Poster generation method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention relates to the artificial intelligence technology, in particular to a poster generation method and device, electronic equipment and a storage medium. Obtaining target poster elements through multiple rounds of inquiry; generating a to-be-audited poster according to the target poster element through a trained poster generation model; through a trained poster auditing model, according to an auditing rule corresponding to the poster to be audited, content auditing is conducted on the poster to be audited, and an auditing result is obtained; and if the audit result is that the content audit is passed, taking the to-be-audited poster as a target poster. According to the embodiment of the invention, the poster generation efficiency and auditing efficiency are improved, and the auditing cost is reduced.

Description

Technical Field

[0001] This application relates to artificial intelligence technology, and more particularly to a poster generation method, apparatus, electronic device, and storage medium. Background Technology

[0002] