Data processing method and apparatus

Abstract

A data processing method, which is applied to the field of artificial intelligence. The method comprises: acquiring images collected with regard to a target scene; on the basis of the images, performing learning to obtain a plurality of primitives of the target scene, wherein the different primitives comprise a 3D distribution of areas at different positions in a space, the 3D distribution being used for indicating target information, and the target information being at least one of semantic information, depth information or optical-flow information; and on the basis of the aggregation of some or all of the plurality of primitives, rendering a 3D model, a semantic map, a depth map or an optical-flow map of the target scene. In the embodiments of the present application, except for RGB images, information of other dimensions, such as a 3D model, a semantic map, a depth map or an optical flow map, is rendered by means of primitives, such that during parameter learning, the reconstruction and rendering effects of the RGB images can be significantly improved by means of the assistance of the information of the other dimensions (i.e., updating parameters on the basis of loss constructed on the basis of the information of the other dimensions).

Description

A data processing method and apparatus

[0001] This application claims priority to Chinese Patent Application No. 202410217743.8, filed on February 27, 2024, entitled “A Data Processing Method and Apparatus”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of artificial intelligence, and more particularly to a data processing method and apparatus thereof. Background Technology

[0003] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a branch of computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0004] 3D Gaussian splashing represents the scene as a Gaussian sphere and obtains the rendering result by projecting the Gaussian sphere, achieving real-time rendering speed. This method starts with a uniformly sampled random point cloud, learns the spherical harmonic function coefficients related to the position and viewing direction of each point, and uses a differentiable splatting-based renderer to project the point cloud onto the imaging plane according to a certain Gaussian distribution to obtain the rendering result.

[0005] However, the existing 3D Gaussian splashing method still produces poor reconstruction and rendering results. Summary of the Invention

[0006] Firstly, this application provides a data processing method, the method comprising: acquiring an image of a target scene; learning multiple primitives of the target scene based on the image, wherein different primitives include 3D distributions of regions at different locations in space, the 3D distributions being used to indicate target information, the target information being at least one of semantic information, depth information, or optical flow information; and rendering a 3D model, semantic map, depth map, or optical flow map of the target scene based on the aggregation of some or all of the multiple primitives. In the embodiments of this application, in addition to RGB images, information of other dimensions, such as 3D models, semantic maps, depth maps, or optical flow maps, is rendered through primitives. During parameter learning, the assistance of other dimensional information (i.e., updating parameters based on the loss constructed from other dimensional information) can significantly improve the reconstruction and rendering effect of RGB images.

[0007] The semantic graph can be an instance segmentation graph, a global segmentation graph, a target segmentation graph, etc., and this application embodiment is not limited to these.

[0008] In one possible implementation, the primitive is a Gaussian sphere.

[0009] In one possible implementation, the plurality of primitives includes a plurality of first primitives for static regions and a plurality of second primitives for dynamic regions in the scene. Separating the primitives for static and dynamic regions allows for joint optimization of the static background and the dynamic target.

[0010] In one possible implementation, the image specifically includes images of multiple frames of a video, the video including a moving object, and the multiple second primitives including multiple primitives for each frame; after learning the multiple primitives of the target scene based on the image, the method further includes: optimizing the multiple second primitives using the motion model of the moving object as a constraint.

[0011] In this embodiment, dynamic target reconstruction and rendering can be based on kinematic constraints. For example, dynamic foreground and background can be modeled and optimized simultaneously based on a scene graph, and kinematic models and 3D bounding box parameters can be optimized simultaneously. The position and rotation matrices of the dynamic object at each moment can be calculated by the kinematic model.

[0012] In one possible implementation, the image specifically includes images of multiple frames of a video, each frame corresponding to multiple second primitives, and the multiple second primitives corresponding to different frames are different.

[0013] In one possible implementation, the method further includes: rendering an RGB image of the target scene based on the aggregation of some or all of the plurality of primitives.

[0014] In one possible implementation, rendering the semantic map, depth map, or optical flow map of the target scene based on the aggregation of some or all of the plurality of primitives includes: aggregating the semantic information of some or all of the plurality of primitives to render the semantic map of the target scene; wherein the aggregation includes a softmax operation on the 3D semantics in the semantic information. This is contrary to most existing methods that apply softmax to the 2D semantic logarithm obtained by accumulating the non-normalized 3D semantic logarithm s_(i). Applying softmax in 2D space leads to the generation of noisy 3D semantic labels. This is because 2D space softmax can generate accurate 2D semantics by adjusting the scale of the 3D semantic logarithm, thus allowing a single sampling point with a large logit value to significantly affect the overall volume rendering result.

[0015] In one possible implementation, the method further includes: obtaining labels for the 3D model, semantic map, depth map, or optical flow map of the image; and updating the plurality of primitives based on the labels.

[0016] Secondly, this application provides a data processing apparatus, the apparatus comprising:

[0017] The acquisition module is used to acquire images of the target scene.

[0018] The processing module is configured to learn multiple primitives of the target scene based on the image, wherein different primitives include 3D distributions of regions at different locations in space, the 3D distributions being used to indicate target information, the target information being at least one of semantic information, depth information, or optical flow information; and to render a 3D model, semantic map, depth map, or optical flow map of the target scene based on the aggregation of some or all of the multiple primitives.

[0019] In one possible implementation, the primitive is a Gaussian sphere.

[0020] In one possible implementation, the plurality of primitives includes a plurality of first primitives in the static region of the scene and a plurality of second primitives in the dynamic region.

[0021] In one possible implementation, the image specifically includes images of multiple frames of a video, the video including moving objects, and the multiple second primitives including multiple primitives for each frame; after learning the multiple primitives of the target scene based on the image, the processing module is further configured to:

[0022] The motion model of the moving object is used as a constraint to optimize the plurality of second primitives.

[0023] In one possible implementation, the image specifically includes images of multiple frames of a video, each frame corresponding to multiple second primitives, and the multiple second primitives corresponding to different frames are different.

[0024] In one possible implementation, the processing module is further configured to:

[0025] The target scene's RGB image is rendered based on the aggregation of some or all of the multiple primitives.

[0026] In one possible implementation, the processing module is specifically used for:

[0027] The semantic information of some or all of the multiple primitives is aggregated to render a semantic graph of the target scene; wherein, the aggregation includes a softmax operation on the 3D semantics in the semantic information.

[0028] In one possible implementation, the processing module is further configured to:

[0029] Obtain the labels of the 3D model, semantic map, depth map, or optical flow map of the image;

[0030] The plurality of primitives are updated according to the labels.

[0031] Thirdly, embodiments of this application provide a data processing apparatus, which may include a memory, a processor, and a bus system, wherein the memory is used to store a program, and the processor is used to execute the program in the memory to perform the methods described in the first aspect above and any of its optional methods.

[0032] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when run on a computer, causes the computer to perform the methods described in the first aspect and any of its optional methods.

[0033] Fifthly, embodiments of this application provide a computer program that, when run on a computer, causes the computer to perform the first aspect and any of its optional methods described above.

[0034] Sixthly, this application provides a chip system including a processor for supporting an execution data processing device in implementing the functions involved in the foregoing aspects, such as transmitting or processing data involved in the foregoing methods; or, information. In one possible design, the chip system further includes a memory for storing program instructions and data necessary for the execution device or training device. This chip system may be composed of chips or may include chips and other discrete devices. Attached Figure Description

[0035] Figure 1A is a schematic diagram of a structural framework for artificial intelligence.

