Model automatic processing method and device based on intelligent optimization and deep learning
By employing intelligent optimization and deep learning methods, the entire process of 3D BIM model processing has been automated, resolving issues related to data compatibility, geometric errors, and rendering efficiency. This has improved processing efficiency, enabled adaptation to complex terminal environments, and maintained high accuracy.
Patent Information
- Application Number
- CN202511025422.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-24
- Publication Date
- 2025-11-14
AI Technical Summary
Traditional 3D BIM model processing workflows suffer from poor data compatibility, frequent geometric errors, low rendering efficiency, and insufficient adaptability to dynamic environments, lacking end-to-end automated solutions.
Employing a method based on intelligent optimization and deep learning, it achieves fully automated processing through format detection, geometric compression, geometric verification, topology optimization, and rendering inspection, combined with graph neural networks (GNN) and generative adversarial networks (GAN). This includes format repair, geometric verification, and rendering error correction, and supports GPU-driven switching of detail levels and AI compensation.
It achieves fully automated error correction, reduces manual intervention, improves geometric compression efficiency by 30%, topology optimization speed by 50%, reduces GPU rendering resource consumption by 40%, and network transmission bandwidth requirements by 60%, while maintaining high-precision visual fidelity.
Smart Images
Figure CN120953398A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of this application relate to the field of multi-module collaborative optimization, specifically to a method and apparatus for automated model processing based on intelligent optimization and deep learning. Background Technology
[0002] As the complexity of 3D BIM (Building Information Modeling) models increases, traditional processing workflows face the following challenges: 1. Poor data compatibility: Diverse model formats easily lead to processing interruptions due to format errors; 2. Frequent geometric errors: Issues such as non-manifold edges and topological defects affect model usability; 3. Low rendering efficiency: High-precision models consume significant computational resources, making it difficult to meet real-time rendering requirements; 4. Insufficient adaptability to dynamic environments: Network transmission jitter or differences in terminal performance can easily lead to model distortion. Existing technologies largely rely on manual intervention or single algorithm optimization, lacking end-to-end automated solutions. Summary of the Invention
[0003] The summary section of this application is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description section below. This summary section is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.
[0004] Some embodiments of this application propose a method, apparatus, computer device, and computer-readable storage medium for automated model processing based on intelligent optimization and deep learning to solve one or more of the technical problems mentioned in the background section above.
[0005] In a first aspect, some embodiments of this application provide an automated model processing method based on intelligent optimization and deep learning. The method includes: performing format detection processing on acquired initial model data to generate a format detection result; in response to determining that the format detection result meets preset format conditions, performing geometric compression processing on the initial model data to generate compressed model data; performing geometric verification processing on the compressed model data to generate a geometric verification result; in response to determining that the geometric verification result meets preset geometric correctness conditions, performing topology optimization processing on the compressed model data to generate model topology data; performing rendering check processing on the model topology data to generate a rendering check result; in response to determining that the rendering check result meets preset normal rendering conditions, performing accelerated rendering processing on the model topology data to generate accelerated model rendering data; and in response to determining that the accelerated model rendering data meets preset jitter conditions, performing compensation processing on the accelerated model rendering data to generate target model data.
[0006] Secondly, some embodiments of this application provide an automated model processing apparatus based on intelligent optimization and deep learning. The apparatus includes: a format detection unit configured to perform format detection processing on acquired initial model data to generate a format detection result; a geometry compression unit configured to perform geometric compression processing on the initial model data in response to determining that the format detection result meets preset format conditions to generate compressed model data; a geometry verification unit configured to perform geometry verification processing on the compressed model data to generate a geometry verification result; a topology optimization unit configured to perform topology optimization processing on the compressed model data in response to determining that the geometry verification result meets preset geometric correctness conditions to generate model topology data; a rendering check unit configured to perform rendering check processing on the model topology data to generate a rendering check result; an accelerated rendering unit configured to perform accelerated rendering processing on the model topology data in response to determining that the rendering check result meets preset normal rendering conditions to generate accelerated model rendering data; and a compensation unit configured to perform compensation processing on the accelerated model rendering data in response to determining that the accelerated model rendering data meets preset jitter conditions to generate target model data.
[0007] Thirdly, this application also provides a computer device, which includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, it implements the method described in any implementation of the first aspect above.
[0008] Fourthly, this application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the method described in any implementation of the first aspect above.
[0009] The above embodiments of this application have the following beneficial effects: Firstly, the automated model processing method based on intelligent optimization and deep learning, as described in some embodiments of this application, achieves fully automated error correction: integrating format repair, geometric verification, and rendering error correction modules to reduce manual intervention. Secondly, it achieves the fusion of deep learning and topology optimization: accelerating topology reconstruction through GNN (Graphical Neural Network) solves the efficiency bottleneck of traditional algorithms. Next, it implements a dynamic adaptive mechanism: supporting GPU (Graphics Processing Unit)-driven LOD (Levels of Detail) switching and AI (Artificial Intelligence) compensation to adapt to complex terminal environments. Then, it achieves a balance between lightweight design and high precision: layered compression algorithms combined with GAN (Generative Adversarial Network) compensation reduce model size while maintaining visual fidelity. Therefore, through multi-module collaborative optimization and deep learning technology, the entire model processing process can be made intelligent. Attached Figure Description
[0010] The above and other features, advantages, and aspects of the embodiments of this application will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.
[0011] Figure 1 This is a flowchart of some embodiments of the model automation processing method based on intelligent optimization and deep learning according to this application;
[0012] Figure 2 The flowchart of the model automation processing method based on intelligent optimization and deep learning according to this application;
[0013] Figure 3 These are schematic diagrams illustrating the structure of some embodiments of the model automation processing device based on intelligent optimization and deep learning according to this application;
[0014] Figure 4 This is a schematic diagram of the structure of a computer device suitable for implementing some embodiments of this application. Detailed Implementation
[0015] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.
[0016] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described herein can be combined with each other.
[0017] It should be noted that the concepts of "first" and "second" mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.
[0018] It should be noted that the terms "a" and "a plurality of" used in this application are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".
[0019] The names of the messages or information exchanged between multiple devices in the embodiments of this application are for illustrative purposes only and are not intended to limit the scope of these messages or information.
[0020] The present application will now be described in detail with reference to the accompanying drawings and embodiments.
[0021] Figure 1 A flowchart 100 is shown, illustrating some embodiments of the automated model processing method based on intelligent optimization and deep learning according to this application. This automated model processing method based on intelligent optimization and deep learning includes the following steps:
[0022] Step 101: Perform format detection processing on the acquired initial model data to generate format detection results.
[0023] In some embodiments, the execution entity of the model automation processing method based on intelligent optimization and deep learning can perform format detection processing on the acquired initial model data to generate a format detection result. The initial model data can be data input from a user terminal received via a wired or wireless connection. Here, the initial model data can be 3D BIM (Building Information Modeling) data. For example, the data format of the initial model data can be, but is not limited to: OBJ (3D model file format), FBX (Filmbox, 3D model file format), and STL (StereoLithography) format. In practice, the execution entity can determine the data format of the initial model data as the format detection result. In practice, the flowchart of this application can be referred to. Figure 2 , Figure 2 A flowchart of the model automation processing method based on intelligent optimization and deep learning according to this application is shown.
[0024] Therefore, it supports input of various raw model data formats, such as OBJ, FBX, and STL.
[0025] It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other currently known or future wireless connection methods.
[0026] Step 102: In response to the determination that the format detection result meets the preset format conditions, the initial model data is geometrically compressed to generate compressed model data.
[0027] In some embodiments, the execution entity may, in response to determining that the format detection result meets preset format conditions, perform geometric compression on the initial model data to generate compressed model data. The preset format conditions may be that the initial model data is in a unified intermediate format (i.e., JSON (JavaScript Object Notation, a lightweight data interchange format)). In practice, in response to determining that the format detection result meets preset format conditions, the execution entity may perform geometric compression on the initial model data using a preset geometric compression algorithm to generate compressed model data. For example, the preset geometric compression algorithm may be an octree-based hierarchical compression algorithm.
[0028] Therefore, an octree-based hierarchical compression algorithm is adopted to dynamically adjust the compression ratio to balance accuracy and size.
[0029] Optionally, before step 103, the above method further includes:
[0030] The first step is to perform format conversion processing on the initial model data in response to the determination that the above format detection result does not meet the above preset format conditions, so as to generate model conversion data.
