Industrial large model-based national wind image generation method, apparatus and device, and medium

Abstract

The invention discloses a national wind image generation method and device based on an industry large model, equipment and a medium, and relates to the technical field of computers. Comprising the following steps: acquiring information data in the national wind field, and preprocessing the information data to obtain target information data; the national breeze fields comprise the drawing field, the calligraphy field, the building field and the clothing field; constructing an initial industry large model, and training the initial industry large model based on the target information data to obtain a preset industry large model; obtaining user input information, and analyzing the user input information by using a preset industry large model to obtain target demand information; the target demand information comprises subject information, behavior information, object information and an artistic relationship; and generating a target national wind image based on the target demand information by using a preset industry large model and a preset texture enhancement technology. Therefore, the national wind style images with consistent styles and reduced details can be generated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer technology, and particularly relates to a method and device for generating a Chinese traditional image based on an industry large model, equipment and a medium. BACKGROUND

[0002] In today's era of rapid development of digitalization and artificial intelligence technology, image generation technology has become an important tool in the fields of artistic creation, cultural heritage, education and training, and commercial design. With the continuous progress of deep learning technology, especially the widespread application of generative adversarial networks and diffusion models (Diffusion Models), text-to-image generation technology has made significant breakthroughs. However, although these technologies have performed well in generating high-quality and diverse images, there are still many technical bottlenecks and challenges in the generation of images in specific styles, especially the generation of Chinese traditional style images.

[0003] Chinese traditional style, as a style with deep cultural roots and unique artistic expression, covers multiple artistic fields such as painting, calligraphy, architecture, and clothing. Its core features lie in the accurate expression of traditional artistic elements, including the smoothness of lines, the coordination of colors, the complexity of textures, and the consistency of overall style. However, existing text-to-image generation technology often fails to meet these requirements when generating Chinese traditional images. The generated images either lack the charm of traditional art or have obvious shortcomings in detail reproduction and artistic expression.

[0004] Firstly, the prior art has significant defects in style consistency. Due to the universality and diversity of training data, general large models often fail to maintain the uniformity of style when generating images of a specific style. Secondly, the prior art performs poorly in detail restoration. One important feature of national style images is their rich details and complex textures. However, general large models often produce blurred or distorted images when generating these complex details. In addition, the prior art also has obvious shortcomings in multi-style fusion and dynamic generation. Existing text-to-image technology can only generate simple mixed effects when dealing with multi-style fusion, and cannot achieve natural and harmonious style transitions. At the same time, in terms of dynamic generation, the existing technology is difficult to adjust the generated image style and content in real time according to the specific needs of users, resulting in a lack of flexibility and personalization in the generated images. Finally, the prior art has hidden dangers in data security and compliance. The widespread application of image generation technology in artistic creation, education and training, and commercial design involves the processing of a large amount of sensitive data, such as personal privacy, business secrets, and cultural heritage protection. However, existing systems have many shortcomings in data encryption, access control, and privacy protection, making it difficult to effectively protect the security and compliance of data. This not only limits the widespread application of technology, but also poses a potential threat to the data security of users.

[0005] In summary, the existing text-to-image technology faces technical bottlenecks in style consistency, detail restoration, multi-style fusion, and data security when generating images of a specific style, especially in the field of national style image generation. These problems not only limit the practical application value of the technology, but also hinder the process of digital transformation in the art industry. Therefore, how to generate national style images with consistent style and restored details is a problem that needs to be solved urgently. SUMMARY

[0006] Therefore, the purpose of the present application is to provide a national style image generation method, device, equipment and medium based on an industry large model, which can generate national style images with consistent style and restored details. The specific solutions are as follows:

[0007] In a first aspect, the present application provides a national style image generation method based on an industry large model, comprising:

[0008] obtaining information data in the national style field, and preprocessing the information data to obtain target information data; the national style field includes the fields of painting, calligraphy, architecture, and clothing; the information data includes structured artistic element data and unstructured image data and text data;

[0009] constructing an initial industry large model, and training the initial industry large model based on the target information data to obtain a preset industry large model;

[0010] obtain user input information, and parse the user input information by using the preset industry large model to obtain target demand information; the target demand information includes subject information, behavior information, object information, and artistic relationship;

[0011] generate a target Chinese style image based on the target demand information by using the preset industry large model and a preset texture enhancement technology.

[0012] Optionally, the information data in the Chinese style field includes:

[0013] If the information data is structured data, API is connected with a target data source in the Chinese style field, so as to update the information data in the Chinese style field in real time;

[0014] If the information data is unstructured data, an image processing technology is used for screening to obtain the information data in the Chinese style field.

[0015] Optionally, the information data is preprocessed to obtain target information data, including:

[0016] If the information data is structured data, a preset segmentation algorithm is used for block processing on the information data to obtain target information data;

[0017] If the information data is unstructured data, a named entity recognition technology is used for semantic annotation on the information data to obtain target information data.

[0018] Optionally, the initial industry large model is trained based on the target information data to obtain a preset industry large model, including:

[0019] The initial industry large model is trained based on the target information data by using a transfer learning technology to obtain a preset industry large model.

[0020] Optionally, the user input information is parsed by using the preset industry large model to obtain target demand information, including:

[0021] The user input information is parsed by using the preset industry large model to obtain target demand information.

[0022] Optionally, the target Chinese style image is generated based on the target demand information by using the preset industry large model and a preset texture enhancement technology, including:

[0023] The Chinese style image is generated based on the target demand information by using the preset industry large model;

[0024] The preset industry large model is used to identify the detail features in the traditional Chinese style image, and the preset texture generation algorithm is used to enhance the identified detail features to obtain the target traditional Chinese style image.

[0025] Optionally, the method for generating national style images based on the industry large model further includes:

[0026] The data in the preset industry big model is protected through sensitive information encryption, access permission control, and compliance verification mechanisms.

[0027] In a second aspect, the present application provides a national style image generation device based on an industry large model, comprising:

[0028] a target information data acquisition module, configured to acquire information data in the field of traditional Chinese culture and pre-process the information data to obtain target information data; the field of traditional Chinese culture includes painting, calligraphy, architecture, and clothing; the information data includes structured art element data and unstructured image data and text data;

[0029] A preset industry big model construction module is used to construct an initial industry big model and train the initial industry big model based on the target information data to obtain a preset industry big model;

[0030] A user input information acquisition module is used to acquire user input information and analyze the user input information using the preset industry macro model to obtain target demand information; the target demand information includes subject information, behavior information, object information and artistic relationship;

[0031] The target national style image generation module is used to generate the target national style image based on the target demand information by using the preset industry large model and the preset texture enhancement technology.

[0032] In a third aspect, the present application provides an electronic device, comprising:

[0033] Memory, used to store computer programs;

[0034] A processor is used to execute the computer program to implement the aforementioned method for generating national style images based on the industry large model.

[0035] In a fourth aspect, the present application provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the aforementioned method for generating national style images based on industry large models.

[0036] The application provides a national style image generation method based on an industry large model. First, information data in the national style field is acquired, and the information data is preprocessed to obtain target information data; the national style field includes the painting field, the calligraphy field, the architecture field and the costume field; the information data includes structured artistic element data and unstructured image data and text data; then, an initial industry large model is constructed, and the initial industry large model is trained based on the target information data to obtain a preset industry large model; subsequently, user input information is acquired, and the user input information is analyzed by using the preset industry large model to obtain target demand information; the target demand information includes subject information, behavior information, object information and artistic relationship; finally, the preset industry large model and a preset texture enhancement technology are used to generate a target national style image based on the target demand information.

