layer 404A and encoder layer 402A are associated with a first resolution, attribute layer 404B and encoder layer 402B are associated with a second resolution, attribute layer 404C and encoder layer 402C are associated with a third resolution, attribute layer 404N and encoder layer 402N are associated with an nth resolution, or a combination thereof. Petition 870250078561, dated 03 / 09 / 2025, pp. 39 / 313 34 / 133

[0097] During operation, encoder portion 160 provides conditional input 167B to attribute generator 164 and input value 105B (xb) to encoder 166, as described with reference to Figure 1. Attribute layer 404A processes conditional input 167B to generate attribute data 163BA associated with the first resolution and provides attribute data 163BA to encoder layer 402A. Encoder layer 402A processes input value 105B and attribute data 163BA to generate an output (associated with the first resolution) that is provided to encoder layer 402B.

[0098] Each subsequent attribute layer 404 processes an output from the previous attribute layer 404 to generate attribute data 163 and provides the attribute data 163 to a corresponding encoder layer 402. For example, attribute layer 404B provides attribute data 163BB associated with the second resolution for encoder layer 402B, attribute layer 404C provides attribute data 163BC associated with the third resolution for encoder layer 402C, attribute layer 404N provides attribute data 163BN associated with the nth resolution for encoder layer 402N, and so on. Attribute data 163B (for example, attribute data 163BA, attribute data 163BB, attribute data 163BC, attribute data 163BN, or a combination thereof) therefore correspond to multi-scale attribute data with different resolutions.

[0099] Each subsequent 402 encoder layer processes an output from the previous 402 encoder layer and attribute data 163 from an attribute layer. Petition 870250078561, dated 03 / 09 / 2025, page 40 / 313 35 / 133 corresponding to generate an output. For example, encoder layer 402B processes the output from encoder layer 402A and attribute data 163BB to generate an output (associated with the second resolution) that is provided to encoder layer 402C, and so on. The output from encoder layer 402N corresponds to encoded data 165B.

[00100] Optionally, in some implementations, attribute generator 164 may use various techniques to generate attribute data 163B. For example, attribute generator 164 may perform wavelet transforms to generate attribute data 163B corresponding to multiscale wavelet transform data. To illustrate, attribute generator 164 may perform a wavelet transform based on conditional input 167B to generate attribute data 163BA (e.g., first wavelet transform data) associated with the first resolution. Attribute generator 164 may perform a wavelet transform based on attribute data 163BA, conditional input 167B, or both to generate attribute data 163BB (e.g., second wavelet transform data) associated with the second resolution.Similarly, the attribute generator 164 can generate the attribute data 163BC (e.g., third wavelet transform data) associated with the third resolution, and so on.

[00101] In one particular aspect, conditional input 167B corresponds to an estimate of conditional input 187B that can be generated in the decoder portion 180 of Figure 5. A technical advantage of using the estimate (e.g., conditional input 167B) of Petition 870250078561, dated 03 / 09 / 2025, p. 41 / 313 36 / 133 information available in decoder portion 180 to generate attribute data 163B which is used to generate encoded data 165B may include a reduced size of encoded data 165B to maintain the accuracy of predicted value 195B generated in decoder portion 180.

[00102] With reference to Figure 5, an example 500 of a decoder portion 180 of a compression network 140 is shown that is operable to perform prediction. For example, the decoder portion 180 is configured to process the encoded data 165B (generated in the encoder portion 160 as described with reference to Figure 4) to generate the predicted value 195B (y) associated with the input value 105B (xb).

[00103] The compression network 140 includes a multi-layered neural network. For example, the attribute generator 184 includes one or more attribute layers 504 coupled to one or more decoder layers 508 of the decoder 186. The one or more attribute layers 504 include an attribute layer 504A, an attribute layer 504B, an attribute layer 504C, one or more additional attribute layers including an attribute layer 504N, or a combination thereof. The one or more decoder layers 508 include a decoder layer 508A, a decoder layer 508B, a decoder layer 508C, one or more additional decoder layers including a decoder layer 508N, or a combination thereof. Each 504 attribute layer output is coupled to a corresponding 508 decoder layer input. For example, a 504A attribute layer output is coupled to a 508A decoder layer input, a Petition 870250078561, dated 03 / 09 / 2025, page 42 / 313 37 / 133 The output of attribute layer 504B is coupled to an input of decoder layer 508B, and so on.

[00104] An output from each preceding 504 attribute layer is coupled to an input of a subsequent 504 attribute layer. For example, an output from attribute layer 504A is coupled to an input from attribute layer 504B, an output from attribute layer 504B is coupled to an input from attribute layer 504C, and so on. An output from each preceding 508 decoder layer is coupled to an input from a subsequent 508 decoder layer. The one or more 508 decoder layers are ordered from a reference number ending with a higher alphabet to a lower alphabet. For example, decoder layer 508B is subsequent to decoder layer 508C, and decoder layer 508A is subsequent to decoder layer 508B.An output from decoder layer 508B is coupled to an input from decoder layer 508A, an output from decoder layer 508C is coupled to an input from decoder layer 508B, and so on. The output of a final decoder layer 508 (e.g., decoder layer 508A) of decoder 186 corresponds to a predicted value 195 associated with an input value 105.

[00105] In some implementations, the 186 decoder is configured to decode multiple resolution orders of the 165B encoded data. In an illustrative example, the 186 decoder corresponds to multiple spatial resolution orders of the 165B encoded data associated with the input value 105 (for example, an image frame). The one or more 504 attribute layers Petition 870250078561, dated 03 / 09 / 2025, page 43 / 313 38 / 133 and the one or more 508 decoder layers correspond to network layers that are associated with multiple resolutions. For example, attribute layer 504A and encoder layer 508A are associated with a first resolution, attribute layer 504B and decoder layer 508B are associated with a second resolution, attribute layer 504C and decoder layer 508C are associated with a third resolution, attribute layer 504N and decoder layer 508N are associated with an nth resolution, or a combination thereof.

[00106] During operation, decoder portion 180 provides conditional input 187B to attribute generator 184 and encoded data 165B to decoder 186, as described with reference to Figure 1. Attribute layer 504A processes conditional input 187B to generate attribute data 183BA associated with the first resolution and provides attribute data 183BA to decoder layer 508A. Each subsequent attribute layer 504 processes an output from the previous attribute layer 504 to generate attribute data 183 and provides attribute data 183 to a corresponding decoder layer 508.For example, attribute layer 504B provides attribute data 183BB associated with the second resolution for decoder layer 508B, attribute layer 504C provides attribute data 183BC associated with the third resolution for decoder layer 508C, attribute layer 504N provides attribute data 183BN associated with the nth resolution for decoder layer 508N, and so on. Attribute data 183B (for example, Petition 870250078561, dated 03 / 09 / 2025, p. 44 / 313 39 / 133 attribute data 183BA, attribute data 183BB, attribute data 183BC, attribute data 183BN or a combination thereof) therefore correspond to multi-scale attribute data with different resolutions.

[00107] The 508N decoder layer processes the 165B encoded data and the 183BN attribute data to generate an output (e.g., associated with the nth resolution) that is provided to a subsequent 508 decoder layer. Each subsequent 508 decoder layer processes an output from the previous 508 decoder layer and 183 attribute data from a corresponding attribute layer to generate an output. For example, the 508B decoder layer processes an output from the 508C decoder layer and the 183BB attribute data to generate an output (e.g., associated with the second resolution). The 508A decoder layer processes the output from the 508B decoder layer and the 183BA attribute data to generate an output that corresponds to the predicted value 195B (y) associated with the input value 105B (xb).

[00108] Optionally, in some implementations, attribute generator 184 may use various techniques similar to those performed by attribute generator 164 in Figure 4 to generate attribute data 183B. For example, attribute generator 184 may perform wavelet transforms to generate attribute data 183B corresponding to multiscale wavelet transform data. To illustrate, attribute generator 184 may perform a wavelet transform based on conditional input 187B to generate attribute data 183BA (e.g., first wavelet transform data) associated with the first resolution. Attribute generator 184 Petition 870250078561, dated 03 / 09 / 2025, page 45 / 313 40 / 133 can perform a wavelet transform based on attribute data 183BA, conditional input 187B, or both to generate attribute data 183BB (e.g., second wavelet transform data) associated with the second resolution. Similarly, attribute generator 184 can generate attribute data 183BC (e.g., third wavelet transform data) associated with the third resolution, and so on.

[00109] Optionally, in some implementations, the decoder portion 180 may include an attribute generator 582 that is configured to process decoder information 587 that is available in the decoder portion 180 to generate attribute data 583 that is used by the decoder 186 to generate the predicted value 195B. As an example, the attribute generator 582 includes one or more attribute layers 506 coupled to one or more decoder layers 508. The one or more attribute layers 506 include an attribute layer 506A, an attribute layer 506B, an attribute layer 506C, one or more additional attribute layers including an attribute layer 506N, or a combination thereof. An output from each attribute layer 506 is coupled to an input of a corresponding decoder layer 508.For example, an output from attribute layer 506A is coupled to an input from decoder layer 508A, an output from attribute layer 506B is coupled to an input from decoder layer 508B, and so on.

[00110] An output from each preceding attribute layer 506 is coupled to an input from a subsequent attribute layer 506. For example, an output from the attribute layer Petition 870250078561, dated 03 / 09 / 2025, page 46 / 313 41 / 133 506A is coupled to an input of attribute layer 506B, an output of attribute layer 506B is coupled to an input of attribute layer 506C, and so on. In a particular aspect, the one or more attribute layers 506 correspond to network layers that are associated with multiple resolutions. For example, attribute layer 506A is associated with the first resolution, attribute layer 506B is associated with the second resolution, attribute layer 506C is associated with the third resolution, attribute layer 506N is associated with the nth resolution, or a combination thereof.

[00111] Decoder portion 180 provides decoder information 587 to attribute generator 582 in addition to (for example, simultaneously with) providing conditional input 187B to attribute generator 184 and encoded data 165B to decoder 186. Attribute layer 506A processes decoder information 587 to generate attribute data 583A associated with the first resolution and provides attribute data 583A to decoder layer 508A. Each subsequent attribute layer 506 processes an output from the previous attribute layer 506 to generate attribute data 583 and provides the attribute data 583 to a corresponding decoder layer 508. For example, attribute layer 506B processes an output from attribute layer 506A to generate attribute data 583b and provides the attribute data 583b associated with the second resolution to decoder layer 508B.Similarly, the 506C attribute layer provides 583C attribute data associated with the third resolution for the 508C decoder layer, and the 506N attribute layer provides 583N attribute data associated with... Petition 870250078561, dated 03 / 09 / 2025, p. 47 / 313 42 / 133 nth resolution for the 508N decoder layer, and so on. The 583 attribute data (e.g., 583A attribute data, 583B attribute data, 583C attribute data, 583N attribute data, or a combination thereof) therefore correspond to multi-scale attribute data with different resolutions.

[00112] The 508N decoder layer processes the 165B encoded data, the 183BN attribute data, and the 583N attribute data to generate an output (e.g., associated with the nth resolution) that is provided to a subsequent 508 decoder layer. Each subsequent 508 decoder layer processes an output from the previous 508 decoder layer, 183 attribute data from a corresponding attribute layer of the 184 attribute generator, and 583 attribute data from a corresponding attribute layer of the 582 attribute generator to generate an output. For example, the 508B decoder layer processes an output from the 508C decoder layer, the 183BB attribute data, and the 583B attribute data to generate an output (e.g., associated with the second resolution).The 508A decoder layer processes the output from the 508B decoder layer, the 183BA attribute data, and the 583A attribute data to generate an output that corresponds to the predicted value 195B (y) associated with the input value 105B (x6).

[00113] Optionally, in some implementations, the 582 attribute generator may use various techniques (e.g., similar to the techniques performed by the 184 attribute generator) to generate the 583 attribute data. For example, the 582 attribute generator may perform wavelet transforms to generate the 583 attribute data. Petition 870250078561, dated 03 / 09 / 2025, p. 48 / 313 43 / 133 corresponding to multiscale wavelet transform data.

[00114] In some implementations, the predicted value 195B(y) corresponds to a reconstructed version of the input value 105B(xft), as further described with reference to Figures 6 to 11. For example, the input value 105B(xft) may correspond to an image unit and the predicted value 195B(y) may correspond to a reconstructed version of the image unit, as further described with reference to Figures 14 to 23. In some implementations, the predicted value 195B(y) corresponds to a predicted future value of one or more input values ​​105, as further described with reference to Figures 12 to 13. In other implementations, the predicted value 195B(y) may correspond to a predicted future value that is associated with the input value 105B. In some respects, the predicted value 195B also includes a prediction of a future value of one or more input values ​​105.In other respects, the predicted value 195B does not include a prediction of a future value of the one or more input values ​​105. In one example, the compression network 140 tracks an object associated with one or more motion values ​​(e.g., the one or more input values ​​105) through one or more pixel frames. To illustrate, the input value 105B might include tracking data (e.g., associated with an image frame) and the predicted value 195B(y) might include a collision avoidance output indicating whether a future position of a vehicle relative to the object (e.g., a future position of the object) is predicted to be. Petition 870250078561, dated 03 / 09 / 2025, page 49 / 313 44 / 133 less than a threshold. The collision avoidance output (e.g., collision or no collision) is a future prediction associated with the tracking data. In some implementations, the collision avoidance output also indicates predicted future tracking data, such as a predicted future position of the vehicle relative to a predicted future position of the object.

[00115] With reference to Figure 6, an example 600 of an encoder portion 160 of a compression network 140 is shown that is operable to perform prediction corresponding to reconstruction. For example, the encoder portion 160 is configured to generate the encoded data 165B which can be processed in a decoder portion 180 of the compression network 140 to generate the predicted value 195B (xb) which corresponds to a reconstructed version of the input value 105B (xb), as further described with reference to Figure 7.

[00116] The conditional input generator 162 uses the local decoder portion 168 to process the encoded data 165A to generate the predicted value 169A, as described with reference to Figure 1. In a particular example, the predicted value 169A (xa) corresponds to an estimate of a reconstructed version of the input value 105A (xa) that can be generated in the decoder portion 180.

[00117] In some implementations, the conditional input generator 162 generates one or more additional predicted values. The one or more additional predicted values ​​may be associated with at least one input value that precedes input value 105B in the one or more input values ​​105, at least one input value that is Petition 870250078561, dated 03 / 09 / 2025, p. 50 / 313 45 / 133 subsequent to the input value 105B in one or more input values ​​105, or both.

[00118] In one example, the one or more additional predicted values ​​may include a predicted value 169C (xc) that is associated with an input value 105C that is subsequent to the input value 105B in the one or more input values ​​105 and that may be encoded as the encoded data 165C independently of (e.g., before) encoding the input value 105B. In an illustrative example, the input value 105C corresponds to a key value (e.g., an I-frame) that may be encoded independently of others among the one or more input values ​​105. The conditional input generator 162 uses the local decoder portion 168 to process the encoded data 165C to generate the predicted value 169C, process one or more additional sets of encoded data to generate one or more additional predicted values, or a combination thereof.

[00119] Optionally, in some implementations, the conditional input generator 162 uses the estimator 170 to generate an estimated value 171B (xb) of the input value 105B (xb) based on one or more predicted values ​​(e.g., predicted value 169A, predicted value 169C, one or more additional predicted values, or a combination thereof). To illustrate, the estimated value 171B (xb) corresponds to an estimate of the input value 105B (xft) that can be generated in the decoder portion 180 based on encoded datasets corresponding to one or more predicted values ​​and independently of the encoded data 165B.

[00120] In a particular implementation, the Petition 870250078561, dated 03 / 09 / 2025, page 51 / 313 46 / 133 Estimator 170 uses one or more estimation techniques to determine the estimated value 171B. For example, the predicted value 169A corresponds to the input value 105A (e.g., a previous image frame), the predicted value 169C corresponds to the input value 105C (e.g., a subsequent image frame), and the estimated value 171B corresponds to an estimated input value between the predicted value 169A and the predicted value 169C. In some implementations, estimator 170 includes a neural network that is configured to process the one or more predicted values ​​to generate the estimated value 171B.