[0031] In some embodiments, the execution entity may, in response to determining that the format detection result does not meet the preset format conditions, perform format conversion processing on the initial model data to generate model conversion data. In practice, in response to determining that the format detection result does not meet the preset format conditions, the execution entity may convert the initial model data into model conversion data representing a unified intermediate format. Here, the unified intermediate format may be JSON (JavaScript Object Notation, a lightweight data interchange format).
[0032] This allows input data to be converted into a unified intermediate format, ensuring compatibility with subsequent processes.
[0033] The second step is to perform error format detection processing on the above model-converted data to generate error format detection results.
[0034] In some embodiments, the aforementioned execution entity can perform error format detection processing on the model conversion data to generate error format detection results. In practice, the aforementioned execution entity can perform error format detection processing on the model conversion data using a preset error format detection algorithm to generate error format detection results. The error format detection result can be information indicating correct format. The error format detection result can also be information indicating format error. For example, the preset error format detection algorithm can be: detecting format errors based on a rule engine. Here, the rules used in the rule engine can include, but are not limited to, at least one of the following: unit consistency rules (unifying the unit of measurement for all geometric data, such as millimeters), topology rules (verifying the application effectiveness of vertex-face indexes), etc.
[0035] The third step is to perform format repair processing on the above-mentioned model conversion data in response to the determination that the above-mentioned error format detection results meet the preset format error conditions, so as to generate model format repair data.
[0036] In some embodiments, the execution entity may, in response to determining that the error format detection result meets a preset format error condition, perform format repair processing on the model conversion data to generate model format repair data. In some embodiments, the execution entity may, in response to determining that the error format detection result meets a preset format error condition, perform format repair processing on the model conversion data to generate model format repair data. The preset format error condition may be: the error format detection result represents a format error. In practice, in response to determining that the error format detection result meets the preset format error condition, the execution entity may use a preset format repair algorithm to perform format repair processing on the model conversion data to generate model format repair data. For example, the preset format repair algorithm may be: an automatically triggered repair program (e.g., the repair program may be a program including, but not limited to, at least one of the following: scaling transformation algorithm, index remapping algorithm, etc.).
[0037] Fourth step: In response to the determination that the above error format detection result does not meet the above preset format error conditions, the above model conversion data is determined as model format repair data.
[0038] In some embodiments, the execution entity may determine the model conversion data as model format repair data in response to determining that the error format detection result does not meet the preset format error condition.
[0039] The fifth step is to perform geometric compression on the model format repair data to generate compressed model data.
[0040] In some embodiments, the execution entity can perform geometric compression on the model format repair data to generate compressed model data. In practice, the execution entity can use the aforementioned preset geometric compression algorithm to perform geometric compression on the model format repair data to generate compressed model data.
[0041] The sixth step is to determine the error log in response to the fact that the model format repair data meets the preset abnormal conditions, and to send the error log to the user terminal.
[0042] In some embodiments, the execution entity may, in response to determining that the model format repair data meets a preset exception condition, determine an error log and send the error log to the user terminal. The preset exception condition may be: the model format repair data is empty. Here, empty model format repair data can indicate that the initial model data cannot be format repaired. In practice, in response to determining that the model format repair data meets the preset exception condition, firstly, the execution entity may add the initial model data, the model format repair data, and preset repair failure information to the error log. The error log is initially empty. For example, the preset repair failure information may be information indicating "unrepairable". Secondly, the execution entity may send the error log to the user terminal.
[0043] Therefore, the rule engine detects format errors and automatically triggers a repair procedure; if the error cannot be repaired, an error log is recorded and feedback is sent to the user.
[0044] Step 103: Perform geometric verification on the compressed model data to generate geometric verification results.
[0045] In some embodiments, the execution entity can perform geometric verification on the compressed model data to generate a geometric verification result. In practice, the execution entity can first perform geometric verification on the compressed model data using a preset geometric verification algorithm to generate a geometric verification result. For example, the preset geometric verification algorithm can be a geometric hash algorithm. The geometric verification result can be a first preset geometric verification result or a second preset geometric verification result. The first preset geometric verification result can indicate the loss of key features (e.g., holes, sharp edges) in the geometrically compressed model data. The second preset geometric verification result can indicate that the key features of the geometrically compressed model data have not been lost.
[0046] Step 104: In response to determining that the geometric verification results meet the preset geometric correctness conditions, topology optimization processing is performed on the model compressed data to generate model topology data.