[0037] As can be seen from the above, the application acquires information data in the national style field, and trains an initial industry large model by using the information data, so that the model can deeply learn the connotation and extension of national style artistic elements, including the fluency of lines, the coordination of colors, the complexity of textures and the consistency of overall style. The generated image is optimized layer by layer by using high-precision texture enhancement technology. Therefore, a national style image with consistent style and restored details can be generated. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of the provided drawings.

[0039] Figure 1 A national style image generation method based on an industry large model disclosed by the present application is shown in the flow chart;

[0040] Figure 2 A specific national style image generation method based on an industry large model disclosed by the present application is shown in the flow chart;

[0041] Figure 3 A national style image generation device based on an industry large model disclosed by the present application is shown in the schematic diagram;

[0042] Figure 4 A structure diagram of an electronic device disclosed by the present application is shown in the structure diagram. DETAILED DESCRIPTION

[0043] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0044] As a style with deep cultural heritage and unique artistic expressiveness, national style covers multiple artistic fields such as painting, calligraphy, architecture, and costumes. Its core features lie in the accurate expression of traditional artistic elements, including the smoothness of lines, the coordination of colors, the complexity of textures, and the consistency of overall style. However, existing text-to-image technology often fails to meet these requirements when generating national style images. The generated images either lack the charm of traditional art or have obvious shortcomings in detail restoration and artistic expressiveness. Therefore, the present application provides a national style image generation scheme based on an industry large model, which can generate national style images with consistent style and detailed restoration.

[0045] Referring to Figure 1 The embodiments of the present application disclose a national style image generation method based on an industry large model, which includes:

[0046] Step S11, information data in the national style field is obtained, and the information data is preprocessed to obtain target information data.

[0047] In this embodiment, the information data in the national style field is obtained by combining automatic crawling with manual screening. The national style field includes but is not limited to the fields of painting, calligraphy, architecture, and costumes. The information data includes structured artistic element data and unstructured image data and text data. Specifically, obtaining information data in the national style field can include: if the information data is structured data, API interfacing with the target data source of the national style field is performed to update the information data in the national style field in real time; if the information data is unstructured data, image processing technology is used to screen and obtain the information data in the national style field. That is, on the one hand, API interfacing with authoritative art data sources is performed to realize real-time synchronization updating of national art data; on the other hand, for unstructured data such as ancient painting scans and calligraphy works, image processing technology is used for preliminary screening to extract core information, thereby constructing a high-quality national art data set.

[0048] Further, the acquired data is preprocessed to ensure system performance. Specifically, the preprocessing of the information data to obtain target information data can include: if the information data is structured data, performing block processing on the information data using a preset segmentation algorithm to obtain target information data; and if the information data is unstructured data, performing semantic labeling on the information data through a named entity recognition technology to obtain target information data. That is, the collected structured guofeng image data is processed in blocks, and a segmentation algorithm dedicated to the art field is used to ensure accurate segmentation of artistic elements. Secondly, the unstructured text data is semantically labeled, and the named entity recognition technology is used to label artistic subjects (such as painters, work names, etc.), artistic behaviors (such as creation, copying, etc.), and artistic concepts (such as ink and brush, meticulous brushwork, etc.). In addition, the system also performs vectorization processing on the data, converts the image data into a high-dimensional vector representation, so as to facilitate subsequent semantic analysis and texture enhancement.

[0049] Step S12, an initial industry large model is constructed, and the initial industry large model is trained based on the target information data to obtain a preset industry large model.

[0050] In this embodiment, the general large model is fine-tuned using massive professional data in the guofeng field, so that the model can deeply learn the connotation and extension of guofeng artistic elements. Specifically, in the model construction process, first, a basic model pre-trained by large-scale general data is selected, and then fine-tuning training is performed on guofeng field data, focusing on optimizing the model's understanding ability of artistic elements, professional expressions, and context. In the fine-tuning process, the transfer learning technology is used to transfer the knowledge learned by the general large model on large-scale general data to guofeng specific data, so as to accelerate the convergence speed of the model and improve the initial performance of the model.

[0051] Step S13, acquiring user input information and using the preset industry large model to analyze the user input information to obtain target demand information.