[00121] Conditional input 167B includes the estimated value 171B (xb), one or more predicted values, such as the predicted value 169A (xa), the predicted value 169C (xc), one or more additional predicted values, or a combination thereof. Attribute generator 164 generates attribute data 163B based on conditional input 167B, and encoder 166 processes input value 105B based on attribute data 163B to generate encoded data 165B, as described with reference to Figure 4.

[00122] With reference to Figure 7, an example 700 of a decoder portion 180 of a compression network 140 is shown that is operable to perform matching prediction and reconstruction. For example, the decoder portion 180 is configured to process the encoded data 165B (generated in the encoder portion 160 as described with reference to Figure 6) to generate the predicted value 195B (xb) which corresponds to a reconstructed version of the input value 105B (xb).

[00123] The conditional input generator 182 obtains the predicted value 195A, as described with reference Petition 870250078561, dated 03 / 09 / 2025, p. 52 / 313 47 / 133 to Figure 1. For example, the decoder portion 180 processes the encoded data 165A to generate the predicted value 195A. In a particular example, the predicted value 195A (xa) corresponds to a reconstructed version of the input value 105A (xa).

[00124] In some implementations, the conditional input generator 182 obtains one or more additional predicted values. The one or more additional predicted values ​​may be associated with at least one input value that precedes input value 105B in the one or more input values ​​105, at least one input value that follows input value 105B in the one or more input values ​​105, or both.

[00125] In one example, the one or more additional predicted values ​​may include a predicted value 195C (xc) that is associated with an input value 105C that is subsequent to the input value 105B in the one or more input values ​​105, and the corresponding encoded data 165C may be decoded to generate the predicted value 195C independently of (e.g., before) decoding the encoded data 165B associated with the input value 105B. In an illustrative example, the input value 105C corresponds to a key value (e.g., an I-frame) and is associated with the encoded data 165C that may be decoded independently of other of the encoded data sets 165.

[00126] Optionally, in some implementations, the conditional input generator 182 uses the estimator 188 to generate an estimated value 193B (xb) from the input value 105B (xb) based on one or more predicted values. Petition 870250078561, dated 03 / 09 / 2025, page 53 / 313 48 / 133 (for example, predicted value 195A, predicted value 195C, one or more additional predicted values, or a combination thereof). To illustrate, the estimated value 193B (xb) corresponds to an estimate of the input value 105B (xft) that can be generated in the decoder portion 180 based on encoded datasets corresponding to one or more predicted values ​​and independently of the encoded data 165B.

[00127] In a particular implementation, estimator 188 uses one or more estimation techniques (similar to the estimation techniques performed by estimator 170 as described with reference to Figure 6) to determine the estimated value 193B. For example, the predicted value 195A corresponds to the input value 105A (e.g., a previous image frame), the predicted value 195C corresponds to the input value 105C (e.g., a subsequent image frame), and the estimated value 193B corresponds to an estimated input value between the predicted value 195A and the predicted value 195C. In some implementations, estimator 188 includes a neural network that is configured to process the one or more predicted values ​​to generate the estimated value 193B.

[00128] Conditional input 187B includes the estimated value 193B (xb), one or more predicted values, such as the predicted value 195A (xa), the predicted value 195C (xc), one or more additional predicted values, or a combination thereof. Attribute generator 184 generates attribute data 183B based on conditional input 187B, and decoder 186 processes encoded data 165B based on attribute data 183B to generate the predicted value 195B (xb), as described with Petition 870250078561, dated 03 / 09 / 2025, page 54 / 313 49 / 133 reference to Figure 5. The predicted value 195B (Xft) corresponds to a reconstructed version of the input value 105B (xft).

[00129] A technical advantage of using an estimate (e.g., conditional input 167B) of information (e.g., conditional input 187B) that is available in the decoder portion 180 to generate the encoded data 165B may include reducing the amount of information that must be encoded as the encoded data 165B to generate a reconstructed version (xft) of the input value 105B (x6).

[00130] With reference to Figure 8, an example 800 of an encoder portion 160 of a compression network 140 is shown that is operable to perform matching prediction based on auxiliary prediction. For example, the encoder portion 160 is configured to generate the encoded data 165B which can be processed in a decoder portion 180 of the compression network 140 based on auxiliary prediction data to generate the predicted value 195B (Xü) which corresponds to a reconstructed version of the input value 105B (xb), as further described with reference to Figure 9.

[00131] It should be understood that the 140 compression network that uses auxiliary prediction data to generate the predicted value 195B (Xft) corresponding to a reconstructed version of the input value 105B (xb) is provided as an illustrative example. In other examples, the 140 compression network may use auxiliary prediction data to generate various other types of predicted values ​​195, such as future predicted values, collision avoidance outputs, Petition 870250078561, dated 03 / 09 / 2025, page 55 / 313 50 / 133 classification outputs, detection outputs, etc.

[00132] The encoder portion 160 includes or is coupled to one or more auxiliary prediction layers 806. In a particular aspect, the encoder portion 160 is configured to provide domain-specific data 805 to the one or more auxiliary prediction layers 806 simultaneously with the provision of input value 105B to the encoder 166 and conditional input 167B to the attribute generator 164. The one or more auxiliary prediction layers 806 process the domain-specific data 805 to generate auxiliary prediction data 807.

[00133] Attribute generator 164 processes conditional input 167B and auxiliary prediction data 807 to generate attribute data 163B. For example, attribute layer 404A processes auxiliary prediction data 807 and conditional input 167B to generate attribute data 163BA. Attribute layer 404B processes an output from attribute layer 404A to generate attribute data 163BB, and so on.

[00134] In some implementations, the auxiliary prediction data 807 corresponds to an estimate of auxiliary prediction data available in the decoder portion 180 that can be used to assist in generating the predicted value 195B corresponding to the input value 105B.

[00135] With reference to Figure 9, an example of a 180 encoder portion of the 140 compression network is shown that is operable to perform prediction matching reconstruction based on auxiliary prediction. For example, the 180 decoder portion is configured to process the 165B encoded data (generated in the portion Petition 870250078561, dated 03 / 09 / 2025, page 56 / 313 51 / 133 of encoder 160 as described with reference to Figure 8) based on auxiliary prediction data to generate the predicted value 195B (xb) which corresponds to a reconstructed version of the input value 105B (xft).

[00136] The decoder portion 180 includes or is coupled to one or more auxiliary prediction layers 906. In a particular aspect, the encoder portion 180 is configured to provide domain-specific data 905 to the one or more auxiliary prediction layers 906 simultaneously with the provision of encoded data 165B to the decoder 186 and conditional input 187B to the attribute generator 184. The one or more auxiliary prediction layers 906 process the domain-specific data 905 to generate auxiliary prediction data 907.

[00137] Attribute generator 184 processes conditional input 187B and auxiliary prediction data 907 to generate attribute data 183B. For example, attribute layer 504A processes auxiliary prediction data 907 and conditional input 187B to generate attribute data 183BA. Attribute layer 504B processes an output from attribute layer 506A to generate attribute data 183BB, and so on.

[00138] In some implementations, the auxiliary prediction data 907 is the same as the auxiliary prediction data 807. For example, the encoder portion 160 and the decoder portion 180 have access to the same auxiliary prediction data. To illustrate, the encoder portion 160 and the decoder portion 180 can be included in the same device, as described with reference to Figure 3, to obtain the auxiliary prediction data from the same source. Petition 870250078561, dated 03 / 09 / 2025, page 57 / 313 52 / 133 or both. In some implementations, the auxiliary prediction data 907 is a predicted (e.g., reconstructed) version, a decoded version, or both, of the auxiliary prediction data 807.

[00139] In one particular aspect, auxiliary prediction data 907 can be used to assist in generating the predicted value 195B corresponding to the input value 105B. In a reconstruction example, auxiliary prediction data 907 might indicate the positions of a person's facial attributes, and the predicted value 195B corresponds to a reconstructed version of the input value 105B (e.g., an image frame) representing the person's face. Having access to the positions of facial attributes can improve the accuracy of the reconstruction. In a collision avoidance example, auxiliary prediction data 907 might indicate a predicted future path of a first vehicle, and the predicted value 195B might correspond to a collision avoidance output indicating whether, given the predicted future path of the first vehicle, a predicted future position of a second vehicle is expected to be within a distance threshold of a predicted future position of an object.

[00140] With reference to Figure 10, an example 1000 of a 160 encoder portion of a compression network is shown that is operable to perform low-latency matching prediction and reconstruction. For example, the 160 encoder portion is configured to generate the 165B encoded data corresponding to the 105B input value independently (e.g., before obtaining) of subsequent values ​​of one or more 105 input values, and the 165B encoded data can be processed in a Petition 870250078561, dated 03 / 09 / 2025, page 58 / 313 53 / 133 decoder portion 180 of the compression network 140 independently of encoded data associated with subsequent values ​​of one or more input values ​​105, as further described with reference to Figure 11.

[00141] The conditional input generator 162 in Figure 1 generates the conditional input 167B corresponding to the input value 105B independently (e.g., before obtaining) subsequent input values, including the input value 105C, from one or more input values ​​105. For example, the conditional input generator 162 uses the estimator 170 to generate the estimated value 171B based on the predicted value 169A, one or more additional predicted values ​​corresponding to one or more input values ​​before the input value 105B in one or more input values ​​105, or a combination thereof. The conditional input 167B includes the estimated value 171B, the predicted value 169A, one or more additional predicted values, or a combination thereof.Attribute generator 164 generates attribute data 163B based on conditional input 167B, and encoder 166 processes input value 105B based on attribute data 163B to generate encoded data 165B, as described with reference to Figure 4. Encoder portion 160 can thus generate encoded data 165B independently of (e.g., before) obtaining input value 105C. A technical advantage of generating encoded data 165B independently of input value 105C may include reduced latency associated with generating encoded data 165B without having to wait for access to input value 105C.

[00142] With reference to Figure 11, an example 1100 of a decoder portion 180 of the network is shown. Petition 870250078561, dated 03 / 09 / 2025, page 59 / 313 54 / 133 compression that is operable to perform prediction matching low-latency reconstruction. For example, the 180 decoder portion is configured to generate the predicted value 195B matching the input value 105B independently (e.g., before obtaining) encoded datasets matching subsequent values ​​of one or more input values ​​105.

[00143] The conditional input generator 182 generates the conditional input 187B corresponding to the input value 105B independently (e.g., before obtaining) encoded data (including the encoded data 165C) corresponding to subsequent input values ​​of one or more input values ​​105. The conditional input generator 182 obtains the predicted value 195A, as described with reference to Figure 1. For example, the decoder portion 180 processes the encoded data 165A to generate the predicted value 195A. The conditional input generator 182 uses the estimator 188 to generate the estimated value 193B based on the predicted value 195A, one or more additional predicted values ​​corresponding to one or more input values ​​before the input value 105B in the one or more input values ​​105, or a combination thereof.Conditional input 187B includes the estimated value 193B, the predicted value 195A, one or more additional predicted values, or a combination thereof.

[00144] Attribute generator 184 generates attribute data 183B based on conditional input 187B, and decoder 186 processes encoded data 165B based on attribute data 183B, as described with reference to Figure 5. Decoder portion 180 can thus generate predicted value 195B independently of (by Petition 870250078561, dated 03 / 09 / 2025, page 60 / 313 55 / 133 example, before) obtaining the 165C encoded data. A technical advantage of generating the 195B predicted value independently of the 165C encoded data may include reduced latency associated with generating the 195B predicted value without having to wait for access to the 165C encoded data.

[00145] With reference to Figure 12, an example 1200 of an encoder portion 160 of a compression network 140 is shown that is operable to perform prediction corresponding to future prediction. For example, the encoder portion 160 is configured to generate encoded data 165B corresponding to an input value 105B (xb) which can be processed in a decoder portion 180 of the compression network 140 to generate a predicted value 195C (xc) which corresponds to a predicted future value. In a particular aspect, the input value 105B (xb) corresponds to a motion value, and the predicted value 195C (xc) corresponds to a predicted future motion value (e.g., a predicted future motion vector).

[00146] The conditional input generator 162 uses the local decoder portion 168 to generate a predicted value 169B (xft). For example, the local decoder portion 168 processes the encoded data 165A associated with the input value 105A (xa) to generate the predicted value 169B corresponding to a predicted future value of the input value 105B (xft). In a particular example, the predicted value 169B (¾) corresponds to an estimate of a predicted future value that can be generated based on the encoded data 165A in the decoder portion 180.

[00147] In some implementations, the generator of Petition 870250078561, dated 03 / 09 / 2025, page 61 / 313 56 / 133 conditional input 162 generates one or more additional predicted values. The one or more additional predicted values ​​may be associated with at least one input value that precedes input value 105B in the one or more input values ​​105, at least one input value that follows input value 105B in the one or more input values ​​105, or both.

[00148] Optionally, in some implementations, the conditional input generator 162 uses the estimator 170 to generate an estimated value 171C (xc) of a predicted future value of the input value 105C (xc) based on one or more predicted values ​​(e.g., the predicted value 169B). To illustrate, the estimated value 171C (xc) corresponds to an estimate of a predicted future value of the input value 105C (xc) that can be generated in the decoder portion 180 based on encoded datasets corresponding to one or more predicted values ​​and independently of the encoded data 165B.

[00149] In a particular implementation, the 170 estimator uses one or more estimation techniques to determine the estimated value 171C. For example, the predicted value 169B corresponds to an estimate of a predicted future value of the input value 105B, one or more additional predicted values ​​correspond to estimates of predicted future values ​​of other input values ​​of the one or more input values ​​105, and the estimated value 171C corresponds to an estimated input value subsequent to the predicted value 169B. In some implementations, the 170 estimator includes a neural network that is configured to process the one or more predicted values ​​to generate the estimated value 171C. Petition 870250078561, dated 03 / 09 / 2025, p. 62 / 313 57 / 133

[00150] Conditional input 167B includes the estimated value 171C (X)), one or more predicted values, such as the predicted value 169B (¾), one or more additional predicted values, or a combination thereof. Attribute generator 164 generates attribute data 163B based on conditional input 167B, and encoder 166 processes input value 105B based on attribute data 163B to generate encoded data 165B, as described with reference to Figure 4.

[00151] With reference to Figure 13, an example 1300 of a decoder portion 180 of a compression network 140 is shown which is operable to perform prediction corresponding to future prediction. For example, the decoder portion 180 is configured to process encoded data 165B (generated in an encoder portion 160 of a compression network 140 as described with reference to Figure 12) corresponding to an input value 105B (xb) to generate a predicted value 195C (xc) which corresponds to a predicted future value.

[00152] The conditional input generator 182 obtains a predicted value 195B (x&). For example, the decoder portion 180 processes the encoded data 165A to generate the predicted value 195B (£6). In a particular example, the predicted value 195B corresponds to a predicted future value of the input value 105B (xb).

[00153] In some implementations, the conditional input generator 182 obtains one or more additional predicted values. The one or more additional predicted values ​​may be associated with at least one input value that precedes input value 105B in the one or more input values ​​105, at least one input value that is Petition 870250078561, dated 03 / 09 / 2025, p. 63 / 313 58 / 133 subsequent to the input value 105B in one or more input values ​​105, or both.

[00154] Optionally, in some implementations, the conditional input generator 182 uses the estimator 188 to generate an estimated value 193C (xc) of a predicted future value of the input value 105C (xc) based on one or more predicted values ​​(e.g., the predicted value 195B). To illustrate, the estimated value 193C (xc) corresponds to an estimate of a predicted future value of the input value 105C (xc) that can be generated in the decoder portion 180 based on encoded datasets corresponding to one or more predicted values ​​and independently of the encoded data 165B.

[00155] In a particular implementation, estimator 188 uses one or more estimation techniques (similar to the estimation techniques performed in estimator 170 as described with reference to Figure 12) to determine the estimated value 193C. For example, the predicted value 195B corresponds to an estimate of a predicted future value of the input value 105B, one or more additional predicted values ​​correspond to estimates of predicted future values ​​of other input values ​​of the one or more input values ​​105, and the estimated value 193C corresponds to an estimated input value subsequent to the predicted value 195B. In some implementations, estimator 188 includes a neural network that is configured to process the one or more predicted values ​​to generate the estimated value 193C.