[0047] In some embodiments, the execution entity may perform topology optimization processing on the compressed model data to generate model topology data in response to determining that the geometric verification result meets a preset geometric correctness condition. The preset geometric correctness condition may be that the geometric verification result is a second preset geometric verification result. In practice, in response to determining that the geometric verification result meets the preset geometric correctness condition, the execution entity may perform topology optimization processing on the compressed model data using a preset topology optimization algorithm to generate model topology data. For example, a pre-defined topology optimization algorithm could be as follows: First, illegal edges are shrunk and overlapping faces are separated using Discrete Curvature Flow. Second, a Graph Neural Network (GNN) is used to rewire the topology connections, ensuring that each edge is shared by only two faces. Then, the hole boundary points are projected onto the implicit manifold surface to generate a parameterized UV mesh (texture coordinate network). Next, skinning lofting based on NURBS (Non-Uniform Rational B-Splines) is used to continuously optimize the curvature, maintaining curvature continuity. Then, redundant UV map coordinates are merged to reduce rendering batches. Finally, a pre-trained Graph Neural Network (GNN) is used to predict the optimal topology, accelerating the optimization process (improving efficiency by more than 40% compared to traditional algorithms).
[0048] Optionally, before step 105, the above method further includes:
[0049] The first step is to perform geometric error repair processing on the compressed model data in response to the determination that the above geometric test results do not meet the above preset geometric correctness conditions, so as to generate model repair data.
[0050] In some embodiments, the execution entity may, in response to determining that the geometric verification result does not meet the preset geometric correctness condition, perform geometric error repair processing on the compressed model data to generate repaired model data. In practice, in response to determining that the geometric verification result does not meet the preset geometric correctness condition, the execution entity may perform geometric error repair processing on the compressed model data using a preset geometric error repair algorithm to generate repaired model data. For example, the preset geometric error repair algorithm may be, but is not limited to: a secondary compression algorithm (here, the secondary compression algorithm may be performing the geometric compression processing step again), or a degradation processing (here, the degradation processing may be reducing the fineness level of the texture).
[0051] Therefore, the integrity of the compressed model is verified by the geometric hash algorithm. If the loss of key features (such as holes or sharp edges) is detected, secondary compression or downgrading is triggered.
[0052] The second step is to perform manifold repair processing on the model repair data in response to the determination that the above model repair data meets the preset non-manifold edge over-limit condition, so as to generate model manifold data.
[0053] In some embodiments, the execution entity may, in response to determining that the model repair data meets a preset non-manifold edge exceeding condition, perform manifold repair processing on the model repair data to generate model manifold data. The preset non-manifold edge exceeding condition may be: the number of non-manifold edges is greater than a preset threshold (e.g., the preset threshold may be 0), and the accuracy of various key regions is inconsistent (the key region may be the region where the key feature is located). In practice, in response to determining that the model repair data meets the preset non-manifold edge exceeding condition, the execution entity may perform manifold repair processing on the model repair data using a preset manifold repair algorithm to generate model manifold data. For example, the preset manifold repair algorithm may be an adaptive subdivision algorithm.
[0054] Therefore, the local mesh density can be adjusted through the Adaptive Subdivision algorithm to ensure the accuracy of critical regions (the accuracy of critical regions must be consistent, and there should be no difference in levels. For example, when a tile is decomposed into 18 levels, the corresponding critical regions should all be at level 18), while repairing non-manifold edges.
[0055] The third step is to perform topology optimization on the above model manifold data to generate model topology data.
[0056] In some embodiments, the execution entity may perform topology optimization processing on the model manifold data to generate model topology data. In practice, the execution entity may use the aforementioned preset topology optimization algorithm to perform topology optimization processing on the model manifold data to generate model topology data.
[0057] Fourth step: In response to the determination that the above model repair data does not meet the preset non-manifold edge over-limit condition, the above model repair data is subjected to topology optimization processing to generate model topology data.
[0058] In some embodiments, the execution entity may perform topology optimization processing on the model repair data to generate model topology data in response to determining that the model repair data does not meet the preset non-manifold edge exceeding condition. In practice, in response to determining that the model repair data does not meet the preset non-manifold edge exceeding condition, the execution entity may perform topology optimization processing on the model repair data using the preset topology optimization algorithm to generate model topology data.