[0036] Figures 1B and 1C are schematic diagrams of the application system framework of this application;

[0037] Figure 1D is a schematic diagram of an optional hardware structure for the terminal;

[0038] Figure 2 is a schematic diagram of a server structure;

[0039] Figure 3 is a schematic diagram of a system architecture according to this application;

[0040] Figure 4 illustrates the process of a cloud service.

[0041] Figure 5 is a flowchart illustrating a data processing method provided in an embodiment of this application;

[0042] Figures 6 to 8 are schematic flowcharts of a data processing method provided in an embodiment of this application;

[0043] Figures 9, 10A, and 10B illustrate one of the beneficial effects of this application;

[0044] Figure 11 is a schematic diagram of a data processing device provided in an embodiment of this application;

[0045] Figure 12 is a schematic diagram of an execution device provided in an embodiment of this application;

[0046] Figure 13 is a schematic diagram of a training device provided in an embodiment of this application;

[0047] Figure 14 is a schematic diagram of a chip structure provided in an embodiment of this application. Detailed Implementation

[0048] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.

[0049] The embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.

[0050] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0051] The terms “substantially,” “about,” and similar terms used herein are used as approximations, not as terms of degree, and are intended to take into account the inherent biases of measurements or calculations known to those skilled in the art. Furthermore, the term “may” used in describing embodiments of this application means “one or more possible embodiments.” The terms “use,” “using,” and “used” used herein are to be considered synonymous with the terms “utilize,” “utilizing,” and “utilized,” respectively. Additionally, the term “exemplary” is intended to refer to an instance or illustration.

[0052] First, the overall workflow of an artificial intelligence system is described, as shown in Figure 1A. Figure 1A is a structural diagram of the main framework of artificial intelligence. The framework is then elaborated from two dimensions: the "intelligent information chain" (horizontal axis) and the "IT value chain" (vertical axis). The "intelligent information chain" reflects a series of processes from data acquisition to processing. For example, it could be the general process of intelligent information perception, intelligent information representation and formation, intelligent reasoning, intelligent decision-making, and intelligent execution and output. In this process, data undergoes a condensation process of "data—information—knowledge—wisdom." The "IT value chain" reflects the value that artificial intelligence brings to the information technology industry, from the underlying infrastructure of human intelligence and information (provided and processed by technology) to the industrial ecosystem of the system.

[0053] (1) Infrastructure

[0054] Infrastructure provides computing power to support artificial intelligence systems, enabling communication with the external world and providing support through a basic platform. This communication occurs through sensors; computing power is provided by intelligent chips (hardware acceleration chips such as CPUs, NPUs, GPUs, ASICs, and FPGAs); and the basic platform includes distributed computing frameworks and related platform guarantees and support, which may include cloud storage and computing, interconnected networks, etc. For example, sensors communicate with the outside world to acquire data, and this data is provided to intelligent chips in the distributed computing system provided by the basic platform for computation.

[0055] (2) Data

[0056] The data at the next layer of infrastructure is used to represent the data sources in the field of artificial intelligence. The data involves graphics, images, voice, text, and IoT data from traditional devices, including business data from existing systems and sensor data such as force, displacement, liquid level, temperature, and humidity.

[0057] (3) Data processing

[0058] Data processing typically includes methods such as data training, machine learning, deep learning, search, reasoning, and decision-making.

[0059] Among them, machine learning and deep learning can perform intelligent information modeling, extraction, preprocessing, and training on data, including symbolization and formalization.

[0060] Reasoning refers to the process in which, in a computer or intelligent system, the machine thinks and solves problems by simulating human intelligent reasoning, based on reasoning control strategies and using formalized information. Typical functions include search and matching.

[0061] Decision-making refers to the process of making decisions based on intelligent information after reasoning, and it typically provides functions such as classification, sorting, and prediction.

[0062] (4) General ability

[0063] After the data processing mentioned above, the results of the data processing can be used to form some general capabilities, such as algorithms or a general system, for example, translation, text analysis, computer vision processing, speech recognition, image recognition, etc.

[0064] (5) Smart Products and Industry Applications

[0065] Intelligent products and industry applications refer to products and applications of artificial intelligence systems in various fields. They are the encapsulation of overall artificial intelligence solutions, productizing intelligent information decision-making and realizing practical applications. Their application areas mainly include: intelligent terminals, intelligent transportation, intelligent healthcare, autonomous driving, smart cities, etc.

[0066] This application can be applied to the field of image processing in the field of artificial intelligence. The following will take image processing as an example to introduce several application scenarios that have been implemented in products.

[0067] In one possible implementation, neural radiation fields can directly synthesize 3D scenes and perform multi-view rendering by inputting a limited number of multi-view images, eliminating the need for complex 3D modeling processes and achieving the effect of obtaining planar images of objects from any viewpoint with low acquisition costs.

[0068] This application can be applied, but is not limited to, to applications with compositing 3D scenes and multi-view rendering capabilities (hereinafter referred to as new view generation applications) or cloud services provided by cloud-side servers, which will be described in detail below:

[0069] I. New Perspective Generation Applications

[0070] The product form of this application embodiment can be a new perspective generation application. This new perspective generation application can run on a terminal device or a cloud-based server.

[0071] In one possible implementation, a new perspective generation application can generate a trained model based on multiple perspective images as input and the corresponding camera intrinsics (such as camera pose information) at the time of shooting. The trained model can then output the corresponding image based on the specified perspective.

[0072] In one possible implementation, a user can open a new perspective generation application installed on a terminal device and input an image and perspective information. The new perspective generation application can process the image using the method provided in the embodiments of this application and present the processing result to the user (the presentation method may include, but is not limited to, displaying, saving, uploading to the cloud, etc.).

[0073] In one possible implementation, a user can open a new perspective generation application installed on a terminal device and input an image. The new perspective generation application can send the image to a cloud-based server. The cloud-based server processes the image using the method provided in this application embodiment and sends the processing result back to the terminal device. The terminal device can then present the processing result to the user (the presentation method may include, but is not limited to, displaying, saving, or uploading to the cloud).

[0074] The following sections will introduce the novel perspective on generating class applications in this application, focusing on both the functional architecture and the product architecture that implements the functions.

[0075] Referring to Figure 1B, which is a schematic diagram of the functional architecture of the novel perspective generation application in this embodiment of the application:

[0076] In one possible implementation, as shown in FIG1B, the new perspective generation application 102 may receive input parameters 101 (e.g., including an image) and produce a processing result 103. The new perspective generation application 102 may execute on at least one computer system (for example) and includes computer code that, when executed by one or more computers, causes the computers to perform the methods provided in the embodiments of this application.

[0077] Referring to Figure 1C, which is a schematic diagram of the entity architecture for running a new perspective generation application in an embodiment of this application:

[0078] Referring to Figure 1C, Figure 1C illustrates a system architecture diagram. The system may include a terminal 100 and a server 200. The server 200 may include one or more servers (Figure 1C illustrates this with one server as an example), and the server 200 may provide the methods provided in the embodiments of this application to one or more terminals.

[0079] The terminal 100 may have a new perspective generation application installed. The application and webpage can provide an interface. The terminal 100 can receive relevant parameters input by the user on the new perspective generation interface and send the parameters to the server 200. The server 200 can obtain the processing result based on the received parameters and return the processing result to the terminal 100.

[0080] It should be understood that in some optional implementations, the terminal 100 can also complete the action of obtaining the processing result based on the received parameters on its own, without the need for the server to cooperate. This application embodiment is not limited to this.