[0052] In this embodiment, the semantic analysis module of the preset industry large model is used to deeply analyze the user input information and identify the artistic subjects, behaviors, objects, and artistic relationships therein. Specifically, the use of the preset industry large model to analyze the user input information to obtain target demand information can include: using the preset industry large model to perform semantic analysis on the user input information to obtain target demand information. That is, the semantic analysis module is introduced as an auxiliary, and through natural language processing technology, the user input text description is deeply analyzed to identify the artistic subjects, behaviors, objects, and artistic relationships. Thus, the artistic logical reasoning ability of the model can be improved.

[0053] Step S14, generating a target Chinese style image based on the target demand information by using the preset industry large model and a preset texture enhancement technology.

[0054] In this embodiment, the high-precision texture enhancement technology is introduced to restore the complex texture and details of the Chinese style image. Specifically, the generation of the target Chinese style image based on the target demand information by using the preset industry large model and the preset texture enhancement technology can include: generating a Chinese style image based on the target demand information by using the preset industry large model; identifying the detail features in the Chinese style image by using the preset industry large model, and enhancing the identified detail features by using a preset texture generation algorithm to obtain the target Chinese style image. The core of the texture enhancement technology is to identify and enhance the texture features of the image by using a deep learning algorithm. The system first uses the industry large model to perform deep semantic analysis on the Chinese style artistic elements, identifies the key detail features in the image, such as the direction of the brush strokes, the thickness of the lines, and the levels of the colors. Then, the texture generation algorithm is used to enhance these features, ensuring that the generated image meets the professional artistic creation standards in terms of detail restoration and artistic expression. For example, when generating ink paintings, the system can accurately restore the smoothness and level of the brush strokes; when generating meticulous paintings, it can clearly present the delicate lines and color levels. In addition, the texture enhancement technology also supports dynamic adjustment, which can optimize the details of the generated image in real time according to the specific needs of the user, further improving the artistic value of the generated image.

[0055] It is worth mentioning that artistic data involves a large amount of sensitive information such as personal privacy, business secrets and cultural heritage protection. Once these data are leaked, not only the legitimate rights and interests of the relevant subjects will be damaged, but also serious legal consequences may be triggered. Therefore, the embodiments of the present application protect the data in the preset industry large model through sensitive information encryption, access permission control and compliance verification mechanism. Specifically, the AES-256 encryption algorithm is used to encrypt the artistic data in storage and transmission, ensuring the confidentiality of the data. All data transmission is carried out through the HTTPS protocol to prevent data from being stolen or tampered with during transmission. A multi-level access permission management system is established, and different access permissions are assigned according to the roles of users (such as artists, designers, ordinary users, etc.) and the permission levels (such as read-only, edit, administrator, etc.). Users need to log in to the system through multi-factor authentication (such as password, fingerprint, dynamic verification code, etc.), ensuring that only authorized users can access and operate related data. The built-in compliance verification module monitors the user operation and data processing process in real time to ensure that it meets the requirements of laws and regulations. For example, when processing user requirements, the system automatically checks whether there is a risk of sensitive information leakage and warns of high-risk operations. At the same time, the system records all operation logs to support post-audit and traceability, providing strong protection for data security. For data involving personal privacy or business secrets, the system uses anonymization processing technology to remove sensitive identification information from the data without affecting the usability of the data. For example, when processing the personal information of artists, the system automatically masks sensitive information such as names and contact information, and only retains the core artistic features of the works.