[0059] Step 105: Perform rendering check processing on the model topology data to generate rendering check results.
[0060] In some embodiments, the execution entity can perform rendering check processing on the model topology data to generate rendering check results. In practice, the execution entity can use a preset rendering check algorithm to perform rendering check processing on the model topology data to generate rendering check results. The rendering check results can characterize whether any anomalies have occurred during the rendering check. For example, the preset rendering check algorithm can be an OpenGL (Open Graphics Library) pre-rendering model or a DirectX API (Direct eXtension API) pre-rendering model. Thus, based on the OpenGL / DirectX API pre-rendering model, problems such as abnormal lighting and texture misalignment can be detected.
[0061] Step 106: In response to determining that the rendering check results meet the preset normal rendering conditions, the model topology data is processed for accelerated rendering to generate accelerated rendering data for the model.
[0062] In some embodiments, the execution entity may, in response to determining that the rendering check result meets preset normal rendering conditions, perform accelerated rendering processing on the model topology data to generate accelerated model rendering data. The preset normal rendering conditions may be that the rendering check result indicates that no anomalies occurred during the rendering check. In practice, in response to determining that the rendering check result meets the preset normal rendering conditions, the execution entity may use a preset accelerated rendering algorithm to perform accelerated rendering processing on the model topology data to generate accelerated model rendering data. For example, the preset accelerated rendering algorithm may be: automatically switching the model LOD level based on the terminal GPU performance to ensure smooth rendering.
[0063] Optionally, before step 107, the above method further includes:
[0064] The first step is to perform rendering repair processing on the model topology data in response to the determination that the above rendering check results do not meet the above preset normal rendering conditions, so as to generate model rendering repair data.
[0065] In some embodiments, the execution entity may, in response to determining that the rendering check result does not meet the preset normal rendering conditions, perform rendering repair processing on the model topology data to generate model rendering repair data. In practice, in response to determining that the rendering check result does not meet the preset normal rendering conditions, the execution entity may use a preset rendering check algorithm to perform rendering repair processing on the model topology data to generate model rendering repair data. For example, the preset rendering check algorithm may be: calling GPU (Graphics Processing Unit) parallel computing resources for real-time repair (such as normal recalculation, texture remapping).
[0066] The second step is to perform accelerated rendering processing on the above model rendering repair data to generate accelerated model rendering data.
[0067] In some embodiments, the execution entity may perform accelerated rendering processing on the model rendering repair data to generate accelerated model rendering data. In practice, the specific implementation method of accelerating rendering processing on the model rendering repair data and the resulting technical effects can be found in step 106 of the above embodiments, and will not be repeated here.
[0068] Step 107: In response to determining that the accelerated rendering data of the model meets the preset jitter conditions, the accelerated rendering data of the model is compensated to generate the target model data.
[0069] In some embodiments, the execution entity may, in response to determining that the accelerated model rendering data meets a preset jitter condition, perform compensation processing on the accelerated model rendering data to generate target model data. The preset jitter condition may be: network jitter or insufficient network bandwidth occurs during the transmission of the accelerated model rendering data. In practice, in response to determining that the accelerated model rendering data meets the preset jitter condition, the execution entity may use a preset compensation algorithm to perform compensation processing on the accelerated model rendering data to generate target model data. For example, the preset compensation algorithm may be: predicting lost geometric details based on a generative adversarial network (GAN), or downgrading to a low-precision model (LOD (Levels of Detail) downgrade).
[0070] Therefore, in network transmission scenarios, if jitter or insufficient bandwidth is detected, the system triggers an AI compensation mechanism—based on generative adversarial networks (GANs) to predict lost geometric details, or downgrades to a low-precision model (LOD downgrade) to ensure real-time performance.
[0071] Optionally, in response to determining that the above-mentioned model accelerated rendering data does not meet the above-mentioned preset jitter conditions, the above-mentioned model accelerated rendering data is determined as the target model data.
[0072] In some embodiments, the execution entity may determine the model accelerated rendering data as target model data in response to determining that the model accelerated rendering data does not meet the preset jitter condition.
[0073] Thus, this application achieves the following effects:
[0074] Efficiency improvements: Geometric compression reduces time by 30%, and deep learning acceleration increases topology optimization speed by 50%;
[0075] Error rate reduction: Automatic formatting error repair success rate ≥95%, non-manifold edge repair accuracy ≥98%;
[0076] Resource optimization: GPU rendering efficiency improved by 40%, network bandwidth requirements reduced by 60%.