[0081] The product form of terminal 100 in Figure 1C will be described next;

[0082] The terminal 100 in this application embodiment can be a mobile phone, tablet computer, wearable device, vehicle device, augmented reality (AR) / virtual reality (VR) device, laptop computer, ultra-mobile personal computer (UMPC), netbook, personal digital assistant (PDA), etc., and this application embodiment does not impose any restrictions on it.

[0083] Figure 1D shows a schematic diagram of an optional hardware structure for terminal 100.

[0084] Referring to Figure 1D, terminal 100 may include components such as a radio frequency unit 110, a memory 120, an input unit 130, a display unit 140, a camera 150 (optional), an audio circuit 160 (optional), a speaker 161 (optional), a microphone 162 (optional), a processor 170, an external interface 180, and a power supply 190. Those skilled in the art will understand that Figure 1D is merely an example of a terminal or multi-functional device and does not constitute a limitation on the terminal or multi-functional device; it may include more or fewer components than illustrated, or combine certain components, or use different components.

[0085] The input unit 130 can be used to receive input numerical or character information, and to generate key signal inputs related to user settings and function control of the portable multi-functional device. Specifically, the input unit 130 may include a touchscreen 131 (optional) and / or other input devices 132. The touchscreen 131 can collect touch operations performed by the user on or near it (such as operations performed by the user using fingers, knuckles, styluses, or any suitable object on or near the touchscreen), and drive the corresponding connection devices according to a pre-set program. The touchscreen can detect the user's touch actions, convert the touch actions into touch signals and send them to the processor 170, and can receive and execute commands sent by the processor 170; the touch signal includes at least touch point coordinate information. The touchscreen 131 can provide an input interface and an output interface between the terminal 100 and the user. In addition, various types of touchscreens, such as resistive, capacitive, infrared, and surface acoustic wave, can be used to implement the touchscreen. Besides the touchscreen 131, the input unit 130 may also include other input devices. Specifically, other input devices 132 may include, but are not limited to, one or more of the following: physical keyboard, function keys (such as volume control buttons, power buttons, etc.), trackball, mouse, joystick, etc.

[0086] The input device 132 can receive input images, camera intrinsic parameters, etc.

[0087] The display unit 140 can be used to display information input by the user or information provided to the user, various menus of the terminal 100, interactive interfaces, file display, and / or playback of any multimedia file. In this embodiment, the display unit 140 can be used to display the interface and processing results of a new perspective generation application.

[0088] The memory 120 can be used to store instructions and data. The memory 120 may primarily include an instruction storage area and a data storage area. The data storage area can store various types of data, such as multimedia files and text. The instruction storage area can store software units such as operating systems, applications, and instructions required for at least one function, or subsets or extended sets thereof. It may also include non-volatile random access memory. It provides the processor 170 with hardware, software, and data resources for managing the computing device, supporting control software and applications. It is also used for storing multimedia files, as well as storing running programs and applications.

[0089] The processor 170 is the control center of the terminal 100. It connects various parts of the terminal 100 via various interfaces and lines. By running or executing instructions stored in the memory 120 and calling data stored in the memory 120, it performs various functions and processes data of the terminal 100, thereby controlling the terminal device as a whole. Optionally, the processor 170 may include one or more processing units; preferably, the processor 170 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 170. In some embodiments, the processor and memory can be implemented on a single chip; in some embodiments, they can also be implemented separately on independent chips. The processor 170 can also be used to generate corresponding operation control signals, send them to the corresponding components of the computing processing device, read and process data in the software, especially read and process data and programs in the memory 120, so that the various functional modules therein perform corresponding functions, thereby controlling the corresponding components to act according to the instructions.

[0090] The memory 120 can be used to store software code related to the data processing method, and the processor 170 can execute the steps of the chip's data processing method, and can also schedule other units (such as the above-mentioned input unit 130 and display unit 140) to achieve the corresponding functions.

[0091] The radio frequency unit 110 (optional) can be used for receiving and transmitting signals during information transmission or calls. For example, it can receive downlink information from the base station and process it for the processor 170; additionally, it can transmit uplink data to the base station. Typically, the RF circuit includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low-noise amplifier (LNA), a duplexer, etc. Furthermore, the radio frequency unit 110 can also communicate wirelessly with network devices and other devices. This wireless communication can use any communication standard or protocol, including but not limited to Global System for Mobile communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, Short Messaging Service (SMS), etc.

[0092] In this embodiment of the application, the radio frequency unit 110 can send an image to the server 200 and receive the processing result sent by the server 200.

[0093] It should be understood that the radio frequency unit 110 is optional and can be replaced with other communication interfaces, such as a network port.

[0094] The terminal 100 also includes a power supply 190 (such as a battery) that supplies power to various components. Preferably, the power supply can be logically connected to the processor 170 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system.

[0095] Terminal 100 also includes an external interface 180, which can be a standard Micro USB interface or a multi-pin connector, which can be used to connect terminal 100 to other devices for communication or to connect a charger to charge terminal 100.

[0096] Although not shown, terminal 100 may also include a flash, a wireless fidelity (WiFi) module, a Bluetooth module, and sensors with various functions, which will not be described in detail here. Some or all of the methods described below can be applied to terminal 100 as shown in Figure 1D.

[0097] The product form of server 200 in Figure 1C is described below;

[0098] Figure 2 provides a schematic diagram of the structure of a server 200. As shown in Figure 2, the server 200 includes a bus 201, a processor 202, a communication interface 203, and a memory 204. The processor 202, the memory 204, and the communication interface 203 communicate with each other via the bus 201.

[0099] Bus 201 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, only one thick line is used in Figure 2, but this does not indicate that there is only one bus or one type of bus.

[0100] The processor 202 can be any one or more of the following processors: central processing unit (CPU), graphics processing unit (GPU), microprocessor (MP), or digital signal processor (DSP).

[0101] Memory 204 may include volatile memory, such as random access memory (RAM). Memory 204 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0102] The memory 204 can be used to store software code related to the data processing method, and the processor 202 can execute the steps of the chip's data processing method, and can also schedule other units to achieve corresponding functions.

[0103] It should be understood that the aforementioned terminal 100 and server 200 can be centralized or distributed devices. The processors (e.g., processor 170 and processor 202) in the aforementioned terminal 100 and server 200 can be hardware circuits (such as application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), general-purpose processors, digital signal processors (DSPs), microprocessors or microcontrollers, etc.) or combinations of these hardware circuits. For example, the processor can be a hardware system with instruction execution capabilities, such as a CPU or DSP, or a hardware system without instruction execution capabilities, such as an ASIC or FPGA, or a combination of the aforementioned hardware systems without instruction execution capabilities and hardware systems with instruction execution capabilities.

[0104] It should be understood that the steps related to the model inference process in the embodiments of this application involve AI-related operations. When performing AI operations, the instruction execution architecture of the terminal device and the server is not limited to the processor-memory architecture described above. The system architecture provided in the embodiments of this application will be described in detail below with reference to Figure 3.

[0105] Figure 3 is a schematic diagram of the system architecture provided in an embodiment of this application. As shown in Figure 3, the system architecture 500 includes an execution device 510, a training device 520, a database 530, a client device 540, a data storage system 550, and a data acquisition system 560.

[0106] The execution device 510 includes a calculation module 511, an I / O interface 512, a preprocessing module 513, and a preprocessing module 514. The calculation module 511 may include a target model / rule 501, while the preprocessing modules 513 and 514 are optional.

[0107] Among them, the execution device 510 can be a terminal device or server that runs the above-mentioned new perspective generation application.