[0056] As can be seen from the above, the embodiments of the present application train the initial industry large model by obtaining information data in the field of national style, so that the model can deeply learn the connotation and extension of national style artistic elements, including the fluency of lines, the coordination of colors, the complexity of textures, and the consistency of overall style. And through high-precision texture enhancement technology, the generated image is optimized layer by layer. Thus, a national style image with consistent style and restored details can be generated. At the same time, the embodiments of the present application attach great importance to data security and compliance, integrating a series of security mechanisms such as sensitive information encryption, access permission control, compliance verification, etc. By using advanced encryption algorithms, the image data in storage and transmission is encrypted to ensure the confidentiality of the data; by strictly controlling access permissions, only authorized users can access and operate related data to prevent data leakage and misuse; by the compliance verification mechanism, the behavior and data processing process of the system are monitored in real time to ensure that they meet the requirements of laws and regulations. These security measures build a solid protective barrier for the system, enabling the system to provide reliable protection for the application of artistic technology when processing sensitive data.

[0057] Referring to Figure 2As shown, the embodiments of the present application disclose a specific industry-based model-based Chinese style image generation method, which comprises:

[0058] In this embodiment, a high-quality Chinese style art dataset is first constructed, covering professional knowledge in multiple fields such as painting, calligraphy, architecture, and costumes. These data types are diverse, including both structured art element databases and unstructured image data and text descriptions. During data collection, the system adopts a combination of automated crawling and manual screening. On the one hand, through API interfacing with authoritative art data sources, real-time synchronous updating of Chinese style art data is achieved; on the other hand, for unstructured data such as ancient painting scans and calligraphy works, the system uses image processing technology for preliminary screening, eliminating irrelevant content and extracting core information. To ensure the accuracy and authority of the data, all data must be reviewed and annotated by art professionals, including keywords of art elements, applicable scenarios, associated elements, and the creation background of the work, etc.

[0059] Further, data preprocessing is a key link to ensure system performance. First, the collected Chinese style image data is processed by block, using art field-specific segmentation algorithms to ensure accurate segmentation of art elements. Second, unstructured text data is semantically annotated, using named entity recognition technology to annotate art subjects (such as painters, work names, etc.), artistic behaviors (such as creation, copying, etc.), and artistic concepts (such as ink and brush, meticulous brushwork, etc.). In addition, the system also performs vectorization processing on the data, converting image data into high-dimensional vector representation for subsequent semantic analysis and texture enhancement.

[0060] In this embodiment, although general large models have shown strong image generation capabilities in a wide range of fields, their performance is not satisfactory when generating images of specific styles, especially Chinese style images. The core features of Chinese style images lie in the accurate expression of traditional art elements, including the smoothness of lines, the coordination of colors, the complexity of textures, and the consistency of overall style. The present application uses massive professional data in the Chinese style field to fine-tune the general large model, enabling the model to deeply learn the connotation and denotation of Chinese style art elements.

[0061] During model construction, a pre-trained base model based on large-scale general data is first selected, such as the Stable Diffusion or DALL·E series model. Then, fine-tuning training is performed on Chinese style field data, focusing on optimizing the model's understanding of art elements, professional expression, and context. During fine-tuning, transfer learning technology is used to transfer the knowledge learned by the general large model on large-scale general data to Chinese style specific data, accelerating the convergence speed of the model and improving the initial performance of the model.

[0062] To enhance the artistic logical reasoning capability of the model, the system introduces a semantic parsing module as an auxiliary. The semantic parsing module, through natural language processing technology, deeply analyzes the text description input by the user, identifies the artistic subject, behavior, object, and artistic relationship therein. For example, when the user inputs "generate a ink landscape painting, with distant mountains, near water, and flying birds", the semantic parsing module can accurately extract key elements such as "ink landscape", "distant mountains", "near water", and "flying birds", and convert them into structured information that the model can understand.