[0077] The above embodiments of this application have the following beneficial effects: Firstly, the automated model processing method based on intelligent optimization and deep learning, as described in some embodiments of this application, achieves fully automated error correction: integrating format repair, geometric verification, and rendering error correction modules reduces manual intervention. Secondly, it achieves the fusion of deep learning and topology optimization: accelerating topology reconstruction through GNN solves the efficiency bottleneck of traditional algorithms. Next, it implements a dynamic adaptive mechanism: supporting GPU-driven LOD switching and AI compensation to adapt to complex terminal environments. Then, it achieves a balance between lightweight design and high precision: layered compression algorithms combined with GAN compensation reduce model size while maintaining visual fidelity. Therefore, through multi-module collaborative optimization and deep learning technology, the entire model processing process can be made intelligent.
[0078] Further reference Figure 3 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a model automation processing device based on intelligent optimization and deep learning. These embodiments of the model automation processing device based on intelligent optimization and deep learning are similar to... Figure 1 Corresponding to the method embodiments shown, this model automation processing device based on intelligent optimization and deep learning can be specifically applied to various electronic devices.
[0079] like Figure 3As shown, the model automation processing device 300 based on intelligent optimization and deep learning in some embodiments includes: a format detection unit 301, a geometric compression unit 302, a geometric verification unit 303, a topology optimization unit 304, a rendering inspection unit 305, an accelerated rendering unit 306, and a compensation unit 307. The system includes: a format detection unit 301 configured to perform format detection processing on the acquired initial model data to generate a format detection result; a geometry compression unit 302 configured to perform geometric compression processing on the initial model data in response to determining that the format detection result meets preset format conditions to generate compressed model data; a geometry verification unit 303 configured to perform geometry verification processing on the compressed model data to generate a geometry verification result; a topology optimization unit 304 configured to perform topology optimization processing on the compressed model data in response to determining that the geometry verification result meets preset geometry correctness conditions to generate model topology data; a rendering check unit 305 configured to perform rendering check processing on the model topology data to generate a rendering check result; an accelerated rendering unit 306 configured to perform accelerated rendering processing on the model topology data in response to determining that the rendering check result meets preset rendering normal conditions to generate accelerated model rendering data; and a compensation unit 307 configured to perform compensation processing on the accelerated model rendering data in response to determining that the accelerated model rendering data meets preset jitter conditions to generate target model data.
[0080] It is understandable that the units and references described in the model automation processing device 200 based on intelligent optimization and deep learning... Figure 1 The steps described in the method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method also apply to the model automation processing device 200 based on intelligent optimization and deep learning and the units contained therein, and will not be repeated here.
[0081] This application also provides a computer device 400. For example... Figure 4 As shown, the computer device 400 includes a bus 401, a processor 402, a memory 403, and a communication interface 404. The processor 402, the memory 403, and the communication interface 404 communicate with each other via the bus 401. The computer device 400 can be a server or a terminal device. It should be understood that this application does not limit the number of processors and memories in the computer device 400.
[0082] Bus 401 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, Figure 4 The bus 401 may be represented by a single line, but this does not mean that there is only one bus or one type of bus. The bus 401 may include a path for transmitting information between various components of the computer device 400 (e.g., memory 403, processor 402, communication interface 404).
[0083] Processor 402 may include any one or more processors such as a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).
[0084] Memory 403 may include volatile memory, such as random access memory (RAM). Memory 403 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).
[0085] The memory 403 stores executable program code, which the processor 402 executes to implement the functions of the aforementioned format detection unit, geometric compression unit, geometric verification unit, topology optimization unit, rendering check unit, accelerated rendering unit, and compensation unit, thereby realizing the aforementioned automated model processing method based on intelligent optimization and deep learning. In other words, the memory 403 stores instructions for executing the aforementioned automated model processing method based on intelligent optimization and deep learning.
[0086] The communication interface 404 uses transceiver modules such as, but not limited to, network interface cards and transceivers to enable communication between the computer device 400 and other devices or communication networks.
[0087] This application also provides a chip, which includes a processor and a data interface. The processor reads instructions stored in the memory through the data interface to execute the above-described model automation processing method based on intelligent optimization and deep learning.