[0108] The data acquisition device 560 is used to acquire training samples. Training samples can be multiple images, etc. After acquiring the training samples, the data acquisition device 560 stores these training samples in the database 530.

[0109] The training device 520 can maintain training samples in the database 530 and obtain the target model / rule 501 from the neural network to be trained (e.g., the neural radiance field (NeRF) in the embodiments of this application).

[0110] It should be noted that in practical applications, the training samples maintained in database 530 may not all come from the data acquisition device 560; they may also be received from other devices. Furthermore, it should be noted that training device 520 may not necessarily train the target model / rule 501 entirely based on the training samples maintained in database 530; it may also obtain training samples from the cloud or other sources for model training. The above description should not be construed as limiting the embodiments of this application.

[0111] The target model / rule 501 trained by the training device 520 can be applied to different systems or devices, such as the execution device 510 shown in Figure 3. The execution device 510 can be a terminal, such as a mobile terminal, tablet computer, laptop computer, augmented reality (AR) / virtual reality (VR) device, vehicle terminal, etc., or it can be a server, etc.

[0112] Specifically, the training device 520 can transfer the trained model to the execution device 510.

[0113] In Figure 3, the execution device 510 is configured with an input / output (I / O) interface 512 for data interaction with external devices. Users can input data (such as images in this embodiment) into the I / O interface 512 through the client device 540.

[0114] Preprocessing modules 513 and 514 are used to preprocess the input data received from the I / O interface 512. It should be understood that preprocessing modules 513 and 514 may be absent, or only one preprocessing module may be used. When preprocessing modules 513 and 514 are absent, the calculation module 511 can be used directly to process the input data.

[0115] During the preprocessing of input data by the execution device 510, or during the calculation module 511 of the execution device 510 performing calculations and other related processes, the execution device 510 can call data, code, etc. in the data storage system 550 for corresponding processing, or store the data, instructions, etc. obtained from the corresponding processing into the data storage system 550.

[0116] Finally, the I / O interface 512 provides the processing result to the client device 540, thereby providing it to the user.

[0117] In the scenario shown in Figure 3, the user can manually provide input data, which can be done through the interface provided by I / O interface 512. Alternatively, the client device 540 can automatically send input data to I / O interface 512. If user authorization is required for the client device 540 to automatically send input data, the user can set the corresponding permissions in the client device 540. The user can view the output results of the execution device 510 on the client device 540, which can be presented in various forms such as display, sound, or animation. The client device 540 can also act as a data acquisition terminal, collecting the input data and output results of the input I / O interface 512 as shown in the figure, and storing them as new sample data in database 530. Alternatively, data can be collected directly from the I / O interface 512 without going through the client device 540, using the input data and output results of the input I / O interface 512 as shown in the figure, and storing them as new sample data in database 530.

[0118] It is worth noting that Figure 3 is merely a schematic diagram of a system architecture provided in an embodiment of this application. The positional relationships between the devices, components, modules, etc., shown in the figure do not constitute any limitation. For example, in Figure 3, the data storage system 550 is an external memory relative to the execution device 510. In other cases, the data storage system 550 can also be placed in the execution device 510. It should be understood that the aforementioned execution device 510 can be deployed in the client device 540.

[0119] From the inference side of the model:

[0120] In this embodiment, the computing module 511 of the execution device 510 can obtain the code stored in the data storage system 550 to implement the steps related to the model reasoning process in this embodiment.

[0121] In this embodiment of the application, the computing module 511 of the execution device 510 may include hardware circuits (such as application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), general-purpose processors, digital signal processors (DSPs), microprocessors or microcontrollers, etc.) or combinations of these hardware circuits. For example, the training device 520 may be a hardware system with instruction execution capabilities, such as a CPU or DSP, or a hardware system without instruction execution capabilities, such as an ASIC or FPGA, or a combination of the aforementioned hardware systems without instruction execution capabilities and hardware systems with instruction execution capabilities.

[0122] Specifically, the computing module 511 of the execution device 510 can be a hardware system with the function of executing instructions. The steps related to the model inference process provided in this application embodiment can be software code stored in the memory. The computing module 511 of the execution device 510 can obtain the software code from the memory and execute the obtained software code to implement the steps related to the model inference process provided in this application embodiment.

[0123] It should be understood that the computing module 511 of the execution device 510 can be a combination of a hardware system without the function of executing instructions and a hardware system with the function of executing instructions. Some steps related to the model reasoning process provided in the embodiments of this application can also be implemented by the hardware system in the computing module 511 of the execution device 510 without the function of executing instructions, which is not limited here.

[0124] From the training side of the model:

[0125] In this embodiment, the training device 520 can obtain the code stored in the memory (not shown in Figure 3, which can be integrated into the training device 520 or deployed separately from the training device 520) to implement the steps related to model training in this embodiment.

[0126] In this embodiment of the application, the training device 520 may include hardware circuits (such as application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), general-purpose processors, digital signal processors (DSPs), microprocessors or microcontrollers, etc.) or combinations of these hardware circuits. For example, the training device 520 may be a hardware system with instruction execution capabilities, such as a CPU or DSP, or a hardware system without instruction execution capabilities, such as an ASIC or FPGA, or a combination of the aforementioned hardware systems without instruction execution capabilities and hardware systems with instruction execution capabilities.

[0127] It should be understood that the training device 520 can be a combination of a hardware system without the function of executing instructions and a hardware system with the function of executing instructions. Some steps related to the training of the neutralization model provided in the embodiments of this application can also be implemented by the hardware system in the training device 520 without the function of executing instructions, which is not limited here.

[0128] II. New Perspectives Provided by Servers for Generating Cloud-like Services:

[0129] In one possible implementation, the server can provide new perspective generation services to the client side through an application programming interface (API).

[0130] In this process, the terminal device can send relevant parameters (such as images from multiple perspectives and corresponding camera intrinsic information) to the server through the API provided by the cloud. The server can obtain the processing results based on the received parameters and return the processing results to the terminal.

[0131] The description of the terminal and server can be found in the above embodiments, and will not be repeated here.

[0132] Figure 4 illustrates the process of generating cloud-like services using a new perspective provided by a cloud platform.

[0133] 1. Activate and purchase content moderation services.

[0134] 2. Users can download the software development kit (SDK) corresponding to the content moderation service. Cloud platforms usually provide multiple development versions of the SDK for users to choose from according to their development environment needs, such as JAVA version SDK, Python version SDK, PHP version SDK, Android version SDK, etc.

[0135] 3. After downloading the corresponding version of the SDK to their local machine according to their needs, users can import the SDK project into their local development environment, configure and debug it in the local development environment, and develop other functions in the local development environment, thus forming an application that integrates new perspective class generation capabilities.

[0136] 4. When a new perspective generation application is used, it can trigger an API call for new perspective generation when needed. When the application triggers the new perspective generation function, it initiates an API request to the running instance of the new perspective generation service in the cloud environment. The API request carries the image and camera intrinsic parameters, and the running instance in the cloud environment processes the image to obtain the processing result.

[0137] 5. The cloud environment returns the processing result to the application, thereby completing one method call provided in this embodiment of the application.

[0138] The embodiments of this application can be applied to most fields requiring image generation, such as autonomous driving sensor simulation, AR / VR, high-precision surveying and mapping, and 3D scene reconstruction. Furthermore, this application can also transfer the image data dimension to other types of camera data, such as depth cameras and event cameras, and should yield comparable experimental results.

[0139] Since the embodiments of this application involve a large number of neural network applications, for ease of understanding, the relevant terms and concepts such as neural networks involved in the embodiments of this application will be introduced below.