[0063] In this embodiment, by combining the logical reasoning capability of the industry large model and the semantic parsing module, complex functions such as multi-style correlation analysis, element analogy reasoning, and personalized demand processing are realized. The image generation process mainly includes the following key steps: demand analysis and semantic understanding: when the user inputs the generation demand, the system first uses the semantic parsing module to deeply analyze the demand, and identifies the artistic subject, behavior, object, and artistic relationship in the demand. For example, the user inputs "generate a meticulous flower and bird painting, with peonies and peacocks", the semantic parsing module can accurately extract key elements such as "meticulous flower and bird", "peonies", and "peacocks", and convert them into structured information that the model can understand. Logical reasoning and image generation: the industry large model generates structured images containing element references, logical deductions, and artistic expressions based on the parsed structured information and high-precision texture enhancement technology. For example, when generating a meticulous flower and bird painting, the model can analyze the morphological characteristics of peonies, the feather texture of peacocks, and the artistic effect of the overall composition, and generate an image that conforms to the traditional artistic style. Multi-round interaction and optimization: to improve the accuracy of the generated image, the system supports multi-round interaction optimization. After the initial image generation, the system further adjusts the generation logic based on user feedback (such as modification suggestions, supplementary demands, etc.), gradually guides the user to perfect the demand description, and finally provides an image that better meets the user's demand. Image visualization and user guidance: for complex demands, the system displays artistic relationships and element logic through visualization tools (such as composition diagrams, color matching diagrams, etc.), helping users better understand the generated content. At the same time, the system provides interactive guidance, helping users clarify the core of the demand through questions or prompts, and reducing the use threshold of the system.

[0064] In this embodiment, to ensure that the system is always in the best state during long-term operation, a system optimization and continuous learning mechanism is designed. During system operation, user feedback and new artistic data are continuously collected to optimize and adjust the system. Through feedback information such as user satisfaction evaluation and modification suggestions, the user's satisfaction with the generated image and improvement suggestions are analyzed. Based on the feedback data, the parameters and processing strategies of the model are adjusted to optimize the effect of image generation. New artistic data and research results are used to continuously train and update the industry large model. Through online learning or incremental learning, the model can learn new knowledge, new elements and new trends in the industry in a timely manner, quickly adapt to the dynamic changes in the art field. A performance monitoring module is established to track key indicators such as image generation accuracy, response time, resource utilization, etc. in real time. According to the monitoring data, the model architecture is optimized, the hyperparameters are adjusted, and the computing resources are reasonably allocated to improve the efficiency and stability of the system. High-precision texture enhancement technology is updated regularly to ensure that the generated image always maintains the leading level of detail performance in the industry.

[0065] As can be seen from the above, the embodiments of the present application fine-tune the general large model using massive professional data in the national style field, enabling the model to deeply learn the connotation and extension of national style artistic elements, including the smoothness of lines, the coordination of colors, the complexity of textures, and the consistency of overall style. In this way, the system can ensure that the generated images maintain high consistency in overall style, avoiding visual inconsistency problems caused by mixed styles. This style consistency not only improves the aesthetics of the image, but also makes it more suitable for artistic creation and cultural heritage scenarios, ensuring that the generated images can directly meet professional needs. The industry large model is used to perform deep semantic analysis on national style artistic elements, identify key detail features in the image, and then enhance these features through texture generation algorithms to ensure that the generated image meets the standards of professional artistic creation in terms of detail restoration and artistic expression. Thus, national style images with consistent style and detail restoration can be generated.

[0066] Referring to Figure 3 The embodiments of the present application disclose a national style image generation device based on an industry large model, as shown in the drawings, comprising:

[0067] The target information data acquisition module 11 is used to acquire information data in the national style field and preprocess the information data to obtain target information data; the national style field includes the painting field, the calligraphy field, the architecture field, and the costume field; the information data includes structured artistic element data and unstructured image data and text data;

[0068] The preset industry large model construction module 12 is used to construct an initial industry large model and train the initial industry large model based on the target information data to obtain a preset industry large model;

[0069] The user input information acquisition module 13 is configured to acquire user input information, and parse the user input information by using the preset industry large model to obtain target demand information; the target demand information includes subject information, behavior information, object information, and artistic relationship;

[0070] The target national style image generation module 14 is configured to generate a target national style image based on the target demand information by using the preset industry large model and a preset texture enhancement technology.