[00156] Conditional entry 187B includes the estimated value 193C (xc), one or more predicted values, such as the predicted value 195B (xft), one or more additional predicted values. Petition 870250078561, dated 03 / 09 / 2025, p. 64 / 313 59 / 133 or a combination thereof. Attribute generator 184 generates attribute data 183B based on conditional input 187B and decoder 186 processes encoded data 165B based on attribute data 183B to generate predicted value 195C (xc), as described with reference to Figure 5.

[00157] A technical advantage of using the compression network 140 to generate a predicted future value may include initiating actions based on the predicted future value. For example, actions may include a preventive action, such as initiating braking, based on the determination that the predicted value 195C corresponds to an alert condition (e.g., a collision). It should be understood that the predicted value 195C corresponding to a predicted future value of one or more input values ​​105 is provided as an illustrative example. In other examples, the predicted value 195C may correspond to a predicted future value (e.g., collision or no collision) that is associated with the input value 105B (e.g., an image frame).

[00158] Figures 14 to 18 describe an example of a 140 compression network that is configured to perform image reconstruction matching prediction. The 140 compression network includes a first 140 compression network associated with motion estimation and a second 140 compression network associated with image reconstruction based on motion estimation.

[00159] Figure 14 includes an example of an encoder portion 160 of the first compression network 140 associated with motion estimation. Figure 15 includes an example of a decoder portion 180 of the first compression network 140 associated with motion estimation. A Petition 870250078561, dated 03 / 09 / 2025, page 65 / 313 60 / 133 Figure 17 includes an example of an encoder portion 160 of the second compression network 140 associated with motion estimation-based image reconstruction. Figure 18 includes an example of a decoder portion 180 of the second compression network 140 associated with motion estimation-based image reconstruction. Figure 16 includes an example of image estimation that can be performed on the encoder portion 160 and the decoder portion 180 of the second compression network 140 associated with motion estimation-based image reconstruction.

[00160] In some examples, a 160 encoder portion of the 140 compression network includes the 160 encoder portion of the first 140 compression network and the 160 encoder portion of the second 140 compression network. Similarly, the 180 decoder portion of the 140 compression network includes the 180 decoder portion of the first 140 compression network and the 180 decoder portion of the second 140 compression network.

[00161] With reference to Figure 14, an example 1400 of an encoder portion 160 of a compression network 140 is shown that is operable to perform prediction corresponding to image reconstruction. For example, the encoder portion 160 is included in the first compression network 140 associated with motion estimation.

[00162] Encoder portion 160 is configured to process one or more image units 1407 to generate encoded datasets 165. The one or more image units 1407 include an image unit 1407A, an image unit 1407B, an image unit 1407C, one or more additional image units or Petition 870250078561, dated 03 / 09 / 2025, page 66 / 313 61 / 133 a combination thereof. As illustrative, non-limiting examples, an image unit 1407 may correspond to a coding unit, a pixel block, a pixel frame, an image frame, or a combination thereof.

[00163] Encoder portion 160 is configured to determine one or more motion values ​​1405 associated with one or more picture units 1407. For example, encoder portion 160, in response to the determination that picture unit 1407B (xb) should be encoded, determines one or more motion values ​​1405 based on a comparison of picture unit 1407B with one or more other picture units of the one or more picture units 1407. To illustrate, encoder portion 160, in response to the determination that picture unit 1407B should be encoded, determines a motion value 1405A (ma^) based on a comparison of picture unit 1407A and picture unit 1407B, determines a motion value 1405B (mc^b) based on a comparison of picture unit 1407C and picture unit 1407B. The 1407B image unit determines one or more additional motion values, or a combination thereof.

[00164] In a particular aspect, the one or more motion values ​​1405 represent motion vectors associated with one or more image units 1407. For example, motion value 1405A represents motion vectors associated with image unit 1407A and image unit 1407B. In a particular aspect, the one or more motion values ​​1405 indicate one or more of linear velocity, linear acceleration, linear position, angular velocity, angular acceleration, or angular position. Petition 870250078561, dated 03 / 09 / 2025, page 67 / 313 62 / 133 associated with one or more 1407 picture units. For example, the motion value 1405A indicates one or more of linear velocity, linear acceleration, linear position, angular velocity, angular acceleration, or angular position associated with picture unit 1407A and picture unit 1407B. In a particular aspect, the compression network 140 is configured to track an object associated with one or more 1405 motion values ​​through one or more picture units. In one particular aspect, the one or more motion values ​​1405 are based on sensor data 226 received from one or more sensors 240, as described with reference to Figure 2. For example, motion value 1405A (ma-^b) is based on the first sensor data 226 associated with (e.g., captured simultaneously with) image unit 1407A and the second sensor data 226 associated with (e.g., captured simultaneously with) image unit 1407B.

[00165] The conditional input generator 162 of the encoder portion 160 generates the conditional input 167B. For example, the conditional input generator 162 uses the local decoder portion 168 to process encoded data associated with the image unit 1407A to generate a first predicted image unit and uses the local decoder portion 168 to process encoded data associated with the image unit 1407C to generate a second predicted image unit. In a particular aspect, the first predicted image unit (xa) corresponds to an estimate of a prediction of the image unit 1407A (xa) that can be generated in the decoder portion 180 of Figure 18. In a particular aspect, the second predicted image unit (xc) Petition 870250078561, dated 03 / 09 / 2025, page 68 / 313 63 / 133 corresponds to an estimate of a predicted image unit 1407C (xc) that can be generated in the decoder portion 180 of Figure 18. The conditional input generator 162 determines each of an estimated motion value 1469A (ma^c) and an estimated motion value 1469B (mc^a) based on the first predicted image unit (xa) and the second predicted image unit (xc). The conditional input 167B includes the estimated motion value 1469A (ma^c), the estimated motion value 1469B (mc^a), or both.

[00166] Attribute generator 164 processes conditional input 167B to generate attribute data 163B, as described with reference to Figure 4. Encoder 166 processes motion value 1405A (ima_^b), motion value 1405B (mc^b), image unit 1407B (xb), and attribute data 163B to generate encoded data 165B associated with image unit 1407B. For example, encoder layer 402A processes motion value 1405A (ma^b), motion value 1405B (mc^b), image unit 1407B (xft), and attribute data 163BA to generate an output. The 402B encoder layer processes an output from the 402A encoder layer and the 163BB attribute data to generate an output, and so on. An output from the 402N encoder layer corresponds to the 165B encoded data (e.g., 1465B encoded data) associated with the motion estimate.One or more motion values ​​of 1405, one or more weights, or a combination thereof, can be used to generate an estimated image unit (xft), as further described with reference to Figure 16.

[00167] With reference to Figure 15, it is shown Petition 870250078561, dated 03 / 09 / 2025, page 69 / 313 64 / 133 an example 1500 of a decoder portion 180 of a compression network 140 that is operable to perform prediction corresponding to image reconstruction. For example, decoder portion 180 is included in the first compression network associated with motion estimation.

[00168] The decoder portion 180 is configured to process encoded datasets 165 to generate one or more predicted motion values ​​1595, one or more weights, or a combination thereof. The one or more predicted motion values ​​1595, the one or more weights, or a combination thereof can be used to generate an estimated image unit (xb), as further described with reference to Figure 16.

[00169] The conditional input generator 182 of the decoder portion 180 generates the conditional input 187B. For example, the conditional input generator 182 obtains a first predicted image unit (xa) that is generated by the decoder portion 180 of the second compression network 140 processing encoded data associated with the image unit 1407A (xa), as further described with reference to Figure 18. In some respects, the conditional input generator 182 obtains a second predicted image unit (xc) that is generated by the decoder portion 180 of the second compression network 140 processing encoded data associated with the image unit 1407C (xc), as further described with reference to Figure 18. In a particular aspect, the image unit 1407A (xa), the image unit 1407C (xc), or both correspond to key units (e.g., I-frames) that can be decoded independently of the decoding of Petition 870250078561, dated 03 / 09 / 2025, page 70 / 313 65 / 133 encoded data associated with the 1407B image unit (xft).

[00170] The conditional input generator 182 determines each of an estimated motion value 1593A (ma^c) and an estimated motion value 1593B (mc^a) based on the first predicted image unit (xa) and the second predicted image unit (xc). The conditional input 187B includes the estimated motion value 1593A (ma^c), the estimated motion value 1593B (mc^a), or both.

[00171] Attribute generator 184 processes conditional input 187B to generate attribute data 183B. For example, attribute layer 504A processes estimated motion value 1593A (ma^c), estimated motion value 1593B (rhc^a), or both, to generate attribute data 183BA. Attribute layer 504B processes an output from attribute layer 504A to generate attribute data 183BB, and so on. Decoder 186 processes the encoded data 165B (e.g., the encoded data 1465B) and the attribute data 183B, as described with reference to Figure 5, to generate a predicted motion value 1595A (ma^b), a predicted motion value 1595B (mc^b), a weight 1565 (a), or a combination thereof.

[00172] In a particular example, decoder portion 180 uses the estimated motion value 1593A (fna^c) and the estimated motion value 1593B (mc^a) corresponding to motion values ​​(e.g., motion vectors) between the first predicted image unit (xa) and the second predicted image unit (xc) as the conditional input 187B and generates the predicted motion value 1595A (mc^b), the predicted motion value 1595B (inc^b), and the weight Petition 870250078561, dated 03 / 09 / 2025, page 71 / 313 66 / 133 1565 (K) which can be used to generate a predicted image unit (xb) associated with the 1407B image unit (xft), as further described with reference to Figures 16 to 18.

[00173] With reference to Figure 16, an example of an image estimator 1600 of a compression network 140 is shown that is operable to perform prediction corresponding to image reconstruction. For example, the image estimator 1600 is included in the estimator 170 of the encoder portion 160, the estimator 188 of the decoder portion 180, or both of the second compression network 140 associated with image reconstruction.

[00174] The image estimator 1600 performs a distortion 1602A of a predicted image unit 1607A (xa) based on a predicted motion value 1695A (ma_+b) to generate an estimated image unit 1609A (a?^i). The image estimator 1600 performs a distortion 1602B of a predicted image unit 1607C (xc) based on a predicted motion value 1695B (mc^b) to generate an estimated image unit 1609B (χδ2). The image estimator 1600 performs a combination 1604 of the estimated image unit 1609A (χδ1) and the estimated image unit 1609B (i?^2) to generate an estimated image unit 1671B (xb). For example, the estimated image unit 1671B (xft) corresponds to a weighted combination of the estimated image unit 1609A (χδ1) and the estimated image unit 1609B (xb2).To illustrate, the image estimator 1600 applies a first weight (e.g., a weight of 1665) to the estimated image unit 1609A (xb1) to generate a first weighted image unit, then applies a second weight (e.g., 1 - the weight 1665) to the estimated image unit 1609B (i?^2) to... Petition 870250078561, dated 03 / 09 / 2025, page 72 / 313 67 / 133 generate a second weighted image unit and combine the first weighted image unit and the second image unit to generate the estimated image unit 1671B (xb).

[00175] The image estimator 1600 that generates the estimated image unit 1671B based on two predicted image units 1607 is provided as an illustrative example. In other examples, the image estimator 1600 may generate the estimated image unit 1671B based on more than two predicted image units 1607. For example, the image estimator 1600 may distort one or more additional predicted image units 1607 to generate one or more additional estimated image units and combine the estimated image units based on various weights to generate the estimated image unit 1671B.

[00176] In a particular aspect, the encoder portion 160 of the second compression network 140 uses the image estimator 1600 to generate a first estimated image unit (xb) corresponding to the image unit 1407B, as further described with reference to Figure 17. In a particular aspect, the decoder portion 180 of the second compression network 140 uses the image estimator 1600 to generate a second estimated image unit (xb) corresponding to the image unit 1407B, as further described with reference to Figure 18.

[00177] With reference to Figure 17, an example 1700 of an encoder portion 160 of the compression network 140 is shown that is operable to perform prediction corresponding to image reconstruction. For example, encoder portion 160 corresponds to an image encoder portion of the second compression network 140 associated with Petition 870250078561, dated 03 / 09 / 2025, page 73 / 313 68 / 133 image reconstruction based on motion estimation.

[00178] Encoder portion 160 is configured to process one or more image units 1407 to generate encoded datasets 165. Conditional input generator 162 uses local decoder portion 168 to generate a predicted image unit 1769A(xa). For example, local decoder portion 168 processes encoded data associated with image unit 1407A(xa) to generate the predicted image unit 1769A(xa). In a particular example, the predicted image unit 1769A(xa) corresponds to an estimate of a predicted image unit associated with image unit 1407A that can be generated in decoder portion 180 of Figure 18. Similarly, conditional input generator 162 uses local decoder portion 168 to generate a predicted image unit 1769C(xc). For example, the local decoder portion 168 processes encoded data associated with picture unit 1407C (xc) to generate the predicted picture unit 17 69C (xc).

[00179] The conditional input generator 162 uses the estimator 170 (e.g., the image estimator 1600) to generate an estimated image unit 1771B (xft) based on the motion value 1405A (ma^) and the motion value 1405B (mc^b) that are generated by the motion estimation performed by the encoder portion 160 of Figure 14. For example, the predicted image unit 1769A (xa) corresponds to the predicted image unit 1607A (xa), the predicted image unit 1769C (xc) corresponds to the predicted image unit 1607C (xc)), the motion value 1405A (ma^b) corresponds to the predicted motion value 1695A (ma^b), and the value of Petition 870250078561, dated 03 / 09 / 2025, page 74 / 313 69 / 133 movement 1405B (mc^ü) corresponds to the predicted movement value 1695B ímc-^b)· The conditional input generator 162 also provides a weight such as weight 1665. In a particular aspect, the weight is based on a configuration setting, default data, user input, or a combination thereof. The estimated image unit 1671B (xb) corresponds to the estimated image unit 1771B ( xb) . The conditional input 167B includes the predicted image unit 1769A (xa), the estimated image unit 1771B (xft), and the predicted image unit 1769C (xc).

[00180] Attribute generator 164 processes conditional input 167B to generate attribute data 163B, as described with reference to Figure 4. For example, attribute layer 404A processes predicted image unit 1769A (xa), estimated image unit 1771B (xft), and predicted image unit 1769C (xc) to generate attribute data 163BA. Attribute layer 404B processes an output from attribute layer 404A to generate attribute data 163BB, and so on. Encoder 166 processes picture unit 1407B (xb) and attribute data 163B to generate encoded data 165B, as described with reference to Figure 4. For example, encoder layer 402A processes picture unit 1407B (xft) and attribute data 163BA to generate an output. Encoder layer 402B processes an output from encoder layer 402A and attribute data 163BB to generate an output, and so on.An output from the 402N encoder layer corresponds to 165B encoded data (e.g., 1765B encoded data) associated with image reconstruction. Petition 870250078561, dated 03 / 09 / 2025, page 75 / 313 70 / 133

[00181] With reference to Figure 18, an example 1800 of a decoder portion 180 of a compression network 140 is shown that is operable to perform prediction corresponding to image reconstruction. For example, the decoder portion 180 corresponds to an image decoder portion of the second compression network 140 associated with image reconstruction based on motion estimation.

[00182] Decoder portion 180 is configured to process encoded datasets 165 to generate one or more predicted image units 1895. Conditional input generator 182 obtains a predicted image unit 1895A ( xa) . For example, decoder portion 180 processes encoded data associated with image unit 1407A (xa) to generate predicted image unit 1895A (xa) . Similarly, conditional input generator 182 obtains a predicted image unit 1895C (xc) . For example, decoder portion 180 processes encoded data associated with image unit 1407C (xc) to generate predicted image unit 1895C (xc).