[0140] (1) Neural Network

[0141] A neural network can be composed of neural units, which can be defined as a computational unit that takes xs (i.e., input data) and an intercept of 1 as input. The output of this computational unit can be:

[0142] Where s = 1, 2, ..., n, where n is a natural number greater than 1, Ws is the weight of xs, and b is the bias of the neural unit. f is the activation function of the neural unit, used to introduce nonlinear characteristics into the neural network to convert the input signal in the neural unit into an output signal. The output signal of this activation function can be used as the input of the next convolutional layer, and the activation function can be the sigmoid function. A neural network is a network formed by connecting multiple of the above-mentioned individual neural units together, that is, the output of one neural unit can be the input of another neural unit. The input of each neural unit can be connected to the local receptive field of the previous layer to extract the features of the local receptive field, which can be a region composed of several neural units.

[0143] (2) Deep Neural Networks

[0144] A Deep Neural Network (DNN) can be understood as a neural network with many hidden layers. There's no specific metric for "many" layers; the commonly used terms "multi-layer neural network" and "deep neural network" are essentially the same thing. DNNs can be categorized into three layers based on their position: input layer, hidden layer, and output layer. Generally, the first layer is the input layer, the last layer is the output layer, and the layers in between are hidden layers. All layers are fully connected, meaning that any neuron in the i-th layer is connected to any neuron in the (i+1)-th layer. Although DNNs seem complex, the operation of each layer is actually not complicated; it can be simply described by the following linear relationship expression: in, It is the input vector. It is the output vector. α is the offset vector, W is the weight matrix (also called coefficients), and α() is the activation function. Each layer is simply an adjustment of the input vector. The output vector is obtained through such a simple operation. Because DNNs have many layers, the coefficients W and the offset vector... The number is quite large. So, how are the specific parameters defined in a DNN? Let's first look at the definition of the coefficient W. Taking a three-layer DNN as an example, the linear coefficient from the 4th neuron in the second layer to the 2nd neuron in the third layer is defined as... The superscript 3 represents the layer number where coefficient W resides, while the subscript corresponds to the output third layer index 2 and the input second layer index 4. In summary, the coefficients from the k-th neuron in layer L-1 to the j-th neuron in layer L are defined as... Note that the input layer does not have a W parameter. In deep neural networks, more hidden layers allow the network to better represent complex real-world situations. Theoretically, the more parameters a model has, the higher its complexity and "capacity," meaning it can accomplish more complex learning tasks.

[0145] (3) A Convolutional Neural Network (CNN) is a deep neural network with a convolutional structure. A CNN contains a feature extractor consisting of convolutional layers and subsampling layers. This feature extractor can be viewed as a filter, and the convolution process can be seen as using a trainable filter to convolve with an input image or a convolutional feature map. A convolutional layer refers to the neuron layer in a CNN that performs convolution processing on the input signal. In a convolutional layer of a CNN, a neuron can be connected to only some of the neurons in its neighboring layers. A convolutional layer typically contains several feature planes, each of which can be composed of rectangularly arranged neural units. Neural units on the same feature plane share weights, which are the convolutional kernels. Shared weights can be understood as the way image information is extracted regardless of location. The underlying principle is that the statistical information of one part of the image is the same as that of other parts. This means that image information learned in one part can also be used in another part. Therefore, the same learned image information can be used for all locations on the image. In the same convolutional layer, multiple convolutional kernels can be used to extract different image information. Generally, the more convolutional kernels there are, the richer the image information reflected by the convolution operation.

[0146] Convolutional kernels can be initialized as matrices of random size, and during the training of a convolutional neural network, they can learn appropriate weights. Furthermore, sharing weights directly reduces the number of connections between layers in the convolutional neural network, while also lowering the risk of overfitting.

[0147] (4) Backpropagation algorithm

[0148] Convolutional neural networks can employ backpropagation (BP) to correct the parameters in the initial super-resolution model during training, thereby reducing the reconstruction error loss. Specifically, forward propagation of the input signal to the output generates an error loss; this error loss information is then propagated back to update the parameters in the initial super-resolution model, leading to convergence of the error loss. The backpropagation algorithm is an error-loss-driven backpropagation process aimed at obtaining the optimal parameters of the super-resolution model, such as the weight matrix.

[0149] (5) Loss Function

[0150] In training a deep neural network, to ensure the output closely approximates the desired predicted value, we compare the network's prediction with the target value. Based on the difference, we update the weight vector of each layer (usually pre-configuring parameters before the initial update). For example, if the prediction is too high, the weight vector is adjusted to predict a lower value. This adjustment continues until the deep neural network predicts the target value or a value very close to it. Therefore, we need to predefine "how to compare the difference between the predicted and target values," which is the loss function or objective function. These are important equations used to measure the difference between the predicted and target values. Taking the loss function as an example, a higher output value (loss) indicates a greater difference, and training the deep neural network becomes a process of minimizing this loss.

[0151] (6) 3D Gaussian Splatting is an emerging method for scene reconstruction and image rendering. Compared with neural rendering methods, splatting abandons the process of emitting a ray from each pixel and sampling a large number of rays. Instead, it represents the scene as a Gaussian sphere and obtains the rendering result by projecting the Gaussian sphere, achieving real-time rendering speed. By shifting from volume-based representation to learning point representation, it improves the memory and runtime performance of existing methods by more than an order of magnitude in terms of training and inference. This method starts with a uniformly sampled random point cloud, learns the spherical harmonic function coefficients related to the position and viewing direction of each point, and uses a differentiable splatting-based renderer to project the point cloud onto the imaging plane according to a certain Gaussian distribution to obtain the rendering result. This is used to train the model to reproduce a set of input training images with a given camera pose.

[0152] (7) Multilayer perceptron, also called artificial neural network (ANN), has multiple hidden layers in addition to the input and output layers. The simplest MLP contains only one hidden layer.

[0153] 3D Gaussian splashing represents the scene as a Gaussian sphere and obtains the rendering result by projecting the Gaussian sphere, achieving real-time rendering speed. This method starts with a uniformly sampled random point cloud, learns the spherical harmonic function coefficients related to the position and viewing direction of each point, and uses a differentiable splatting-based renderer to project the point cloud onto the imaging plane according to a certain Gaussian distribution to obtain the rendering result.

[0154] However, the existing 3D Gaussian splashing method still produces poor reconstruction and rendering results.

[0155] To address the aforementioned problems, embodiments of this application provide a data processing method. The data processing method of this application embodiment will be described in detail below with reference to the accompanying drawings.

[0156] Referring to Figure 5, which is a flowchart of a data processing method provided in an embodiment of this application, the data processing method provided in this application may include steps 501 to 503, which will be described in detail below.

[0157] 501. Acquire images for the target scene;

[0158] In one possible implementation, an image can be acquired. The image can be a picture captured by a camera, or one or more frames from a captured video. The video can be taken from multiple perspectives of the same scene. The scene can be a dynamic scene, that is, a scene with moving objects.

[0159] In one possible implementation, the image may be a pre-taken picture stored in the terminal device's image library or downloaded via the Internet. This application does not limit the specific content of the image.

[0160] 502. Based on the image, learn multiple primitives of the target scene, wherein different primitives include 3D distributions of regions at different locations in space, and the 3D distributions are used to indicate target information, wherein the target information is at least one of semantic information, depth information, or optical flow information.

[0161] In one possible implementation, multiple primitives of the target scene can be learned from the image. These primitives can include a 3D distribution of regions at different locations in space. For example, the primitives could be Gaussian spheres.