[0071] As can be seen from the above, the embodiments of the present application acquire information data in the national style field, and train an initial industry large model by using the information data, so that the model can deeply learn the connotation and denotation of national style artistic elements, including the fluency of lines, the coordination of colors, the complexity of textures, and the consistency of overall style. The generated image is optimized layer by layer by using a high-precision texture enhancement technology. Therefore, a national style image with consistent style and restored details can be generated.

[0072] In some specific embodiments, the target information data acquisition module 11 can specifically include:

[0073] The first information data acquisition unit is configured to, if the information data is structured data, perform API interfacing with a target data source in the national style field to update the information data in the national style field in real time;

[0074] The second information data acquisition unit is configured to, if the information data is unstructured data, use image processing technology to screen to acquire the information data in the national style field;

[0075] The first information data processing unit is configured to, if the information data is structured data, use a preset segmentation algorithm to perform block processing on the information data to obtain target information data;

[0076] The second information data processing unit is configured to, if the information data is unstructured data, perform semantic labeling on the information data by using a named entity recognition technology to obtain target information data.

[0077] In some specific embodiments, the preset industry large model construction module 12 can specifically include:

[0078] The industry large model training unit is configured to train the initial industry large model based on the target information data by using a transfer learning technology to obtain a preset industry large model.

[0079] In some specific embodiments, the user input information acquisition module 13 can specifically include:

[0080] The target demand information acquisition unit is configured to perform semantic analysis on the user input information by using the preset industry large model to obtain target demand information.

[0081] In some embodiments, the target national style image generation module 14 can specifically include:

[0082] The national style image generation unit is configured to generate a national style image based on the target demand information by using the preset industry large model.

[0083] The texture enhancement unit is configured to identify detailed features in the national style image by using the preset industry large model, and perform enhancement processing on the identified detailed features by using a preset texture generation algorithm to obtain a target national style image.

[0084] In some embodiments, the national style image generation device based on the industry large model can further include:

[0085] The security protection unit is configured to protect data in the preset industry large model by sensitive information encryption, access permission control, and compliance verification mechanism.

[0086] Further, the present application also discloses an electronic device, Figure 4 The electronic device 20 shown in the figure is not considered as any limitation on the use range of the present application. The electronic device 20 can specifically include at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the related steps in the national style image generation method based on the industry large model disclosed in any of the preceding embodiments. In addition, the electronic device 20 in the present embodiment can be an electronic computer.

[0087] In the present embodiment, the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol followed by the communication interface 24 can be any communication protocol applicable to the technical solution of the present application, which is not limited here; the input / output interface 25 is used to obtain external input data or output data to the outside, and the specific interface type can be selected according to the specific application needs, which is not limited here.

[0088] In addition, the memory 22 can be a read-only memory, a random access memory, a magnetic disk or an optical disk, etc. as a carrier for storing resources, and the resources stored thereon can include an operating system 221, a computer program 222, etc. The storage mode can be temporary storage or permanent storage.

[0089] The operating system 221 is used to manage and control various hardware devices on the electronic device 20 and the computer program 222, and can be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of completing the industry large model-based Chinese style image generation method disclosed in any of the preceding embodiments and executed by the electronic device 20, the computer program 222 can further include a computer program capable of completing other specific work.

[0090] Further, the present application also discloses a computer readable storage medium for storing a computer program; wherein the computer program is executed by a processor to implement the industry large model-based Chinese style image generation method disclosed above. For specific steps of the method, please refer to the corresponding content disclosed in the preceding embodiments, which will not be described here.

[0091] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. For the same or similar parts between the embodiments, please refer to each other. For the device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and please refer to the method part for the relevant part.