[00183] The conditional input generator 182 uses the estimator 188 (e.g., the image estimator 1600) to generate an estimated image unit 1893B (xb) based on the predicted motion value 1595A (ma_^b) and the predicted motion value 1595B (mc^b) that are generated by the motion estimation performed by the decoder portion 180 of Figure 15. For example, the predicted image unit 1895A (xa) corresponds to the predicted image unit 1607A (xa)), the predicted image unit 1895C (xc) corresponds to the predicted image unit 1607C (xc)), the predicted motion value 1595A Petition 870250078561, dated 03 / 09 / 2025, page 76 / 313 71 / 133 (ina^b) corresponds to the predicted motion value 1695A (ma^b), the predicted motion value 1595B (mc_^b) corresponds to the predicted motion value 1695B (mc^), and the weight 1565 (a) from Figure 15 corresponds to the weight 1665. The estimated image unit 1671B (xb) corresponds to the estimated image unit 1893B (xb). The conditional input 187B includes the predicted image unit 1895A (xa), the estimated image unit 1893B (xb), and the predicted image unit 1895C (xc).

[00184] Attribute generator 184 processes conditional input 187B to generate attribute data 183B, as described with reference to Figure 5. For example, attribute layer 504A processes predicted image unit 1895A (xa), estimated image unit 1893B (xb), predicted image unit 1895C (xc) to generate attribute data 183BA. Attribute layer 504B processes an output from attribute layer 504A to generate attribute data 183BB, and so on. Decoder 186 processes encoded data 165B (e.g., encoded data 1765B) and attribute data 183B, as described with reference to Figure 5, to generate a predicted image unit 1895B (xb).A technical advantage of using the 140 compression network to process the 1407B image unit to generate the 1465B encoded data and the 1765B encoded data, and processing the 1465B encoded data and the 1765B encoded data to generate the predicted 1895B image unit, may include a reduced size of the 1465B encoded data, the 1765B encoded data, or both.

[00185] Figures 19 to 23 describe an example of a 140 compression network that is configured for Petition 870250078561, dated 03 / 09 / 2025, page 77 / 313 72 / 133 perform prediction corresponding to low-latency image reconstruction. The 140 compression network includes a first 140 compression network associated with low-latency motion estimation and a second 140 compression network associated with low-latency image reconstruction based on low-latency motion estimation.

[00186] Figure 19 includes an example of a 160 encoder portion of the first 140 compression network associated with low-latency motion estimation. Figure 20 includes an example of a 180 decoder portion of the first 140 compression network associated with low-latency motion estimation. Figure 22 includes an example of a 160 encoder portion of the second 140 compression network associated with low-latency image reconstruction based on low-latency motion estimation. Figure 23 includes an example of a 180 decoder portion of the second 140 compression network associated with low-latency image reconstruction based on low-latency motion estimation.Figure 21 includes an example of image estimation that can be performed on the encoder portion 160 and the decoder portion 180 of the second compression network 140 associated with low-latency image reconstruction based on low-latency motion estimation.

[00187] In some examples, a 160 encoder portion of the 140 compression network includes the 160 encoder portion of the first 140 compression network and the 160 encoder portion of the second 140 compression network. Similarly, the 180 decoder portion of the 140 compression network includes the 180 decoder portion of the Petition 870250078561, dated 03 / 09 / 2025, page 78 / 313 73 / 133 first compression network 140 and the decoder portion 180 of the second compression network 140.

[00188] With reference to Figure 19, a diagram is shown of an illustrative aspect of an example of a 160 encoder portion of a 140 compression network that is operable to perform matching prediction for low-latency image reconstruction. For example, the 160 encoder portion is included in the first 140 compression network associated with low-latency motion estimation.

[00189] Encoder portion 160 is configured to process one or more picture units 1407 to generate encoded datasets 165. Encoder portion 160 is configured to determine one or more motion values ​​1905 associated with one or more picture units 1407. For example, encoder portion 160, in response to the determination that one picture unit 1407D (x^ ) of one or more picture units 1407 should be encoded, determines one or more motion values ​​1905 based on a comparison of the picture unit 1407D with one or more other picture units of one or more picture units 1407.To illustrate, the encoder portion 160, in response to the determination that the 1407D picture unit should be encoded, determines a motion value 1905A (mc^d) based on a comparison of the 1407C picture unit (xc) and the 1407D picture unit (xd), determines one or more additional motion values ​​based on a comparison of the 1407D picture unit and one or more picture units that precede the 1407D picture unit in the one or more 1407 picture units, or a combination thereof. Petition 870250078561, dated 03 / 09 / 2025, page 79 / 313 74 / 133

[00190] In a particular aspect, the one or more motion values ​​1905 represent motion vectors associated with one or more image units 1407. For example, the motion value 1905A (rnc^d) represents motion vectors associated with image unit 1407C and image unit 1407D. In a particular aspect, the one or more motion values ​​1905 indicate one or more of the following linear velocity, linear acceleration, linear position, angular velocity, angular acceleration, or angular position associated with one or more image units 1407. For example, the motion value 1905A indicates one or more of the following linear velocity, linear acceleration, linear position, angular velocity, angular acceleration, or angular position associated with image unit 1407C and image unit 1407D. In one particular aspect, the one or more motion values ​​1905 are based on sensor data 226 received from one or more sensors 240, as described with reference to Figure 2.For example, the motion value 1905A (rnc^d) is based on the first sensor data 226 associated (e.g., captured simultaneously with) with the image unit 1407C and the second sensor data 226 associated (e.g., captured simultaneously with) with the image unit 1407D.

[00191] The conditional input generator 162 of the encoder portion 160 generates the conditional input 167D. For example, the conditional input generator 162 uses the local decoder portion 168 to process encoded data associated with the image unit 1407A to generate a first predicted image unit (xa), uses the local decoder portion 168 to process encoded data associated with Petition 870250078561, dated 03 / 09 / 2025, page 80 / 313 75 / 133 image unit 1407B to generate a second predicted image unit (xft) and uses the local decoder portion 168 to process encoded data associated with image unit 1407C to generate a third predicted image unit (xc). In a particular aspect, the first predicted image unit (xa) corresponds to an estimate of a prediction of image unit 1407A (xa) that can be generated in decoder portion 180 of Figure 23. Conditional input generator 162 determines an estimated motion value 1969A (m^c) based on the second predicted image unit (¾) and the third predicted image unit (xc). Conditional input generator 162 determines an estimated motion value 1969B (fna_^b) based on the first predicted image unit (xa) and the second predicted image unit (x^). The conditional input 167D includes the estimated movement value 1969A (mb^c), the estimated movement value 1969B (ma_^b), or both.

[00192] Attribute generator 164 processes conditional input 167D to generate attribute data 163D, as described with reference to Figure 4. For example, attribute layer 404A processes estimated motion value 1969A and estimated motion value 1969B to generate attribute data 163DA. Attribute layer 404B processes an output from attribute layer 404A to generate attribute data 163DB, and so on. Attribute data 163D includes attribute data 163DA, attribute data 163DB, attribute data 163DC, one or more additional sets of attribute data that include attribute data 163DN, or a combination thereof. Encoder 166 processes motion value 1905A (mc^d), image unit 1407D (xd), and the Petition 870250078561, dated 03 / 09 / 2025, page 81 / 313 76 / 133 attribute data 163D to generate the encoded data 165D associated with the image unit 1407D. For example, encoder layer 402A processes the motion value 1905A (rnc^d), the image unit 1407D (xd), and the attribute data 163DA to generate an output. Encoder layer 402B processes an output from encoder layer 402A and the attribute data 163DB to generate an output, and so on. An output from encoder layer 402N corresponds to the encoded data 165D (e.g., encoded data 1965D) associated with the motion estimate. One or more motion values ​​1905, one or more weights, or a combination thereof can be used to generate an estimated image unit (xd), as further described with reference to Figure 21.

[00193] The 160 encoder portion can thus generate the 165D encoded data independently of (e.g., before) obtaining any subsequent image units to the 1407D image unit in one or more 1407 image units. A technical advantage of generating the 165D encoded data independently of subsequent image units may include reduced latency associated with generating the 165D encoded data without having to wait for access to subsequent image units.

[00194] With reference to Figure 20, an example 2000 of a decoder portion 180 of a compression network 140 is shown that is operable to perform matching prediction for low-latency image reconstruction. For example, the decoder portion 180 is included in the first compression network associated with low-latency motion estimation. Petition 870250078561, dated 03 / 09 / 2025, page 82 / 313 77 / 133

[00195] The decoder portion 180 is configured to process encoded datasets 165 to generate one or more predicted motion values ​​2095, one or more weights, or a combination thereof. The one or more predicted motion values ​​2095, the one or more weights, or a combination thereof can be used to generate an estimated image unit (xb), as further described with reference to Figure 21.

[00196] The conditional input generator 182 of the decoder portion 180 generates the conditional input 187D. For example, the conditional input generator 182 obtains a first predicted image unit (xa) that is generated by the decoder portion 180 of the second compression network 140 processing encoded data associated with image unit 1407A (xa), as further described with reference to Figure 23. Similarly, in some respects, the conditional input generator 182 obtains a second predicted image unit (xft) and a third predicted image unit (xc) associated with image unit 1407B (xb)) and image unit 1407C (xc)), respectively, as further described with reference to Figure 23. In a particular aspect, image unit 1407A (xa), image unit 1407B (xft), image unit 1407C (xc), or a combination thereof, precede image unit 1407D (xd)) which is being encoded.

[00197] The conditional input generator 182 determines an estimated motion value 2093A (mb^c) based on a comparison of the second predicted image unit (xb) and the third predicted image unit (xc). In this way Petition 870250078561, dated 03 / 09 / 2025, page 83 / 313 78 / 133 similarly, conditional input generator 182 determines an estimated motion value 2093B (ma_b) based on a comparison of the first predicted image unit (xa) and the second predicted image unit (xb). Conditional input 187D includes the estimated motion value 2093A (mb_c), the estimated motion value 2093B (ina_b), or both.

[00198] Attribute generator 184 processes conditional input 187D to generate attribute data 183D, as described with reference to Figure 5. For example, attribute layer 504A processes estimated motion value 2093A (mb_c), estimated motion value 2093B (ma-+b), or both, to generate attribute data 183DA. Attribute layer 504B processes an output from attribute layer 504A to generate attribute data 183DB, and so on. Attribute data 183D includes attribute data 183DA, attribute data 183DB, attribute data 183DC, one or more additional sets of attribute data that include attribute data 183DN, or a combination thereof. Decoder 186 processes the 165D encoded data (e.g., the 1965D encoded data) and the 183D attribute data, as described with reference to Figure 5, to generate a predicted motion value 2095A (inc_d), a weight 2065 (K), or both.For example, the 508N decoder layer processes the 165D encoded data and the 183DN attribute data to generate an output. The 508B decoder layer processes an output from the 508C decoder layer and the 183DB attribute data to generate an output, and so on. An output from the 508A decoder layer corresponds to the value of... Petition 870250078561, dated 03 / 09 / 2025, page 84 / 313 79 / 133 predicted motion 2095A (mc^d), to weight 2065 (a) or to both associated with image unit 1407D.

[00199] In a particular example, decoder portion 180 uses the estimated motion value 2093A (mb^c) and the estimated motion value 2093B (ma^b) corresponding to motion values ​​(e.g., motion vectors) between predicted image units that precede image unit 1407D as the conditional input 187D and generates the predicted motion value 2095A (mc^d) and the weight 2065 (a) which can be used to generate a predicted image unit (xd) associated with image unit 1407D (xd), as further described with reference to Figures 21 to 23. A technical advantage of generating the predicted motion value 2095A (mc^d) independently of the encoded data associated with any image units subsequent to image unit 1407D may include reduced latency associated with the generation of the predicted motion value. 2095A (mc^d).

[00200] With reference to Figure 21, an example of an image estimator 2100 of a compression network 140 is shown that is operable to perform matching prediction for low-latency image reconstruction. For example, image estimator 2100 is included in estimator 170 of encoder portion 160, estimator 188 of decoder portion 180, or both of the second compression network 140 associated with low-latency image reconstruction.

[00201] The image estimator 2100 performs a distortion 2102 of a predicted image unit 2107C (xc)) based on a predicted motion value 2195A (mc^d) to generate an estimated image unit 2109A (xd1). The estimator Petition 870250078561, dated 03 / 09 / 2025, page 85 / 313 Image estimator 2100 uses the predicted image unit 2107C (xc) as an estimated image unit 2109B (xd2). Image estimator 2100 performs a combination 2104 of the estimated image unit 2109A (xdi) and the estimated image unit 2109B (xd2) to generate an estimated image unit 2171D (xd). For example, the estimated image unit 2171D (xd) corresponds to a weighted combination of the estimated image unit 2109A (xdl) and the estimated image unit 2109B (A?d2). To illustrate, the image estimator 2100 applies a first weight (e.g., a weight of 2165) to the estimated image unit 2109A (xd1) to generate a first weighted image unit, applies a second weight (e.g., 1 to the weight 2165) to the estimated image unit 2109B (xd2) to generate a second weighted image unit, and combines the first weighted image unit and the second image unit to generate the estimated image unit 2171D (xd).

[00202] The image estimator 2100 that generates the estimated image unit 2171D based on a single predicted image unit 2107 is provided as an illustrative example. In other examples, the image estimator 2100 may generate the estimated image unit 2171D based on multiple predicted image units 2107. For example, the image estimator 2100 may distort one or more additional predicted image units 2107 to generate one or more additional estimated image units and combine the estimated image units based on various weights to generate the estimated image unit 2171D.

[00203] In one particular aspect, the 160 encoder portion of the second 140 compression network uses the Petition 870250078561, dated 03 / 09 / 2025, page 86 / 313 81 / 133 image estimator 2100 to generate a first estimated image unit (xd) corresponding to the image unit 1407D, as further described with reference to Figure 22. In a particular aspect, the decoder portion 180 of the second compression network 140 uses the image estimator 2100 to generate a second estimated image unit (xd) corresponding to the image unit 1407D, as further described with reference to Figure 23.

[00204] With reference to Figure 22, an example 2200 of an encoder portion 160 of a compression network 140 is shown that is operable to perform matching prediction for low-latency image reconstruction. For example, encoder portion 160 corresponds to an image encoder portion of the second compression network 140 associated with low-latency image reconstruction based on low-latency motion estimation.

[00205] Encoder portion 160 is configured to process one or more image units 1407 to generate encoded datasets 165. Conditional input generator 162 uses local decoder portion 168 to generate a predicted image unit 2269C (xc). For example, local decoder portion 168 processes encoded data associated with image unit 1407C (xc) to generate the predicted image unit 2269C (xc). In a particular example, the predicted image unit 2269C (xc) corresponds to an estimate of a predicted image unit associated with image unit 1407C that can be generated in decoder portion 180 of Figure 23.

[00206] The conditional input generator 162 uses Petition 870250078561, dated 03 / 09 / 2025, page 87 / 313 82 / 133 The estimator 170 (e.g., image estimator 2100) is used to generate an estimated image unit 2271D (xd) based on the motion value 1905A (mc^d) that is generated by the motion estimation performed in the encoder portion 160 of Figure 19. For example, the predicted image unit 2269C (xc) corresponds to the predicted image unit 2107C (xc), and the motion value 1905A (mc^d) corresponds to the predicted motion value 2195A (mc^d). The conditional input generator 162 also provides a weight such as weight 2165. In a particular aspect, the weight is based on a configuration setting, default data, user input, or a combination thereof. The estimated image unit 2171D (xd) corresponds to the estimated image unit 2271D (xd). The conditional input 167D associated with the image unit 1407D includes the predicted image unit 2269C (xc) and the estimated image unit 2271D (xd).

[00207] The attribute generator 164 processes the conditional input 167D to generate attribute data 163D, as described with reference to Figure 4. For example, attribute layer 404A processes the conditional input 167D to generate attribute data 163DA. Attribute layer 404B processes an output from attribute layer 404A to generate attribute data 163DB, and so on. The attribute data 163D includes attribute data 163DA, attribute data 163DB, attribute data 163DC, one or more additional sets of attribute data that include attribute data 163DN, or a combination thereof. Encoder 166 processes the image unit 1407D (xd) to generate the encoded data 165D. For example, the encoder layer Petition 870250078561, dated 03 / 09 / 2025, page 88 / 313 83 / 133 Layer 402A processes the 1407D (xd) image unit and 163DA attribute data to generate an output. Layer 402B processes an output from layer 402A and 163DB attribute data to generate an output, and so on. An output from layer 402N corresponds to 165D encoded data (e.g., 2265D encoded data) associated with low-latency image reconstruction.