[0162] In one possible implementation, the plurality of primitives includes a plurality of first primitives in the static region of the scene and a plurality of second primitives in the dynamic region.

[0163] In one possible implementation, the image specifically includes images of multiple frames of a video, each frame corresponding to multiple second primitives, and the multiple second primitives corresponding to different frames are different. That is, a fixed 3D distribution can be obtained for static regions, and a dynamic 3D distribution can be obtained for dynamic regions that changes with time frames.

[0164] Taking a Gaussian distribution in 3D as an example, assuming the scene consists of a static region and N rigid dynamic objects, a complete 3D scene comprises a static background region model and N dynamic object models. The static region is represented using a static Gaussian in the world coordinate system. Each dynamic object uses a dynamic Gaussian in the normal coordinate system and a set of rigid transformations. Model the data, where t represents the timestamp.

[0165] 3D Gaussians can be used to model static and dynamic regions. Each Gaussian consists of a 3D covariance matrix ∑∈R. 3×3 and a 3D position μ∈R 3 And an opacity α∈R + definition:

[0166] Each Gaussian represents a color vector c∈R 3 , parameterized as SH coefficients.

[0167] In one possible implementation, the 3D distribution is used to indicate target information, which is at least one of semantic information, depth information, or optical flow information. Specifically, it can be based on the distribution of each primitive (e.g., the semantic logarithm s∈R of a 3D Gaussian). S Additional modeling can be performed to allow rendering of semantic graphs (e.g., 2D semantic labels). Furthermore, additional modeling based on the distribution of each primitive can be performed to allow rendering of optical flow graphs; for example, 3D positions μ can be projected onto the image space at two different time stamps (t1 and t2) and motion calculated to obtain rendered optical flow.

[0168] In one possible implementation, the image specifically includes images of multiple frames of a video, the video including moving objects (or dynamic objects), and the multiple second primitives including multiple primitives for each frame; after learning the multiple primitives of the target scene based on the image, the motion model of the moving object can be used as a constraint to optimize the multiple second primitives. In this embodiment, dynamic target reconstruction and rendering can be based on kinematic constraints. For example, referring to Figure 7, dynamic foreground and background can be simultaneously modeled and optimized based on a 3D Gaussian Splatting scene graph, introducing kinematic models and 3D bounding box parameters for simultaneous optimization. The position and rotation matrix of the dynamic object at each moment can be calculated by the kinematic model.

[0169] 503. Based on the aggregation of some or all of the primitives among the plurality of primitives, render the 3D model, semantic map, depth map or optical flow map of the target scene.

[0170] In this embodiment, in addition to RGB images, other dimensions of information, such as 3D models, semantic maps, depth maps, or optical flow maps, can be rendered using primitives. During parameter learning, the reconstruction and rendering effects of RGB images can be significantly improved by using the assistance of other dimensions of information (that is, updating parameters based on the loss constructed from other dimensions of information).

[0171] In one possible implementation, the RGB image of the target scene can also be rendered based on the aggregation of some or all of the multiple primitives.

[0172] For example, taking a Gaussian distribution as an example, for the RGB image branch, the combination of static and dynamic Gaussian distributions can be sorted and projected onto the image plane using α mixing:

[0173] In the above formula, α′ j It is determined by the 2D Gaussian and 3D opacity α of the projection.

[0174] Compared to single-object scenes, urban scenes typically involve more complex lighting conditions, and images are often captured using automatic white balance and automatic exposure. NeRF-based methods typically input the appearance embedding of each frame along with the 3D location into a neural network to calculate color, thereby compensating for exposure. However, when using 3D Gaussian, no neural network is capable of handling appearance embeddings. Embodiments of this application can map the camera's extrinsic parameters to an affine matrix A∈R using a task network (e.g., an MLP). 3×3 and vector b∈R 3 This generates an exposure affine matrix for each camera: Modeling the exposure can improve the rendering quality of the experimental portion.

[0175] In one possible implementation, the semantic information of some or all of the multiple primitives can be aggregated to render a semantic graph of the target scene; wherein, the aggregation includes a softmax operation on the 3D semantics in the semantic information.

[0176] For example, taking a Gaussian distribution as an example, for semantic branches: 2D semantic labels can be obtained through α-mixing based on the 3D semantic logit s using the semantic information rendering formula:

[0177] Before α mixing, the 3D semantic logarithm s can be analyzed. i Performing a softmax operation, which is similar to how most existing methods apply softmax by accumulating the non-normalized 3D semantic logarithm s i The resulting 2D semantic logarithm is the opposite. Applying softmax in 2D space leads to the generation of noisy 3D semantic labels. This is because 2D space softmax can generate accurate 2D semantics by adjusting the scale of the 3D semantic logarithm, allowing a single sample point with a large logit value to significantly affect the overall volume rendering result. For example, although the target rendering label is "tree", unwanted floating-point labels like "car" may not be penalized as long as there is a 3D Gaussian spectrum along this ray that provides a large logit value for "tree". Conversely, embodiments of this application can eliminate such floating-point labels by normalizing the logarithm in 3D space.

[0178] For example, taking a Gaussian distribution as an example, for the optical flow branch: 3D Gaussian representation can also render optical flow. Given two timestamps (t1 and t2), the optical flow at each 3D Gaussian center μ is first calculated as follows: Specifically, μ can be projected onto the 2D image space based on the camera's intrinsic and extrinsic parameters to obtain:

[0179] The motion vector is calculated based on the above formula. Next, the optical flow map is rendered using the optical flow rendering formula based on volume rendering:

[0180] This rendering process assumes that any pixel in a 2D Gaussian splash shares the same optical flow direction with the corresponding Gaussian center, but with a scaling factor.

[0181] In one possible implementation, during parameter learning, labels for the 3D model, semantic map, depth map, or optical flow map of the image can be obtained; based on the labels, the multiple primitives are updated. This application proposes a complete large-scale scene understanding framework based on 3D Gaussian Spatting, achieving joint optimization of multiple branches such as RGB, semantics, optical flow, and moving targets, without relying on manual annotation of moving targets, and realizing joint optimization of static background and dynamic targets based on 3D-GS.

[0182] The following is a specific implementation process:

[0183] Referring to Figure 6, which is a schematic diagram of an application architecture of this application embodiment, this application can obtain the semantic segmentation image sequence, optical flow image prediction sequence, and 3D detection tracking box pseudo-label sequence corresponding to the input image based on a pre-trained model; given an input RGB image, a three-dimensional color point cloud is obtained using the SFM (structure-from-motion) model, and a background 3D Gaussian sphere is initialized; a 3D Gaussian sphere is randomly initialized within the predicted 3D detection box, and a foreground 3D Gaussian sphere is placed according to the position of the 3D detection box; the position, velocity, direction, acceleration, and angular velocity parameters of the dynamic vehicle at each moment are initialized based on the predicted 3D detection box; the kinematic model for the dynamic target optimization process can be a unicycle model or a bicycle model. Figure 8 is a schematic diagram of the dynamic target optimization process based on the unicycle model:

[0184] The unicycle model is shown in the following formula: θ t+1 =θ t +w t ;

[0185] Here, v t w represents forward speed. t It refers to angular velocity. Compared to directly and independently optimizing the transformation of a dynamic vehicle in each frame, this model integrates physical constraints, enabling smoother motion modeling of moving objects and making them less prone to local minima.