[0088] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that a computing device can store, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct the computing device to execute the aforementioned automated model processing method based on intelligent optimization and deep learning.
[0089] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0090] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of this application.
Claims
1. A method for automated model processing based on intelligent optimization and deep learning, characterized in that, include: The acquired initial model data is subjected to format detection processing to generate format detection results; In response to determining that the format detection result meets the preset format conditions, the initial model data is geometrically compressed to generate compressed model data; The compressed data of the model is subjected to geometric verification processing to generate geometric verification results; In response to determining that the geometric verification result meets the preset geometric correctness conditions, the model compressed data is subjected to topology optimization processing to generate model topology data; The model topology data is subjected to rendering inspection processing to generate rendering inspection results; In response to determining that the rendering check result meets the preset normal rendering conditions, the model topology data is subjected to accelerated rendering processing to generate accelerated model rendering data. In response to determining that the accelerated rendering data of the model meets the preset jitter conditions, the accelerated rendering data of the model is compensated to generate target model data.
2. The automated model processing method based on intelligent optimization and deep learning according to claim 1, characterized in that, Before performing geometric verification processing on the compressed model data to generate geometric verification results, the method further includes: In response to determining that the format detection result does not meet the preset format conditions, the initial model data is subjected to format conversion processing to generate model conversion data; The model-converted data is subjected to error format detection processing to generate error format detection results; In response to determining that the error format detection result meets the preset format error conditions, the model conversion data is subjected to format repair processing to generate model format repair data; In response to determining that the error format detection result does not meet the preset format error condition, the model conversion data is determined as model format repair data; Geometric compression is performed on the model format repair data to generate compressed model data; In response to determining that the model format repair data meets preset abnormal conditions, an error log is determined and the error log is sent to the user terminal.
3. The automated model processing method based on intelligent optimization and deep learning according to claim 1, characterized in that, Before performing rendering inspection processing on the model topology data to generate rendering inspection results, the method further includes: In response to determining that the geometric inspection result does not meet the preset geometric correctness condition, geometric error repair processing is performed on the compressed model data to generate model repair data; In response to determining that the model repair data meets the preset non-manifold edge exceeding condition, the model repair data is subjected to manifold repair processing to generate model manifold data; The model manifold data is subjected to topology optimization processing to generate model topology data; In response to determining that the model repair data does not meet the preset non-manifold edge overlimit condition, the model repair data is subjected to topology optimization processing to generate model topology data.
4. The automated model processing method based on intelligent optimization and deep learning according to claim 1, characterized in that, Before performing compensation processing on the model accelerated rendering data to generate target model data in response to determining that the model accelerated rendering data meets a preset jitter condition, the method further includes: In response to determining that the rendering check result does not meet the preset normal rendering conditions, the model topology data is subjected to rendering repair processing to generate model rendering repair data; The model rendering repair data is processed to accelerate rendering, thereby generating accelerated model rendering data.
5. The automated model processing method based on intelligent optimization and deep learning according to claim 1, characterized in that, The method further includes: In response to determining that the accelerated rendering data of the model does not meet the preset jitter condition, the accelerated rendering data of the model is determined as the target model data.
6. A model automation processing device based on intelligent optimization and deep learning, characterized in that, include: The format detection unit is configured to perform format detection processing on the acquired initial model data to generate format detection results; A geometric compression unit is configured to perform geometric compression processing on the initial model data in response to determining that the format detection result meets preset format conditions, so as to generate compressed model data; A geometric verification unit is configured to perform geometric verification processing on the compressed model data to generate geometric verification results; A topology optimization unit is configured to perform topology optimization processing on the model compressed data in response to determining that the geometric verification result meets a preset geometric correctness condition, so as to generate model topology data. A rendering inspection unit is configured to perform rendering inspection processing on the model topology data to generate rendering inspection results; An accelerated rendering unit is configured to perform accelerated rendering processing on the model topology data in response to determining that the rendering check result meets preset normal rendering conditions, so as to generate accelerated rendering data of the model. The compensation unit is configured to perform compensation processing on the model accelerated rendering data in response to determining that the model accelerated rendering data meets a preset jitter condition, so as to generate target model data.
7. A computer device, characterized in that, The computer device includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, it implements the steps of the method as described in any one of claims 1-5.
8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-5.