[00208] The 160 encoder portion can thus generate the 165D encoded data independently of (e.g., before) obtaining any subsequent image units to the 1407D image unit in one or more 1407 image units. A technical advantage of generating the 165D encoded data independently of subsequent image units may include reduced latency associated with generating the 165D encoded data without having to wait for access to subsequent image units.

[00209] With reference to Figure 23, an example 2300 of a decoder portion 180 of a compression network 140 is shown that is operable to perform matching prediction for low-latency image reconstruction. For example, the decoder portion 180 corresponds to an image decoder portion of the second compression network 140 associated with low-latency image reconstruction based on low-latency motion estimation.

[00210] Decoder portion 180 is configured to process encoded datasets 165 to generate one or more predicted image units 2395. Conditional input generator 182 obtains a predicted image unit 2395C (xc). For example, decoder portion 180 processes encoded data associated with Petition 870250078561, dated 03 / 09 / 2025, page 89 / 313 84 / 133 image unit 1407C (xc) to generate the predicted image unit 2395C (xc).

[00211] The conditional input generator 182 uses the estimator 188 (e.g., the image estimator 2100) to generate an estimated image unit 2393D (xd) based on the predicted motion value 2095A (mc^d) that is generated by the motion estimation performed by the decoder portion 180 of Figure 20. For example, the predicted image unit 2395C (xc) corresponds to the predicted image unit 2107C (xc)), the predicted motion value 2095A (mc_^d) corresponds to the predicted motion value 2195A (mc-^d), and the weight 2065 (a) of Figure 20 corresponds to the weight 2165. The estimated image unit 2171D (xd) corresponds to the estimated image unit 2393D (xd). The conditional input 187D associated with the image unit 1407D includes the predicted image unit 2395C (xc) and the estimated image unit 2393D (xd).

[00212] The attribute generator 184 processes the conditional input 187D to generate attribute data 183D, as described with reference to Figure 5. For example, attribute layer 504A processes the predicted image unit 2395C (xc) and the estimated image unit 2393D (xd) to generate attribute data 183DA. Attribute layer 504B processes an output from attribute layer 504A to generate attribute data 183DB, and so on. Attribute data 183D includes attribute data 183DA, attribute data 183DB, attribute data 183DC, one or more additional sets of attribute data that include attribute data 183DN, or a combination thereof. Decoder 186 processes the encoded data 165B (by Petition 870250078561, dated 03 / 09 / 2025, p. 90 / 313 85 / 133 example, the encoded data 2265D) and the attribute data 183D, as described with reference to Figure 5, to generate a predicted image unit 2395D (xd). A technical advantage of using the compression network 140 may include a reduced size of the encoded data 1965D, the encoded data 2265D, or both. A technical advantage of generating the predicted image unit 2395D (xd) independently of the encoded data of any subsequent image units of one or more image units 1407 may include reduced latency associated with the generation of the predicted image unit 2395D (xd).

[00213] Figure 24 depicts a 2400 implementation of a 2402 integrated circuit that includes one or more 2490 processors. The one or more 2490 processors include a compression network component 2460, such as the encoder portion 160, the decoder portion 180, or both of the compression network 140 of Figure 1. The 2402 integrated circuit also includes a signal input 2404, such as one or more bus interfaces, to enable the input data 2428 to be received for processing. The integrated circuit 2402 also includes a signal output 2406, as a bus interface, to enable the sending of output data 2450. For example, in implementations where the compression network component 2460 includes the encoder portion 160, the input data 2428 may include the input value(s) 105 and the output data 2450 may include the encoded data sets 165.In implementations where the compression network component 2460 includes the encoder portion 180, the input data 2428 may include the encoded datasets 165 and the data. Petition 870250078561, dated 03 / 09 / 2025, page 91 / 313 86 / 133 output 2450 may include the predicted value(s) 195. In implementations where the compression network component 2460 includes the encoder portion 160 and the decoder portion 180, the input data 2428 may include the input value(s) 105 and the output data 2450 may include the predicted value(s) 195. The integrated circuit 2402 enables the implementation of at least one portion of the compression network 140 as a component in various systems, such as a mobile phone or tablet depicted in Figure 25, a headset depicted in Figure 26, a wearable electronic device depicted in Figure 27, a voice-controlled speaker system depicted in Figure 28, a camera depicted in Figure 29, a virtual reality, mixed reality, or augmented reality headset depicted in Figure 30 or a vehicle depicted in Figure 31 or Figure 32.

[00214] Figure 25 depicts an implementation 2500 in which the compression network component 2460 is integrated into a mobile device 2502, such as a phone or tablet, as illustrative and non-limiting examples. The mobile device 2502 includes one or more microphones 2510, one or more speakers 2520, and a display screen 2504. Furthermore, the compression network component 2460 is integrated into the mobile device 2502 and is illustrated by dashed lines to indicate internal components that are not generally visible to a user of the mobile device 2502. In a particular example, the compression network component 2460 operates to improve encoding efficiency to reduce the amount of resources used for transmission or storage of encoded data. In a non-limiting illustrative example, the compression network component 2460 operates Petition 870250078561, dated 03 / 09 / 2025, page 92 / 313 87 / 133 to encode video data captured by a camera on mobile device 2502 for transmission to another device, decode video data received from another device, or a combination thereof. In another illustrative, non-limiting example, compression network component 2460 operates to encode data for storage on mobile device 2502 and decode the data upon retrieval from storage.

[00215] Figure 26 depicts an implementation 2600 in which the compression network component 2460 is implemented in a headset device 2602. The headset device 2602 includes a microphone 2610 and a camera 2620. In a particular example, the compression network component 2460 operates to improve encoding efficiency to reduce the amount of resources used for transmission or storage of encoded data. In a non-limiting illustrative example, the compression network component 2460 operates to encode data, such as audio data captured by the microphone 2610, video data captured by the camera 2620, and / or head tracking data (e.g., sensor data from an IMU of the headset device 2602) for transmission to another device, to decode data, such as audio data received from another device for playback in the headset device's headphones 2602, or a combination thereof.In another illustrative, non-limiting example, the compression network component 2460 operates to encode data for storage in the headset device 2602 and decode the data upon retrieval from storage. Petition 870250078561, dated 03 / 09 / 2025, page 93 / 313 88 / 133

[00216] Figure 27 depicts an implementation 2700 in which the compression network component 2460 is implemented in a wearable electronic device 2702, illustrated as a smartwatch. The compression network component 2460, a microphone 2710, a speaker 2720, and a display screen 2704 are integrated into the wearable electronic device 2702. In a particular example, the compression network component 2460 operates to improve encoding efficiency to reduce the amount of resources used for transmission or storage of encoded data.In a non-limiting illustrative example, the compression network component 2460 operates to encode data, such as audio data captured by microphone 2710, for transmission to another device; alternatively, the compression network component 2460 operates to decode data, such as audio data received from another device for playback on speaker 2720 or video data for playback on display screen 2704; or both. In another non-limiting illustrative example, the compression network component 2460 operates to encode data for storage on the wearable electronic device 2702 and decode the data upon retrieval from storage.

[00217] In a particular example, the 2702 wearable electronic device includes a haptic device that provides haptic notification (e.g., vibration) in response to the detection of activity associated with the operation of the 2460 compression network component. For example, the haptic notification might cause a user to look at the 2702 wearable electronic device to see a Petition 870250078561, dated 03 / 09 / 2025, page 94 / 313 89 / 133 notification displayed that data (e.g., audio or video data) has been received from a remote device and is available for playback on the 2702 wearable electronic device. The 2702 wearable electronic device can thus alert a user with a hearing impairment or a user wearing a headset about such notifications.

[00218] Figure 28 is an implementation 2800 in which the compression network device 2460 is implemented in a wireless loudspeaker and voice-activated device 2802. The wireless loudspeaker and voice-activated device 2802 may have wireless network connectivity and is configured to perform a wizard operation. One or more processors 2890, including the compression network component 2460, a microphone 2810, a camera 2820, a loudspeaker 2804, or a combination thereof, are included in the wireless loudspeaker and voice-activated device 2802. In a particular example, the compression network component 2460 operates to improve encoding efficiency to reduce the amount of resources used for transmission or storage of encoded data.In a non-limiting illustrative example, the compression network component 2460 operates to encode video data captured by camera 2820 for transmission to another device, decode audio data for playback on speaker 2804, decode video data received from another device for playback on a display screen (not shown) of the speaker and wireless voice-activated device 2802, or a combination thereof. In another non-limiting illustrative example, the compression network component 2460 operates to encode data for storage on the speaker and voice-activated device. Petition 870250078561, dated 03 / 09 / 2025, page 95 / 313 90 / 133 wireless 2802 and decode the data by retrieving it from storage.

[00219] During operation, in response to receiving a verbal command from a user via the microphones 2810, the speaker and the wireless voice-activated device 2802 can perform assistant operations, such as via the execution of a voice activation system (e.g., an integrated assistant application). Assistant operations may include adjusting a temperature, playing music, turning on lights, etc. For example, assistant operations may include initiating data transmission to a remote device, receiving data from a remote device, and / or storing / retrieving data from a local memory of the speaker and wireless voice-activated device 2802, each of which is conducted more efficiently due to the operation of the compression network component 2460.

[00220] Figure 29 depicts an implementation 2900 in which the device 2460 is implemented in a portable electronic device corresponding to a camera device 2902. The compression network component 2460 and a microphone 2910 are integrated into the camera device 2902. In a particular example, the compression network component 2460 operates to improve encoding efficiency to reduce the amount of resources used for transmission or storage of encoded data. In a non-limiting illustrative example, the compression network component 2460 operates to encode data captured by the camera device 2902, such as image data, video data, audio data, or a combination thereof, for transmission to another device. In another non-limiting illustrative example, the Petition 870250078561, dated 03 / 09 / 2025, page 96 / 313 The 91 / 133 compression network component 2460 operates to encode data, such as video or image data captured on the camera device 2902, for storage in the camera device's memory 2902, and also to decode the data upon retrieval from the memory.

[00221] Figure 30 depicts an implementation 3000 in which the compression network component 2460 is implemented in a portable electronic device corresponding to a virtual reality, mixed reality, or augmented reality headset 3002. The compression network component 2460, a microphone 3010, and a camera 3020 are integrated into the headset 3002. A visual interface device, such as a display screen, is positioned in front of the user's eyes to enable the display of augmented reality, mixed reality, or virtual reality images or scenes to the user while using the headset 3002. In a particular example, the compression network component 2460 operates to improve encoding efficiency to reduce the amount of resources used for transmission or storage of encoded data.In an illustrative and non-limiting example, the 2460 compression network component operates to decode data received by the 3002 headset, such as video data, audio data, other data associated with providing a virtual, augmented, or mixed reality experience for a user of the 3002 headset, or a combination thereof. In another illustrative and non-limiting example, the 2460 compression network component operates to encode data captured by the 3002 headset, such as head tracking data from a 3002 headset IMU, for transmission to a remote device. Petition 870250078561, dated 03 / 09 / 2025, page 97 / 313 92 / 133 (for example, a virtual reality (VR) session server). In another illustrative, non-limiting example, the 2460 compression network component operates to encode data, such as video, audio, or image data captured or received by headset 3002, for storage in headset 3002 memory, and also operates to decode the data by retrieving it from memory.

[00222] Figure 31 depicts an implementation 3100 in which the compression network component 2460 corresponds to, or is integrated into, a vehicle 3102, illustrated as a manned or unmanned aerial device (e.g., a package delivery drone). The compression network component 2460, a microphone 3110, a loudspeaker 3120, a camera 3104, or a combination thereof are integrated into the vehicle 3102. User voice activity detection can be performed based on audio signals received from the microphone 3110 of the vehicle 3102, such as delivery instructions provided by an authorized user of the vehicle 3102.

[00223] In a particular example, the 2460 compression network component operates to improve encoding efficiency to reduce the amount of resources used for transmission or storage of encoded data. In a non-limiting illustrative example, the 2460 compression network component operates to encode video data captured by camera 3104 of vehicle 3102 for transmission to another device, decode video data received from another device, or a combination thereof. In another non-limiting illustrative example, the 2460 compression network component operates to encode Petition 870250078561, dated 03 / 09 / 2025, page 98 / 313 93 / 133 data for storage in vehicle 3102 and decode the data by retrieving it from storage.

[00224] Figure 32 depicts another implementation 3200 in which the compression network component 2460 corresponds to, or is integrated into, a vehicle 3202, illustrated as a car. The vehicle 3202 includes one or more processors 290, one or more processors 292, one or more processors 390, or a combination thereof, including the compression network component 2460. The vehicle 3202 also includes a microphone 3210, a camera 3204, or both.

[00225] User voice activity detection can be performed based on audio signals received from microphone 3210 of vehicle 3202. In some implementations, user voice activity detection can be performed based on an audio signal received from internal microphones (e.g., microphones 3210), such as for a voice command from an authorized passenger. For example, user voice activity detection can be used to detect a voice command from an operator of vehicle 3202 (e.g., from a parent to set a volume to 5 or to set a destination for an autonomous vehicle) and disregard the voice of another passenger (e.g., a child's voice command to set the volume to 10 or other passengers discussing another location).In some implementations, user voice activity detection can be performed based on an audio signal received from external microphones (e.g., microphone 3210), such as by an authorized vehicle user. In one particular implementation, this is in response to receiving a verbal command. Petition 870250078561, dated 03 / 09 / 2025, page 99 / 313 94 / 133 identified as user speech, a voice activation system initiates one or more vehicle operations 3202 based on one or more detected keywords (e.g., unlock, start engine, play music, display weather forecast or other voice command) in an input signal, such as providing feedback or information via a display 3220 or one or more loudspeakers (e.g., a loudspeaker 3230).

[00226] In a particular example, the 2460 compression network component operates to improve encoding efficiency to reduce the amount of resources used for transmission or storage of encoded data. In a non-limiting illustrative example, the 2460 compression network component operates to encode video data captured by camera 3204 of vehicle 3202 for transmission to another device, decode video data received from another device, or a combination thereof. In another non-limiting illustrative example, the 2460 compression network component operates to encode data for storage in vehicle 3202 and decode the data upon retrieval from storage.

[00227] Camera 3204 can capture one or more image frames while vehicle 3202 is in operation. Compression network component 2460 can process one or more motion values ​​(e.g., the one or more input values ​​105) associated with the one or more image frames to generate encoded datasets 165. The one or more motion values ​​may include the one or more image frames, one or more speed measurements, one or more acceleration measurements, other sensor data, a path of Petition 870250078561, dated 03 / 09 / 2025, page 100 / 313 95 / 133 vehicle navigation 3202 or a combination thereof. Vehicle 3202 may store the encoded data 165 in vehicle 3202, transmit the encoded data sets 165 to another device (e.g., a server), or both.

[00228] In some implementations, the compression network component 2460 generates one or more predicted values ​​195 based on the encoded datasets 165. In a collision avoidance example, the compression network component 2460 can track an object in one or more image frames. The compression network component 2460 generates one or more predicted values ​​195 corresponding to predicted future motion values. For example, one or more predicted values ​​195 indicate whether a predicted future position of vehicle 3202 is within a distance threshold of a predicted future position of the object.In some implementations, the compression network component 2460, in response to the determination that the predicted future position of vehicle 3202 is within the distance threshold of the predicted future position of the object, initiates one or more collision avoidance actions, such as generating an alert, initiating braking, activating an alarm, sending an alert to an emergency vehicle, or a combination thereof.

[00229] With reference to Figure 33, a particular implementation of a method 3300 for generating encoded data using a compression network is shown. In a particular aspect, one or more operations of the method 3300 are performed by at least one of the conditional input generator 162, the attribute generator 164, the encoder 166, the local decoder portion 168, the estimator 170, the encoder portion 160, the compression network 140 of Figure 1, the Petition 870250078561, dated 03 / 09 / 2025, page 101 / 313 96 / 133 one or more processors 290, device 202, system 200 of Figure 2, one or more processors 390, device 302, system 300 of Figure 3, compression network component 2460 of Figure 24 or a combination thereof.