[0186] Although an initial state (x1, y1, θ1) can be defined and v can be recursively derived based on velocity. t and w t However, this recursive parameterization is difficult to optimize. In practice, I defined a set of trainable states. and trainable speed We added a regularization term to ensure the vehicle's state conformed to the characteristics of the unicycle model in the formula. Additionally, we adjusted the vehicle's vertical position. The model is designed to be optimized parameters.

[0187] Render all foreground and background 3D Gaussian spheres to obtain the model's predicted RGB, semantic, and optical flow maps; in addition, the depth of each 3D Gaussian sphere can be rendered using a similar volume rendering method to obtain a depth map; extract the center point of the 3D Gaussian sphere to obtain the semantic point cloud.

[0188] The embodiments of this application achieve state-of-the-art (SOTA) image rendering quality index (PSNR) in the field of new perspective image synthesis; the scene geometry estimation capability is greatly improved compared to existing technologies; and the cost is lower, the image fidelity is higher, and the scene diversity is better compared to CG image synthesis methods. For example, refer to Figures 9, 10A, and 10B, where Figure 9 is a schematic diagram of a static scene, Figure 10A is a schematic diagram of a dynamic scene, and Figure 10B is a schematic diagram of the 3D geometry of the scene.

[0189] Referring to Figure 11, which is a schematic diagram of the structure of a data processing apparatus provided in an embodiment of this application, as shown in Figure 11, the data processing apparatus 1100 provided in this embodiment includes:

[0190] The acquisition module 1101 is used to acquire images of the target scene.

[0191] The specific description of the acquisition module 1101 can be found in the description of step 501 in the above embodiments, and will not be repeated here.

[0192] The processing module 1102 is configured to learn multiple primitives of the target scene based on the image, wherein different primitives include 3D distributions of regions at different locations in space, the 3D distributions being used to indicate target information, the target information being at least one of semantic information, depth information, or optical flow information; and to render a 3D model, semantic map, depth map, or optical flow map of the target scene based on the aggregation of some or all of the multiple primitives.

[0193] The specific description of the processing module 1102 can be found in the description of steps 502 to 503 in the above embodiments, and will not be repeated here.

[0194] In one possible implementation, the primitive is a Gaussian sphere.

[0195] In one possible implementation, the plurality of primitives includes a plurality of first primitives in the static region of the scene and a plurality of second primitives in the dynamic region.

[0196] In one possible implementation, the image specifically includes images of multiple frames of a video, the video including moving objects, and the multiple second primitives including multiple primitives for each frame; after learning the multiple primitives of the target scene based on the image, the processing module 1102 is further configured to:

[0197] The motion model of the moving object is used as a constraint to optimize the plurality of second primitives.

[0198] In one possible implementation, the image specifically includes images of multiple frames of a video, each frame corresponding to multiple second primitives, and the multiple second primitives corresponding to different frames are different.

[0199] In one possible implementation, the processing module 1102 is further configured to:

[0200] The target scene's RGB image is rendered based on the aggregation of some or all of the multiple primitives.

[0201] In one possible implementation, the processing module 1102 is specifically used for:

[0202] The semantic information of some or all of the multiple primitives is aggregated to render a semantic graph of the target scene; wherein, the aggregation includes a softmax operation on the 3D semantics in the semantic information.

[0203] In one possible implementation, the processing module 1102 is further configured to:

[0204] Obtain the labels of the 3D model, semantic map, depth map, or optical flow map of the image;

[0205] The plurality of primitives are updated according to the labels.

[0206] The following describes an execution device provided in an embodiment of this application. Please refer to Figure 12, which is a schematic diagram of the structure of an execution device provided in an embodiment of this application. The execution device 1200 can specifically be a virtual reality (VR) device, a mobile phone, a tablet, a laptop, a smart wearable device, a monitoring data processing device, or a server, etc., and is not limited here. Specifically, the execution device 1200 includes: a receiver 1201, a transmitter 1202, a processor 1203, and a memory 1204 (the number of processors 1203 in the execution device 1200 can be one or more; Figure 12 shows one processor as an example). The processor 1203 may include an application processor 12031 and a communication processor 12032. In some embodiments of this application, the receiver 1201, transmitter 1202, processor 1203, and memory 1204 can be connected via a bus or other means.

[0207] Memory 1204 may include read-only memory and random access memory, and provides instructions and data to processor 1203. A portion of memory 1204 may also include non-volatile random access memory (NVRAM). Memory 1204 stores processor and operation instructions, executable modules, or data structures, or subsets thereof, or extended sets thereof, wherein the operation instructions may include various operation instructions for implementing various operations.

[0208] Processor 1203 controls the operation of the execution device. In specific applications, the various components of the execution device are coupled together through a bus system, which may include not only the data bus, but also power buses, control buses, and status signal buses. However, for clarity, all buses in the diagram are referred to as the bus system.

[0209] The methods disclosed in the embodiments of this application can be applied to or implemented by the processor 1203. The processor 1203 can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by the integrated logic circuits in the hardware of the processor 1203 or by instructions in software form. The processor 1203 can be a general-purpose processor, a digital signal processor (DSP), a microprocessor, or a microcontroller, and may further include an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The processor 1203 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the field, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 1204. Processor 1203 reads the information from memory 1204 and, in conjunction with its hardware, completes the steps involved in the model inference process described above.

[0210] Receiver 1201 can be used to receive input digital or character information, and to generate signal inputs related to the settings and function control of the execution device. Transmitter 1202 can be used to output digital or character information through the first interface; transmitter 1202 can also be used to send instructions to the disk group through the first interface to modify the data in the disk group; transmitter 1202 may also include a display device such as a display screen.

[0211] This application embodiment also provides a training device. Referring to Figure 13, Figure 13 is a schematic diagram of a training device provided in this application embodiment. Specifically, the training device 1300 is implemented by one or more servers. The training device 1300 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 1313 (e.g., one or more processors) and a memory 1332, and one or more storage media 1330 (e.g., one or more mass storage devices) for storing application programs 1342 or data 1344. The memory 1332 and storage media 1330 can be temporary or persistent storage. The program stored in the storage media 1330 may include one or more modules (not shown in the figure), each module may include a series of instruction operations on the training device. Furthermore, the CPU 1313 may be configured to communicate with the storage media 1330 and execute the series of instruction operations in the storage media 1330 on the training device 1300.

[0212] The training device 1300 may also include one or more power supplies 1326, one or more wired or wireless network interfaces 1350, one or more input / output interfaces 1358; or, one or more operating systems 1341, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0213] In this embodiment, the central processing unit 1313 is used to perform actions related to model training in the above embodiments.

[0214] This application also provides a computer program product that, when run on a computer, causes the computer to perform steps as performed by the aforementioned execution device, or causes the computer to perform steps as performed by the aforementioned training device.

[0215] This application also provides a computer-readable storage medium storing a program for signal processing, which, when run on a computer, causes the computer to perform steps as performed by the aforementioned execution device, or causes the computer to perform steps as performed by the aforementioned training device.

[0216] The execution device, training device, or terminal device provided in this application embodiment can specifically be a chip. The chip includes a processing unit and a communication unit. The processing unit can be, for example, a processor, and the communication unit can be, for example, an input / output interface, pins, or circuits. The processing unit can execute computer execution instructions stored in the storage unit to cause the chip within the execution device to execute the data processing method described in the above embodiments, or to cause the chip within the training device to execute the data processing method described in the above embodiments. Optionally, the storage unit can be a storage unit within the chip, such as a register or cache. Alternatively, the storage unit can be a storage unit located outside the chip within the wireless access device, such as a read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions, such as random access memory (RAM).