[0092] The skilled person can further realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware, computer software or a combination of the two. In order to clearly show the interchangeability of hardware and software, the components and steps of the examples have been described in the above description. Whether the functions are realized by hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0093] The steps of the method or algorithm described in combination with the embodiments disclosed herein can be directly implemented by hardware, a software module executed by a processor, or a combination of the two. The software module can be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0094] Finally, it needs to be pointed out that in this document, relational terms such as first and second and the like can only be intended to distinguish one entity or operation from another entity or operation without necessarily requiring or implying any actual such relationship or order between such entities or operations. Moreover, the terms "comprising", "including", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without more limitations, an element defined by the statement "comprising a" does not exclude the existence of additional identical elements in the process, method, article, or apparatus including the stated element.

[0095] The above detailed description of the technical solutions provided by the present application has been provided, and the principles and implementation modes of the present application have been described by applying specific examples. The above description of the examples is only for the purpose of helping to understand the method of the present application and its core idea; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed, and the above description of the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method for generating national style images based on a large industry model, characterized in that: include: Acquire information data in the national wind field, and pre-process the information data to obtain target information data; The fields of national style include painting, calligraphy, architecture, and clothing; the information data includes structured art element data and unstructured image data and text data; Constructing an initial industry big model, and training the initial industry big model based on the target information data to obtain a preset industry big model; Obtaining user input information, and parsing the user input information using the preset industry macro model to obtain target demand information; The target demand information includes subject information, behavior information, object information and artistic relationship; The preset industry large model and the preset texture enhancement technology are used to generate a target national style image based on the target demand information.

2. The method for generating national style images based on industry large models according to claim 1 is characterized in that: The information data obtained in the field of national style includes: If the information data is structured data, then API docking is performed with the target data source in the national style field so as to synchronously update the information data in the national style field in real time; If the information data is unstructured data, image processing technology is used to filter it to obtain the information data in the national style field.

3. The method for generating national style images based on industry large models according to claim 1 is characterized in that: The preprocessing of the information data to obtain target information data includes: If the information data is structured data, the information data is segmented using a preset segmentation algorithm to obtain target information data; If the information data is unstructured data, semantic annotation is performed on the information data using named entity recognition technology to obtain target information data.

4. The method for generating national style images based on industry large models according to claim 1 is characterized in that: The training of the initial industry macro model based on the target information data to obtain a preset industry macro model includes: The initial industry big model is trained based on the target information data using transfer learning technology to obtain a preset industry big model.

5. The method for generating national style images based on industry large models according to claim 1 is characterized in that: The utilizing of the preset industry macro model to analyze the user input information to obtain target demand information includes: The preset industry macro model is used to perform semantic parsing operations on the user input information to obtain target demand information.

6. The method for generating national style images based on industry large models according to claim 1 is characterized in that: The generating of a target national style image based on the target demand information by using the preset industry large model and the preset texture enhancement technology includes: Generate a national style image based on the target demand information using the preset industry macro model; The preset industry large model is used to identify the detail features in the traditional Chinese style image, and the preset texture generation algorithm is used to enhance the identified detail features to obtain the target traditional Chinese style image.

7. The method for generating national style images based on industry large models according to any one of claims 1 to 6, characterized in that: Also includes: The data in the preset industry big model is protected through sensitive information encryption, access permission control, and compliance verification mechanisms.

8. A national style image generation device based on an industry large model, characterized in that: include: The target information data acquisition module is used to acquire information data in the field of national style and pre-process the information data to obtain target information data; The fields of national style include painting, calligraphy, architecture, and clothing; the information data includes structured art element data and unstructured image data and text data; A preset industry big model construction module is used to construct an initial industry big model and train the initial industry big model based on the target information data to obtain a preset industry big model; A user input information acquisition module is used to acquire user input information and analyze the user input information using the preset industry macro model to obtain target demand information; The target demand information includes subject information, behavior information, object information and artistic relationship; The target national style image generation module is used to generate the target national style image based on the target demand information by using the preset industry large model and the preset texture enhancement technology.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor is used to execute the computer program to implement the method for generating national style images based on industry large models as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that Used to store a computer program, wherein when the computer program is executed by a processor, it implements the method for generating Chinese style images based on an industry large model as described in any one of claims 1 to 7.

Images