[00230] Method 3300 involves obtaining a conditional input from a compression network, where the conditional input is based on one or more predicted first motion values, in 3302. For example, the conditional input generator 162 of Figure 1 obtains the conditional input 167B from the compression network 140, as described with reference to Figure 1. The conditional input 167B is based on a predicted value 169A (e.g., a predicted motion value).

[00231] Method 3300 also includes processing, using the compression network, the conditional input and one or more motion values ​​to generate encoded data associated with the one or more motion values, in 3304. For example, encoder portion 160 uses attribute generator 164 of compression network 140 to process conditional input 167B to generate attribute data 163B and uses encoder 166 of compression network 140 to process attribute data 163B and input value 105B to generate encoded data 165B associated with input value 105B, as described with reference to Figure 1.

[00232] Method 3300 thus enables the reduction of the amount of information to be provided to the decoder portion 180 as the encoded data 165A by generating the encoded data 165A based on an estimate (e.g., conditional input 167B) of information (e.g., conditional input 187B) that can be generated in the decoder portion 180. Petition 870250078561, dated 03 / 09 / 2025, page 102 / 313 97 / 133

[00233] The 3300 method of Figure 33 can be implemented by a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), a processing unit such as a central processing unit (CPU), a digital signal processor (DSP), a controller, another hardware device, a firmware device, or any combination thereof. As an example, the 3300 method of Figure 33 can be implemented by a processor that executes instructions, as described with reference to Figure 35.

[00234] With reference to Figure 34, a particular implementation of a method 3400 for generating predicted data from encoded data using a compression network is shown. In a particular aspect, one or more operations of the method 3400 are performed by at least one of the conditional input generator 182, the attribute generator 184, the decoder 186, the estimator 188, the decoder portion 180, the compression network 140 of Figure 1, the one or more processors 292, the device 260, the system 200 of Figure 2, the one or more processors 390, the device 302, the system 300 of Figure 3, the compression network component 2460 of Figure 24, or a combination thereof.

[00235] Method 3400 involves obtaining encoded data associated with one or more motion values, in 3402. For example, decoder portion 180 obtains, from encoder portion 160, the encoded data 165B associated with the input value 105B, as described with reference to Figure 1. Petition 870250078561, dated 03 / 09 / 2025, page 103 / 313 98 / 133

[00236] Method 3400 also includes obtaining a conditional input from a compression network, where the conditional input is based on one or more predicted first movement values, in 3404. For example, the conditional input generator 182 obtains the conditional input 187B from the compression network 140, as described with reference to Figure 1. The conditional input 187B is based on the predicted value 195A.

[00237] Method 3400 additionally includes processing, using the compression network, the encoded data and the conditional input to generate one or more second predicted motion values, in 3406. For example, decoder portion 180 uses attribute generator 184 of compression network 140 to process conditional input 187B to generate attribute data 183B and uses decoder 186 of the compression network 140 to process the attribute data 183B and the encoded data 165B to generate the predicted value 195B, as described with reference to Figure 1.

[00238] The 3400 method thus enables the reduction of the amount of information to be obtained by the decoder portion 180 as the encoded data 165A by generating the predicted value 195B based on the information (e.g., the conditional input 187B) that can be estimated (e.g., the conditional input 167B) in the encoder portion 160.

[00239] Method 3400 in Figure 34 can be implemented by an FPGA device, an ASIC, a processing unit such as a CPU, a DSP, a controller, another hardware device, a firmware device, or any combination thereof. As an example, method 3400 in Figure Petition 870250078561, dated 03 / 09 / 2025, page 104 / 313 99 / 133 can be performed by a processor that executes instructions, as described with reference to Figure 35.

[00240] With reference to Figure 35, a block diagram of a particular illustrative implementation of a device is depicted and generically designated by the numerical reference 3500. In various implementations, device 3500 may have a greater or lesser number of components than that illustrated in Figure 35. In an illustrative implementation, device 3500 may correspond to one or more of device 202, device 260 from Figure 2, or device 302 from Figure 3. In an illustrative implementation, device 3500 may perform one or more of the operations described with reference to Figures 1 to 34.

[00241] In a particular implementation, the 3500 device includes a 3506 processor (e.g., a CPU). The 3500 device may include one or more additional 3510 processors (e.g., one or more DSPs). In a particular aspect, the one or more 290 processors, the one or more 292 processors of Figure 2, the one or more 390 processors of Figure 3, or a combination thereof, correspond to the 3506 processor, the 3510 processors, or a combination thereof. The 3510 processors may include a speech and music encoder-decoder (CODEC) 3508 that includes a voice encoder (vocoder) 3536, a vocoder decoder 3538, or both. The 3510 processors may include the compression network component 2460, one or more applications 262, or a combination thereof.

[00242] The 3500 device may include a 3586 memory and a 3534 CODEC. The 3586 memory may include Petition 870250078561, dated 03 / 09 / 2025, page 105 / 313 100 / 133 instructions 3556 that are executable by one or more additional processors 3510 (or by processor 3506) to implement the described functionality with reference to the compression network component 2460. The device 3500 may include a modem 3570 coupled, via a transceiver 3550, to an antenna 3552. In a particular aspect, the modem 3570 corresponds to the modem 270, the modem 280, or both of Figure 2.

[00243] The device 3500 may include a display 3528 coupled to a display controller 3526. One or more loudspeakers 3520, one or more microphones 3524, or a combination thereof, may be coupled to the CODEC 3534. The CODEC 3534 may include a digital-to-analog converter (DAC) 3502, an analog-to-digital converter (ADC) 3504, or both. In a specific implementation, the CODEC 3534 can receive analog signals from one or more microphones 3524, convert the analog signals into digital signals using the analog-to-digital converter 3504, and provide the digital signals to the speech and music codec 3508. The speech and music codec 3508 can process the digital signals, and the digital signals can be further processed by the compression network component 2460, by one or more applications 262, or a combination thereof.In one particular implementation, the speech and music codec 3508 can provide digital signals to the codec 3534. The codec 3534 can convert the digital signals into analog signals using the digital-to-analog converter 3502 and can provide the analog signals to one or more loudspeakers 3520.

[00244] In a particular implementation, the Petition 870250078561, dated 03 / 09 / 2025, page 106 / 313 Device 3500 may be included in a packaged system or system-on-chip device 3522. In a particular implementation, memory 3586, processor 3506, processors 3510, display controller 3526, CODEC 3534, and modem 3570 are included in the packaged system or system-on-chip device 3522. In a particular implementation, an input device 3530, one or more sensors 240, and a power supply 3544 are coupled to the packaged system or system-on-chip device 3522. Furthermore, in a particular implementation, as illustrated in Figure 35, the display 3528, the input device 3530, one or more speakers 3520, one or more microphones 3524, one or more sensors 240, the antenna 3552, and the power supply are included. 3544 are external to the packaged system or system-on-chip device 3522.In one particular implementation, each of the display 3528, the input device 3530, the one or more speakers 3520, the one or more microphones 3524, the one or more sensors 240, the antenna 3552 and the power supply 3544 can be coupled to a component of the packaged system device or system-on-chip device 3522, such as an interface or a controller.

[00245] Device 3500 may include a smart speaker, a speakerbar, a mobile communication device, a smartphone, a cell phone, a laptop computer, a computer, a tablet, a personal digital assistant, a display device, a television, a game console, a music player, a radio, a digital video player, a digital video disc (DVD) player, a tuner, a camera, a device of Petition 870250078561, dated 03 / 09 / 2025, page 107 / 313 102 / 133 navigation, a vehicle, a headset, an augmented reality headset, a mixed reality headset, a virtual reality headset, an aerial vehicle, a home automation system, a voice-activated device, a wireless speaker and voice-activated device, a portable electronic device, a car, a computing device, a communication device, an Internet of Things (IoT) device, a virtual reality (XR) device, a base station, a mobile device, or any combination thereof.

[00246] In conjunction with the implementations described, an apparatus includes means for obtaining encoded data associated with one or more motion values. For example, the means for obtaining encoded data may correspond to the decoder portion 180, the encoder portion 160, the compression network 140 of Figure 1, the modem 280, one or more processors 292, the device 260, the system 200 of Figure 2, the storage device 392, the device 302, the system 300 of Figure 3, the processor 3506, the processors 3510, the antenna 3552, the transceiver 3550, the modem 3570, one or more other circuits or components configured to obtain encoded data associated with one or more motion values, or any combination thereof.

[00247] The apparatus also includes means for obtaining conditional input from a compression network, where the conditional input is based on one or more predicted first motion values. For example, the means for obtaining a conditional input may correspond to estimator 188, conditional input generator 182, decoder portion Petition 870250078561, dated 03 / 09 / 2025, page 108 / 313 103 / 133 180, to the compression network 140 of Figure 1, to one or more processors 292, to device 260, to system 200 of Figure 2, to device 302, to system 300 of Figure 3, to image estimator 1600 of Figure 16, to image estimator 2100 of Figure 21, to processor 3506, to processors 3510, to one or more other circuits or components configured to obtain the conditional input or any combination thereof.

[00248] The apparatus additionally includes means for processing, using the compression network, the encoded data and the conditional input to generate one or more second predicted motion values. For example, the means for processing the encoded data and the conditional input may correspond to attribute generator 184, decoder 186, decoder portion 180, compression network 140 of Figure 1, one or more processors 292, device 260, system 200 of Figure 2, device 302, system 300 of Figure 3, processor 3506, processors 3510, one or more other circuits or components configured to process the encoded data and the conditional input, or any combination thereof.

[00249] Also in conjunction with the implementations described, an apparatus includes means for obtaining a conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values. For example, the means for obtaining a conditional input may correspond to the local decoder portion 168, the estimator 170, the conditional input generator 162, the encoder portion 160, the compression network 140 of Figure 1, or one or more processors. Petition 870250078561, dated 03 / 09 / 2025, page 109 / 313 104 / 133 290, to device 202, to system 200 of Figure 2, to device 302, to system 300 of Figure 3, to image estimator 1600 of Figure 16, to image estimator 2100 of Figure 21, to processor 3506, to processors 3510, to one or more other circuits or components configured to obtain the conditional input, or any combination thereof.

[00250] The apparatus also includes means for processing, using the compression network, the conditional input and one or more motion values ​​to generate encoded data associated with one or more motion values. For example, the means for processing the conditional input and one or more motion values ​​may correspond to attribute generator 164, encoder 166, encoder portion 160, compression network 140 of Figure 1, one or more processors 290, device 202, system 200 of Figure 2, device 302, system 300 of Figure 3, processor 3506, processors 3510, one or more other circuits or components configured to process the conditional input and one or more motion values, or any combination thereof.

[00251] In some implementations, a non-transient computer-readable medium (for example, a computer-readable storage device such as 3586 memory) includes instructions (for example, the 3556 instructions) that, when executed by one or more processors (for example, the one or more 3510 processors or the 3506 processor), cause the one or more processors to obtain encoded data (for example, the 165B encoded data) associated with one or more movement values ​​(for example, the 105B input value). The instructions, when executed by Petition 870250078561, dated 03 / 09 / 2025, page 110 / 313 105 / 133 one or more processors also cause the one or more processors to obtain conditional input (e.g., conditional input 187B) from a compression network (e.g., compression network 140), where the conditional input is based on one or more predicted first movement values ​​(e.g., predicted value 195A). The instructions, when executed by the one or more processors, additionally cause the one or more processors to process, using the compression network, the encoded data and the conditional input to generate one or more predicted second movement values ​​(e.g., predicted value 195B).

[00252] In some implementations, a non-transient computer-readable medium (for example, a computer-readable storage device such as memory 3586) includes instructions (for example, instructions 3556) that, when executed by one or more processors (for example, one or more 3510 processors or processor 3506), cause the one or more processors to obtain a conditional input (for example, conditional input 167B) from a compression network (for example, compression network 140), where the conditional input is based on one or more predicted movement values ​​(for example, predicted value 169A).The instructions, when executed by one or more processors, also cause the one or more processors to process, using the compression network, the conditional input and one or more movement values ​​(for example, the input value 105B) to generate encoded data (for example, the encoded data 165B) associated with the one or more movement values. Petition 870250078561, dated 03 / 09 / 2025, page 111 / 313 106 / 133

[00253] The particular aspects of disclosure are described below in sets of interrelated examples:

[00254] According to Example 1, a device includes one or more processors configured to obtain encoded data associated with one or more motion values; obtain conditional input from a compression network, where the conditional input is based on one or more predicted first motion values; and process, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values.

[00255] Example 2 includes the device from Example 1, where one or more motion values ​​are based on the output of one or more sensors.

[00256] Example 3 includes the device from Example 1 or Example 2, wherein one or more sensors include an inertial measurement unit (IMU).

[00257] Example 4 includes the device of any of Examples 1 to 3, wherein the one or more motion values ​​represent one or more motion vectors associated with one or more image units.

[00258] Example 5 includes the device from Example 4, where one or more image units include an encoding unit.

[00259] Example 6 includes the device from Example 4 or Example 5, where one image unit of one or more image units includes a block of pixels.

[00260] Example 7 includes the device from any of Examples 4 to 6, where an image unit Petition 870250078561, dated 03 / 09 / 2025, p. 112 / 313 107 / 133 of one or more image units includes a pixel frame.

[00261] Example 8 includes the device from any of Examples 1 to 7, where the one or more second predicted motion values ​​represent future motion vectors.

[00262] Example 9 includes the device from any of Examples 1 to 8, where the one or more second predicted motion values ​​correspond to a reconstructed version of the one or more motion values.

[00263] Example 10 includes the device from any of Examples 1 to 9, in which one or more processors are integrated into at least one of the following: a headset, a mobile communication device, an extended reality (XR) device, or a vehicle.

[00264] Example 11 includes the device from any of Examples 1 to 10, wherein the one or more motion values ​​indicate one or more of linear velocity, linear acceleration, linear position, angular velocity, angular acceleration, or angular position.

[00265] Example 12 includes the device from any of Examples 1 through 11, where the compression network includes a multi-layered neural network.

[00266] Example 13 includes the device from any of Examples 1 to 12, wherein the compression network includes a video decoder, and wherein the video decoder has multiple decoder layers configured to decode multiple resolution orders of the encoded data associated with one or more motion values.

[00267] Example 14 includes the device from any of Examples 1 to 13, wherein one or more Petition 870250078561, dated 03 / 09 / 2025, page 113 / 313 108 / 133 processors are configured to track an object associated with one or more motion values ​​across one or more pixel frames.

[00268] Example 15 includes the device from Example 14, where the one or more second predicted motion values ​​represent a collision avoidance output associated with a vehicle.

[00269] Example 16 includes the device from Example 15, where the collision avoidance output indicates a predicted future position of the vehicle relative to the object.

[00270] Example 17 includes the device from Example 15 or Example 16, where the collision avoidance output indicates a predicted future position of the vehicle and a predicted future position of the object.

[00271] Example 18 includes the device from any of Examples 1 to 17 and additionally includes a modem configured to receive a bitstream from an encoder device, wherein the bitstream includes the encoded data.

[00272] Example 19 includes the device from any of Examples 1 through 18, wherein one or more processors are configured to process the encoded data and the attribute data to generate the one or more second predicted motion values, and wherein the attribute data are based on conditional input.

[00273] Example 20 includes the device from Example 19 and additionally includes a modem configured to receive a bitstream from an encoder device, where the bitstream includes attribute data and encoded data.

[00274] Example 21 includes the device from Example 19 or Example 20, where one or more processors Petition 870250078561, dated 03 / 09 / 2025, p. 114 / 313 109 / 133 are configured to process conditional input using network compression to generate attribute data.

[00275] Example 22 includes the device from any of Examples 19 to 21, where the attribute data correspond to multiscale attribute data with different spatial resolutions.

[00276] Example 23 includes the device from any of Examples 19 to 22, where the attribute data includes wavelet transform data.