[0217] Specifically, please refer to Figure 14, which is a schematic diagram of a chip structure provided in an embodiment of this application. The chip can be represented as a neural network processor (NPU) 1400. The NPU 1400 is mounted as a coprocessor on the host CPU, and tasks are assigned by the host CPU. The core part of the NPU is the arithmetic circuit 1403, which is controlled by the controller 1404 to extract matrix data from the memory and perform multiplication operations.

[0218] In some implementations, the arithmetic circuit 1403 internally includes multiple processing engines (PEs). In some implementations, the arithmetic circuit 1403 is a two-dimensional pulsating array. The arithmetic circuit 1403 can also be a one-dimensional pulsating array or other electronic circuitry capable of performing mathematical operations such as multiplication and addition. In some implementations, the arithmetic circuit 1403 is a general-purpose matrix processor.

[0219] For example, suppose we have an input matrix A, a weight matrix B, and an output matrix C. The arithmetic circuit retrieves the corresponding data of matrix B from the weight memory 1402 and caches it in each PE of the arithmetic circuit. The arithmetic circuit retrieves the data of matrix A from the input memory 1401 and performs matrix operations with matrix B. The partial result or the final result of the obtained matrix is ​​stored in the accumulator 1408.

[0220] Unified memory 1406 is used to store input and output data. Weight data is directly transferred to weight memory 1402 via Direct Memory Access Controller (DMAC) 1405. Input data is also transferred to unified memory 1406 via DMAC.

[0221] BIU stands for Bus Interface Unit, which is used for interaction between the AXI bus and the DMAC and the Instruction Fetch Buffer (IFB) 1409.

[0222] The Bus Interface Unit (BIU) 1410 is used by the instruction fetch memory 1409 to fetch instructions from external memory, and also by the memory access controller 1405 to fetch the original data of the input matrix A or the weight matrix B from external memory.

[0223] The DMAC is mainly used to move input data from external memory DDR to unified memory 1406, or to weight data to weight memory 1402, or to input data to input memory 1401.

[0224] The vector computation unit 1407 includes multiple processing units that further process the output of the computation circuit 1403 when needed, such as vector multiplication, vector addition, exponential operations, logarithmic operations, size comparisons, etc. It is mainly used for computation in non-convolutional / fully connected layers of neural networks, such as batch normalization, pixel-level summation, and upsampling of feature planes.

[0225] In some implementations, the vector computation unit 1407 can store the processed output vector in the unified memory 1406. For example, the vector computation unit 1407 can apply a linear function, or a nonlinear function, to the output of the computation circuit 1403, such as performing linear interpolation on feature planes extracted from a convolutional layer, or, for example, accumulating a vector of values ​​to generate activation values. In some implementations, the vector computation unit 1407 generates normalized values, pixel-level summed values, or both. In some implementations, the processed output vector can be used as activation input to the computation circuit 1403, for example, for use in subsequent layers of the neural network.

[0226] The instruction fetch buffer 1409 connected to the controller 1404 is used to store the instructions used by the controller 1404;

[0227] Unified memory 1406, input memory 1401, weighted memory 1402, and instruction fetch memory 1409 are all on-chip memories. External memory is proprietary to this NPU hardware architecture.

[0228] The processor mentioned above can be a general-purpose central processing unit, a microprocessor, an ASIC, or one or more integrated circuits used to control the execution of the above program.

[0229] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0230] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0231] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0232] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A data processing method, characterized in that, The method includes: Acquire images for the target scene; Based on the image, multiple primitives of the target scene are learned. The different primitives include the 3D distribution of regions at different locations in the space where the target scene is located. The 3D distribution is used to indicate target information, which is at least one of semantic information, depth information, or optical flow information. Based on the aggregation of some or all of the primitives, render a 3D model, semantic map, depth map, or optical flow map of the target scene.

2. The method according to claim 1, characterized in that, The basic element is a Gaussian sphere.

3. The method according to claim 1 or 2, characterized in that, The plurality of primitives includes a plurality of first primitives in the static region of the scene and a plurality of second primitives in the dynamic region.

4. The method according to claim 3, characterized in that The image specifically includes images of multiple frames of a video, the video including moving objects, and the multiple second primitives including multiple primitives of each frame; After learning multiple primitives of the target scene based on the image, the method further includes: The motion model of the moving object is used as a constraint to optimize the plurality of second primitives.

5. The method according to claim 3 or 4, characterized in that, The image specifically includes images of multiple frames of the video, each frame of the image corresponds to multiple second primitives, and the multiple second primitives corresponding to different frames are different.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: The target scene's RGB image is rendered based on the aggregation of some or all of the multiple primitives.

7. The method according to any one of claims 1 to 6, characterized in that, The step of rendering a semantic map, depth map, or optical flow map of the target scene based on the aggregation of some or all of the plurality of primitives includes: The semantic information of some or all of the multiple primitives is aggregated to render a semantic graph of the target scene; wherein, the aggregation includes a softmax operation on the 3D semantics in the semantic information.

8. The method according to any one of claims 1 to 7, characterized in that, The method further includes: Obtain the labels of the 3D model, semantic map, depth map, or optical flow map of the image; The plurality of primitives are updated according to the labels.

9. A data processing apparatus, characterized in that, The device includes: The acquisition module is used to acquire images of the target scene. The processing module is configured to learn multiple primitives of the target scene based on the image, wherein different primitives include 3D distributions of regions at different locations in space, the 3D distributions being used to indicate target information, the target information being at least one of semantic information, depth information, or optical flow information; and to render a 3D model, semantic map, depth map, or optical flow map of the target scene based on the aggregation of some or all of the multiple primitives.

10. The apparatus according to claim 9, characterized in that, The basic element is a Gaussian sphere.

11. The apparatus according to claim 9 or 10, characterized in that, The plurality of primitives includes a plurality of first primitives in the static region of the scene and a plurality of second primitives in the dynamic region.

12. The apparatus according to claim 11, characterized in that, The image specifically includes images of multiple frames of a video, the video including moving objects, and the multiple second primitives including multiple primitives for each frame; after learning the multiple primitives of the target scene based on the image, the processing module is further configured to: The motion model of the moving object is used as a constraint to optimize the plurality of second primitives.

13. The apparatus according to claim 11 or 12, characterized in that, The image specifically includes images of multiple frames of the video, each frame of the image corresponds to multiple second primitives, and the multiple second primitives corresponding to different frames are different.

14. The apparatus according to any one of claims 9 to 13, characterized in that, The processing module is further configured to: The target scene's RGB image is rendered based on the aggregation of some or all of the multiple primitives.

15. The apparatus according to any one of claims 9 to 14, characterized in that, The processing module is specifically used for: The semantic information of some or all of the multiple primitives is aggregated to render a semantic graph of the target scene; wherein, the aggregation includes a softmax operation on the 3D semantics in the semantic information.

16. The apparatus according to any one of claims 9 to 15, characterized in that, The processing module is further configured to: Obtain the labels of the 3D model, semantic map, depth map, or optical flow map of the image; The plurality of primitives are updated according to the labels.

17. A computer storage medium, characterized in that, The computer storage medium stores one or more instructions, which, when executed by one or more computers, cause the one or more computers to perform the operation of the method according to any one of claims 1 to 8.

18. A computer program product, characterized in that, Includes computer-readable instructions that, when executed on a computer device, cause the computer device to perform the method as described in any one of claims 1 to 8.

19. A system comprising at least one processor and at least one memory; the processor and the memory are connected via a communication bus and communicate with each other. The at least one memory is used to store code; The at least one processor is used to execute the code to perform the method as described in any one of claims 1 to 8.

Images