[00277] Example 24 includes the device from any of Examples 1 to 23, wherein the encoded data includes encoded motion estimate data associated with an image unit (e.g., xb), and wherein one or more processors are configured to: determine estimated motion values ​​(e.g., ma^c, ™c^a) based on a comparison of a previously reconstructed image unit (e.g., xa) and a subsequently reconstructed image unit (e.g., xc); process the estimated motion values ​​to generate motion estimate attribute data corresponding to the image unit;and process, using the compression network, the encoded motion estimate data and the motion estimate attribute data to generate the one or more second predicted motion values, where the one or more second predicted motion values ​​include one or more reconstructed motion values ​​(e.g., ina_,b, ^c >b) corresponding to the one or more motion values.

[00278] Example 25 includes the device from Example 24, wherein the encoded data includes reconstruction encoded data, and wherein one or more Petition 870250078561, dated 03 / 09 / 2025, page 115 / 313 110 / 133 processors are configured to: generate an estimated image unit (e.g., xb) based on one or more reconstructed motion values ​​(e.g., ma^b, mc^b); process the estimated image unit to generate reconstruction attribute data corresponding to the image unit; and process, using the compression network, the encoded reconstruction data and the reconstruction attribute data to generate a reconstructed image unit (e.g., Xb).

[00279] Example 26 includes the device from Example 25, wherein the one or more predicted first motion values ​​include a first reconstructed motion value (e.g., fna^b) and a second reconstructed motion value (e.g., mc^b), and wherein the one or more processors are configured to: determine a first estimated version of the image unit based on applying the first reconstructed motion value (e.g., ina^b) to the previous reconstructed image unit (e.g., Xa); determine a second estimated version of the image unit based on applying the second reconstructed motion value (e.g., mc^b) to the subsequent reconstructed image unit (e.g., Xc); and generate the estimated image unit (e.g., xb) based on a combination of the first estimated version of the image unit and the second estimated version of the image unit.

[00280] Example 27 includes the device from any of Examples 1 to 23, where the encoded data includes motion estimation encoded data associated with an image unit (e.g., x^), and where one or more processors are configured to: determine Petition 870250078561, dated 03 / 09 / 2025, pp. 116 / 313 111 / 133 a first estimated motion value (e.g., mb^c) based on a comparison of a first reconstructed prior image unit (e.g., Xc) and a second reconstructed prior image unit (e.g., Xb); process at least the first estimated motion value to generate motion estimate attribute data corresponding to the image unit; and process, using the compression network, the encoded motion estimate data and the motion estimate attribute data to generate the one or more second predicted motion values, wherein the one or more second predicted motion values ​​include one or more reconstructed motion values ​​(e.g., mc^d) corresponding to the one or more motion values.

[00281] Example 28 includes the device of Example 27, wherein the encoded data includes reconstruction encoded data, and wherein one or more processors are configured to: generate an estimated image unit (e.g., xd) based on one or more reconstructed motion values ​​(e.g., mc^a); process the estimated image unit to generate reconstruction attribute data corresponding to the image unit; and process, using the compression network, the reconstruction encoded data and the reconstruction attribute data to generate a reconstructed image unit (e.g., xd).

[00282] Example 29 includes the device from Example 28, where one or more processors are configured to determine a second estimated motion value (e.g., ma^b) based on a comparison of the second reconstructed image unit (e.g., xb) and a Petition 870250078561, dated 03 / 09 / 2025, page 117 / 313 112 / 133 third reconstructed previous image unit (e.g., Xa), where the motion estimate attribute data corresponding to the image unit are additionally based on processing the second estimated motion value.

[00283] Example 30 includes the device from Example 28 or Example 29, wherein the one or more predicted first motion values ​​include a first reconstructed motion value (e.g., fnc^d), and wherein the one or more processors are configured to determine the estimated image unit (e.g., x^) based on applying the first reconstructed motion value (e.g., rnc^a) to the first reconstructed previous image unit (e.g., Xc), wherein the reconstruction attribute data is based on the estimated image unit (e.g., xp).

[00284] According to Example 31, a method includes obtaining, from a device, encoded data associated with one or more motion values; obtaining, from the device, a conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values; and processing, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values.

[00285] Example 32 includes the method from Example 31, where one or more motion values ​​are based on the output of one or more sensors.

[00286] Example 33 includes the method of Example 31 or Example 32, where one or more sensors include an inertial measurement unit (IMU).

[00287] Example 34 includes the method of Petition 870250078561, dated 03 / 09 / 2025, page 118 / 313 113 / 133 any of Examples 31 to 33, where one or more motion values ​​represent one or more motion vectors associated with one or more image units.

[00288] Example 35 includes the method from Example 34, where one image unit of one or more image units includes an encoding unit.

[00289] Example 36 includes the method from Example 34 or Example 35, where one image unit of one or more image units includes a block of pixels.

[00290] Example 37 includes the method of any of Examples 34 to 36, where an image unit of one or more image units includes a pixel frame.

[00291] Example 38 includes the method of any of Examples 31 to 37, where the one or more second predicted motion values ​​represent future motion vectors.

[00292] Example 39 includes the method from any of Examples 31 to 38, where the one or more second predicted motion values ​​correspond to a reconstructed version of the one or more motion values.

[00293] Example 40 includes the method of any of Examples 31 to 39, where the device includes at least one of a headset, a mobile communication device, an extended reality (XR) device, or a vehicle.

[00294] Example 41 includes the method of any of Examples 31 to 40, wherein one or more motion values ​​indicate one or more of linear velocity, linear acceleration, linear position, angular velocity, angular acceleration, or angular position. Petition 870250078561, dated 03 / 09 / 2025, p. 119 / 313 114 / 133

[00295] Example 42 includes the method of any of Examples 31 to 41, where the compression network includes a multi-layered neural network.

[00296] Example 43 includes the method of any of Examples 31 to 42, wherein the compression network includes a video decoder, and wherein the video decoder has multiple decoder layers configured to decode multiple resolution orders of the encoded data associated with one or more motion values.

[00297] Example 44 includes the method of any of Examples 31 to 43, additionally including the tracking of an object associated with one or more motion values ​​through one or more pixel frames.

[00298] Example 45 includes the method from Example 44, where the one or more second predicted motion values ​​represent a collision avoidance output associated with a vehicle.

[00299] Example 46 includes the method from Example 45, where the collision avoidance output indicates a predicted future position of the vehicle relative to the object.

[00300] Example 47 includes the method from Example 45 or Example 46, where the collision avoidance output indicates a predicted future position of the vehicle and a predicted future position of the object.

[00301] Example 48 includes the method of any of Examples 31 to 47 and additionally includes receiving a bitstream via modem from an encoder device, where the bitstream includes the encoded data.

[00302] Example 49 includes the method from any of Examples 31 to 48 and additionally includes the Petition 870250078561, dated 03 / 09 / 2025, pp. 120 / 313 115 / 133 processing of encoded data and attribute data to generate the one or more second predicted motion values, where the attribute data is based on conditional input.

[00303] Example 50 includes the method from Example 49 and additionally includes receiving a bitstream via a modem from an encoder device, where the bitstream includes attribute data and encoded data.

[00304] Example 51 includes the method from Example 49 or Example 50 and additionally includes processing the conditional input using the compression network to generate the attribute data.

[00305] Example 52 includes the method of any of Examples 31 through 51, wherein the processing of the encoded data and the conditional input includes: processing the conditional input using the compression network to generate attribute data; and processing the encoded data and the attribute data to generate the one or more second predicted motion values.

[00306] Example 53 includes the method of any of Examples 49 to 52, where the attribute data correspond to multiscale attribute data with different spatial resolutions.

[00307] Example 54 includes the method from any of Examples 49 through 53, where the attribute data includes wavelet transform data.

[00308] Example 55 includes the method of any of Examples 31 to 54 and additionally includes: determining estimated motion values ​​(e.g., ma^c, ™c^a) based on a comparison of a previous image unit. Petition 870250078561, dated 03 / 09 / 2025, pp. 121 / 313 116 / 133 reconstructed (e.g., xa) and a subsequent reconstructed image unit (e.g., Xc), wherein the encoded data includes encoded motion estimate data associated with an image unit (e.g., xb); process the estimated motion values ​​to generate motion estimate attribute data corresponding to the image unit; and process, using the compression network, the encoded motion estimate data and the motion estimate attribute data to generate the one or more second predicted motion values, wherein the one or more second predicted motion values ​​include one or more reconstructed motion values ​​(e.g., fna^b,inc^b) corresponding to the one or more motion values.

[00309] Example 56 includes the method of Example 55 and additionally includes: generating an estimated image unit (e.g., xb) based on one or more reconstructed motion values ​​(e.g., ma^,™c^b)'; processing the estimated image unit to generate reconstruction attribute data corresponding to the image unit; and processing, using the compression network, reconstruction-encoded data and reconstruction attribute data to generate a reconstructed image unit (e.g., Xb), wherein the encoded data includes the reconstruction-encoded data.

[00310] Example 57 includes the method of Example 56 and additionally includes: determining a first estimated version of the image unit based on applying a first reconstructed motion value (e.g., ma^b) to the previously reconstructed image unit (e.g., Xa), wherein one or more first predicted motion values Petition 870250078561, dated 03 / 09 / 2025, page 122 / 313 117 / 133 include the first reconstructed motion value (e.g., m^); determine a second estimated version of the image unit based on applying a second reconstructed motion value (e.g., mc^b) to the subsequent reconstructed image unit (e.g., lèc), where the first one or more predicted motion values ​​include the second reconstructed motion value (e.g., mc^b); and generate the estimated image unit (e.g., xb) based on a combination of the first estimated version of the image unit and the second estimated version of the image unit.

[00311] Example 58 includes the method of any of Examples 31 to 54 and additionally includes: determining a first estimated motion value (e.g., mb^c) based on a comparison of a first reconstructed prior image unit (e.g., Xc) and a second reconstructed prior image unit (e.g., Xb), wherein the encoded data includes encoded motion estimate data associated with an image unit (e.g., xd); processing at least the first estimated motion value to generate motion estimate attribute data corresponding to the image unit;and process, using the compression network, the encoded motion estimate data and the motion estimate attribute data to generate the one or more predicted second motion values, where the one or more predicted second motion values ​​include one or more reconstructed motion values ​​(e.g., mc^d) corresponding to the one or more motion values.

[00312] Example 59 includes the method from Example Petition 870250078561, dated 03 / 09 / 2025, page 123 / 313 118 / 133 and additionally includes: generating an estimated image unit (e.g., xd) based on one or more reconstructed motion values ​​(e.g., mc^d); processing the estimated image unit to generate reconstruction attribute data corresponding to the image unit; and processing, using the compression network, reconstruction-encoded data and reconstruction attribute data to generate a reconstructed image unit (e.g., lèd), wherein the encoded data includes the reconstruction-encoded data.

[00313] Example 60 includes the method from Example 59 and additionally includes determining a second estimated motion value (e.g., ma^b) based on a comparison of the second reconstructed previous image unit (e.g., lèb) and a third reconstructed previous image unit (e.g., lca), wherein the motion estimate attribute data corresponding to the image unit are additionally based on the processing of the second estimated motion value.

[00314] Example 61 includes the method of Example 59 or Example 60 and additionally includes determining the estimated image unit (e.g., xd) based on applying a first reconstructed motion value (e.g., fnc^d) to the first previously reconstructed image unit (e.g., xc), wherein the one or more first predicted motion values ​​include the first reconstructed motion value (e.g., fàc^,d), and wherein the reconstruction attribute data are based on the estimated image unit (e.g., x^). Petition 870250078561, dated 03 / 09 / 2025, page 124 / 313 119 / 133

[00315] According to Example 62, a device includes: a memory configured to store instructions; and a processor configured to execute the instructions to perform the method of any of Examples 31 to 61.

[00316] According to Example 63, a non-transient, computer-readable medium stores instructions that, when executed by a processor, cause the processor to perform the method of any of Examples 31 to 61.

[00317] According to Example 64, an apparatus includes means for carrying out the method of any of Examples 31 to 61.

[00318] According to Example 65, a device includes one or more processors configured to obtain conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values; and to process, using the compression network, the conditional input and one or more motion values ​​to generate encoded data associated with the one or more motion values.

[00319] Example 66 includes the device from Example 65, where one or more motion values ​​are based on the output of one or more sensors.

[00320] Example 67 includes the device of Example 66, where one or more sensors include an inertial measurement unit (IMU).

[00321] Example 68 includes the device of any of Examples 65 to 67, where the one or more motion values ​​represent one or more motion vectors associated with one or more image units. Petition 870250078561, dated 03 / 09 / 2025, p. 125 / 313 120 / 133

[00322] Example 69 includes the device of Example 68, where one or more image units include an encoding unit.

[00323] Example 70 includes the device of Example 68 or Example 69, where one or more image units include a block of pixels.

[00324] Example 71 includes the device of any of Examples 68 to 70, where one or more image units include a pixel frame.

[00325] Example 72 includes the device from any of Examples 65 to 71 and additionally includes a modem configured to transmit a bitstream to a decoder device, wherein the bitstream includes the encoded data.

[00326] Example 73 includes the device of any of Examples 65 to 72, in which one or more processors are integrated into at least one of a headset, a mobile communication device, an extended reality (XR) device, or a vehicle.

[00327] Example 74 includes the device of any of Examples 65 to 73, in which one or more motion values ​​indicate one or more of linear velocity, linear acceleration, linear position, angular velocity, angular acceleration, or angular position.

[00328] Example 75 includes the device from any of Examples 65 to 74, where the compression network includes a multi-layered neural network.

[00329] Example 76 includes the device from any of Examples 65 to 75, wherein the compression network includes a video encoder, and wherein the encoder of Petition 870250078561, dated 03 / 09 / 2025, page 126 / 313 121 / 133 video has multiple encoder layers configured to encode multiple higher resolution orders of video data associated with one or more motion values.

[00330] Example 77 includes the device of any of Examples 65 to 76, additionally comprising a modem configured to transmit a bitstream to a decoder device, wherein the bitstream includes the encoded data.

[00331] Example 78 includes the device from any of Examples 65 to 77, in which one or more processors are configured to: process the conditional input using the compression network to generate attribute data; and process, using the compression network, the one or more motion values ​​and the attribute data to generate the encoded data.

[00332] Example 79 includes the device of Example 78, further comprising a modem configured to transmit a bitstream to a decoder device, wherein the bitstream includes attribute data and encoded data.

[00333] Example 80 includes the device from Example 78 or Example 79, where the attribute data includes multiscale attribute data with different spatial resolutions.

[00334] Example 81 includes the device from any of Examples 78 to 80, where the attribute data includes wavelet transform data.

[00335] Example 82 includes the device from any of Examples 65 to 81, where one or more processors are configured to: determine, based on Petition 870250078561, dated 03 / 09 / 2025, p. 127 / 313 122 / 133 a comparison of an image unit (e.g., Xb) and a previous image unit (e.g., Xa), a first motion value (e.g., ma^b) from one or more motion values; determine, based on a comparison of the image unit and a subsequent image unit (e.g., xc), a second motion value (e.g., mc^b) from one or more motion values; determine estimated motion values ​​(e.g., ma^c, mc^a) based on a comparison of a reconstructed previous image unit (e.g., xa) and a reconstructed subsequent image unit (e.g., xc); process the estimated motion values ​​to generate motion estimate attribute data corresponding to the image unit;and process, using the compression network, the motion estimate attribute data, the first motion value, the second motion value, and the image unit to generate encoded motion estimate data associated with the image unit, wherein the encoded data includes the encoded motion estimate data.

[00336] Example 83 includes the device from Example 82, wherein the one or more predicted first motion values ​​include a first reconstructed motion value (e.g., rna^b) and a second reconstructed motion value (e.g., mc^b), and wherein the one or more processors are configured to: determine a first estimated version of the image unit based on applying the first reconstructed motion value (e.g., fna^b) to a previously reconstructed image unit (e.g., Xa); determine a second estimated version of the Petition 870250078561, dated 03 / 09 / 2025, page 128 / 313 123 / 133 image unit based on applying the second reconstructed motion value (e.g., mc^b) to a subsequent reconstructed image unit (e.g., jVc); generate an estimated image unit (e.g., xb) based on a combination of the first estimated version of the image unit and the second estimated version of the image unit; process the estimated image unit (e.g., xb) to generate reconstruction attribute data; and process, using the compression network, the image unit and the reconstruction attribute data to generate coded reconstruction data associated with the image unit, wherein the coded data includes the coded reconstruction data.

[00337] Example 84 includes the device from any of Examples 65 to 81, wherein one or more processors are configured to: determine, based on a comparison of an image unit (e.g., Xd) and a previous image unit (e.g., Xc), a motion value (e.g., mc^d) from one or more motion values; determine a first estimated motion value (e.g., mb^c) based on a comparison of a first reconstructed previous image unit (e.g., Xc) and a second reconstructed previous element (e.g., xb); process at least the first estimated motion value to generate motion estimate attribute data corresponding to the image unit; and process, using the compression network, the motion estimate attribute data, the motion value, and the image unit to generate encoded motion estimate data associated with the image unit, wherein the encoded data includes the Petition 870250078561, dated 03 / 09 / 2025, page 129 / 313 124 / 133 coded motion estimate data.

[00338] Example 85 includes the device from Example 84, wherein one or more processors are configured to determine a second estimated motion value (e.g., fàa^b) based on a comparison of the second reconstructed prior element (e.g., JVb) and a third reconstructed prior element (e.g., Jva), wherein the motion estimation attribute data corresponding to the image unit are further based on processing the second estimated motion value.

[00339] Example 86 includes the device of Example 84 or Example 85, wherein the one or more predicted first motion values ​​include a first reconstructed motion value (e.g., mc^d), and wherein the one or more processors are configured to: determine an estimated image unit (e.g., xd) based on applying the first reconstructed motion value (e.g., mc^d) to the first reconstructed image unit (e.g., xc); process the estimated image unit (e.g., xd) to generate reconstruction attribute data; and process, using the compression network, the image unit and the reconstruction attribute data to generate reconstruction-encoded data associated with the image unit, wherein the encoded data includes the reconstruction-encoded data.

[00340] According to Example 87, a method includes obtaining, on a device, conditional input from a compression network, where the conditional input is based on one or more predicted first motion values; and Petition 870250078561, dated 03 / 09 / 2025, page 130 / 313 125 / 133 process, using the compression network, the conditional input and one or more motion values ​​to generate encoded data associated with the one or more motion values.

[00341] Example 88 includes the method from Example 87, additionally includes processing, on the device, the conditional input using the compression network to generate attribute data; and processing, using the compression network, the one or more motion values ​​and the attribute data to generate the encoded data.

[00342] Example 89 includes the method from Example 87 or Example 88, where the attribute data includes attribute data from multiple scales with different spatial resolutions.

[00343] Example 90 includes the method from any of Examples 87 to 89, where one or more motion values ​​are based on the output of one or more sensors.

[00344] Example 91 includes the method of Example 90, where one or more sensors include an inertial measurement unit (IMU).

[00345] Example 92 includes the method of any of Examples 87 to 91, where the one or more motion values ​​represent one or more motion vectors associated with one or more image units.

[00346] Example 93 includes the method from Example 92, where one image unit of one or more image units includes an encoding unit.

[00347] Example 94 includes the method from Example 92 or Example 93, where one image unit of one or more image units includes a block of pixels.

[00348] Example 95 includes the method of Petition 870250078561, dated 03 / 09 / 2025, pp. 131 / 313 126 / 133 any of Examples 92 to 94, where an image unit of one or more image units includes a pixel frame.

[00349] Example 96 includes the method of any of Examples 87 to 95 and additionally includes transmitting a bitstream via a modem to a decoder device, where the bitstream includes the encoded data.

[00350] Example 97 includes the method of any of Examples 87 through 96, wherein the device includes at least one of a headset, a mobile communication device, an extended reality (XR) device, or a vehicle.

[00351] Example 98 includes the method of any of Examples 87 to 97, wherein one or more motion values ​​indicate one or more of linear velocity, linear acceleration, linear position, angular velocity, angular acceleration, or angular position.

[00352] Example 99 includes the method of any of Examples 87 to 98, where the compression network includes a multi-layered neural network.

[00353] Example 100 includes the method of any of Examples 97 to 99, wherein the compression network includes a video encoder, and wherein the video encoder has multiple encoder layers configured to encode multiple higher orders of resolution of the video data associated with one or more motion values.

[00354] Example 101 includes the method of any of Examples 87 to 100, additionally including transmitting a bitstream via a modem to a decoder device, where the bitstream includes the data. Petition 870250078561, dated 03 / 09 / 2025, p. 132 / 313 127 / 133 encoded.

[00355] Example 102 includes the method of any of Examples 87 to 101 and additionally includes: processing the conditional input using the compression network to generate attribute data; and processing, using the compression network, the one or more motion values ​​and the attribute data to generate the encoded data.

[00356] Example 103 includes the method of Example 102, additionally including the transmission of a bitstream via a modem to a decoder device, wherein the bitstream includes attribute data and encoded data.

[00357] Example 104 includes the method from Example 102 or Example 103, where the attribute data includes attribute data from multiple scales with different spatial resolutions.

[00358] Example 105 includes the method from any of Examples 102 through 104, where the attribute data includes wavelet transform data.

[00359] Example 106 includes the method of any of Examples 87 to 105 and includes: determining, based on a comparison of an image unit (e.g., Xb) and a previous image unit (e.g., Xa), a first motion value (e.g., ma^b) from one or more motion values; determining, based on a comparison of the image unit and a subsequent image unit (e.g., Xc), a second motion value (e.g., rnc^b) from one or more motion values; determining estimated motion values ​​(e.g., rna^c, mc^a) based on a comparison of a reconstructed previous image unit (e.g., Xa) and a unit of Petition 870250078561, dated 03 / 09 / 2025, page 133 / 313 128 / 133 subsequent reconstructed image (e.g., Xc); process the estimated motion values ​​to generate motion estimate attribute data corresponding to the image unit; and process, using the compression network, the motion estimate attribute data, the first motion value, the second motion value, and the image unit to generate encoded motion estimate data associated with the image unit, wherein the encoded data includes the encoded motion estimate data.

[00360] Example 107 includes the method of Example 106 and additionally includes: determining a first estimated version of the image unit based on applying a first reconstructed motion value (e.g., ma^b) to a previously reconstructed image unit (e.g., xa), wherein the one or more predicted first motion values ​​include the first reconstructed motion value (e.g., ma^b); determining a second estimated version of the image unit based on applying a second reconstructed motion value (e.g., mc^b) to a subsequently reconstructed image unit (e.g., Xc), wherein the one or more predicted first motion values ​​include the second reconstructed motion value (e.g., mc^b); generating an estimated image unit (e.g., xb) based on a combination of the first estimated version of the image unit and the second estimated version of the image unit;process the estimated image unit (e.g., xb) to generate reconstruction attribute data; and process, using the compression network, the image unit and the data from; Petition 870250078561, dated 03 / 09 / 2025, page 134 / 313 129 / 133 reconstruction attribute to generate coded reconstruction data associated with the image unit, where the coded data includes the coded reconstruction data.

[00361] Example 108 includes the method of any of Examples 87 to 105 and additionally includes: determining, based on a comparison of an image unit (e.g., xd) and a previous image unit (e.g., Xc), a motion value (e.g., mc_d) from one or more motion values; determining a first estimated motion value (e.g., fàb_c) based on a comparison of a first reconstructed previous image unit (e.g., Xc) and a second reconstructed previous element (e.g., xb); processing at least the first estimated motion value to generate motion estimate attribute data corresponding to the image unit;and process, using the compression network, the motion estimate attribute data, the motion value, and the image unit to generate encoded motion estimate data associated with the image unit, wherein the encoded data includes the encoded motion estimate data.

[00362] Example 109 includes the method from Example 108 and additionally includes determining a second estimated motion value (e.g., ma_b) based on a comparison of the second reconstructed prior element (e.g., Xb) and a third reconstructed prior element (e.g., Xa), wherein the motion estimate attribute data corresponding to the image unit is additionally based on processing the second value of Petition 870250078561, dated 03 / 09 / 2025, page 135 / 313 130 / 133 estimated movement.

[00363] Example 110 includes the method of Example 108 or Example 109 and additionally includes determining an estimated image unit (e.g., x^) based on applying a first reconstructed motion value (e.g., fnc^d) to the first previously reconstructed image unit (e.g., ièc), wherein the one or more predicted first motion values ​​include the first reconstructed motion value (e.g., mc^d); processing the estimated image unit (e.g., x^) to generate reconstruction attribute data; and processing, using the compression network, the image unit and the reconstruction attribute data to generate coded reconstruction data associated with the image unit, wherein the coded data includes the coded reconstruction data.

[00364] According to Example 111, a device includes: a memory configured to store instructions; and a processor configured to execute the instructions to perform the method of any of Examples 87 to 110.

[00365] According to Example 112, a nontransient, computer-readable medium stores instructions that, when executed by a processor, cause the processor to perform the method of any of Examples 87 through 110.

[00366] According to Example 113, an apparatus includes means for carrying out the method of any of Examples 87 to 110.

[00367] According to Example 114, a nontransient, computer-readable medium includes instructions that, when executed by one or more processors, do Petition 870250078561, dated 03 / 09 / 2025, page 136 / 313 131 / 133 with which one or more processors obtain encoded data associated with one or more motion values; obtain conditional input from a compression network, where the conditional input is based on one or more predicted first motion values; and process, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values.

[00368] According to Example 115, an apparatus includes: means for obtaining encoded data associated with one or more motion values; means for obtaining conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values; and means for processing, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values.

[00369] According to Example 116, a nontransient, computer-readable medium includes instructions that, when executed by one or more processors, cause the one or more processors to obtain a conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values; and to process, using the compression network, the conditional input and one or more motion values ​​to generate encoded data associated with the one or more motion values.

[00370] According to Example 117, an apparatus includes: means for obtaining conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values; and means for processing, using the compression network, the conditional input and one or more motion values ​​to generate data. Petition 870250078561, dated 03 / 09 / 2025, page 137 / 313 132 / 133 codes associated with one or more movement values.

[00371] Those skilled in the art will further understand that the various illustrative logic blocks, configurations, modules, circuits, and algorithm steps described in connection with the implementations disclosed in the present invention may be implemented as electronic hardware, computer software executed by a processor, or combinations of both. Various illustrative components, blocks, configurations, modules, circuits, and steps have been described above in terms of their functionalities. Whether such functionality is implemented in the form of hardware or processor-executable instructions depends on the particular application and the design constraints imposed on the system as a whole. Those skilled in the art may implement the described functionality in various ways for each particular application; such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

[00372] The steps of a method or algorithm described in connection with the implementations disclosed in the present invention may be incorporated directly into hardware, in a software module executed by a processor, or in a combination of both. A software module may reside in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), or electrically erasable read-only programmable memory. Petition 870250078561, dated 03 / 09 / 2025, page 138 / 313 133 / 133 (EEPROM - electrically erasable programmable read-only memory), registers, hard disk, a removable disk, a compact disc read-only memory (CD-ROM), or any other form of non-transient storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from and write information to the storage medium. Alternatively, the storage medium may be an integral part of the processor. The processor and the storage medium may reside in an application-specific integrated circuit (ASIC). The ASIC may reside in a computing device or a user terminal. Alternatively, the processor and the storage medium may reside as distinct components in a computing device or user terminal.

[00373] The above description of the disclosed aspects is provided to enable a person skilled in the art to perform or use the disclosed aspects. Various modifications to these aspects will be readily apparent to those skilled in the art, and the principles defined in the present invention can be applied to other aspects without departing from the scope of the disclosure. Thus, the present disclosure is not intended to be limited to the aspects shown in the present invention, but should be applied to the widest possible scope consistent with the innovative principles and attributes as defined by the following claims. Petition 870250078561, dated 03 / 09 / 2025, page 139 / 313

Claims

1 / 5 CLAIMS 1. Device characterized by comprising: one or more processors configured to: obtain encoded data associated with one or more motion values; obtain conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values; and process, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values.

2. Device according to claim 1, characterized in that one or more motion values ​​are based on an output from one or more sensors.

3. Device according to claim 2, characterized in that one or more sensors include an inertial measurement unit (IMU).

4. Device, according to claim 1, characterized in that one or more motion values ​​represent one or more motion vectors associated with one or more image units.

5. Device according to claim 4, characterized in that one or more image units include an encoding unit.

6. Device according to claim 4, characterized in that one or more image units include a block of pixels.

7. Device according to claim 4, characterized in that one or more image units include a pixel frame. Petition 870250078561, dated 03 / 09 / 2025, pp. 274 / 313 2 / 5 8. Device according to claim 1, characterized in that one or more second predicted motion values ​​represent future motion vectors.

9. Device according to claim 1, characterized in that the one or more second predicted motion values ​​correspond to a reconstructed version of the one or more motion values.

10. Device according to claim 1, characterized in that one or more processors are integrated into at least one of a headset, a mobile communication device, an extended reality (XR) device, or a vehicle.

11. Device according to claim 1, characterized in that one or more motion values ​​indicate one or more of linear velocity, linear acceleration, linear position, angular velocity, angular acceleration, or angular position.

12. Device according to claim 1, characterized in that the compression network includes a multi-layered neural network.

13. Device according to claim 1, characterized in that the compression network includes a video decoder, and wherein the video decoder has multiple decoder layers configured to decode multiple resolution orders of the encoded data associated with one or more motion values.

14. Device according to claim 1, characterized in that one or more processors are configured to track an object associated with one or more motion values ​​through one or more pixel frames. Petition 870250078561, dated 03 / 09 / 2025, pp. 275 / 313 3 / 5 15. Device according to claim 14, characterized in that the one or more second predicted motion values ​​represent a collision avoidance output associated with a vehicle.

16. Device according to claim 15, characterized in that the collision avoidance output indicates a predicted future position of the vehicle relative to the object.

17. Device according to claim 15, characterized in that the collision avoidance output indicates a predicted future position of the vehicle and a predicted future position of the object.

18. Device according to claim 1, characterized by further comprising a modem configured to receive a bitstream from an encoder device, wherein the bitstream includes the encoded data.

19. A method characterized by comprising: obtaining, from a device, encoded data associated with one or more motion values; obtaining, from the device, a conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values; and processing, using the compression network, the encoded data and the conditional input to generate one or more predicted second motion values.

20. Method, according to claim 19, characterized in that the processing of the encoded data and the conditional input includes: processing the conditional input using the compression network to generate attribute data; and processing the encoded data and the attribute data to generate the one or more predicted second movement values.

21. Method according to claim 20, characterized in that the attribute data correspond to attribute data of multiple scales with different spatial resolutions.

22. Method according to claim 20, characterized in that the attribute data include multiscale wavelet transform data.

23. Device characterized by comprising: one or more processors configured to: obtain conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values; and process, using the compression network, the conditional input and one or more motion values ​​to generate encoded data associated with the one or more motion values.

24. Device according to claim 23, characterized in that one or more motion values ​​are based on an output from one or more sensors.

25. Device according to claim 24, characterized in that one or more sensors include an inertial measurement unit (IMU).

26. Device according to claim 23, characterized in that one or more motion values ​​represent one or more motion vectors associated with one or more image units.

27. Device according to claim 23, characterized by further comprising a modem configured to transmit a bitstream to a decoder device, wherein the bitstream includes the encoded data.

28. A method characterized by comprising: obtaining, in a device, a conditional input from a compression network, wherein the conditional input is based on one or more predicted first motion values; and processing, using the compression network, the conditional input and one or more motion values ​​to generate encoded data associated with the one or more motion values.

29. A method according to claim 28, characterized by further comprising: processing, in the device, the conditional input using the compression network to generate attribute data; and processing, using the compression network, the one or more motion values ​​and the attribute data to generate the encoded data.

30. Method according to claim 29, characterized in that the attribute data include attribute data from multiple scales with different spatial resolutions. Petition 870250078561, dated 03 / 09 / 2025, pp. 278 / 313

Images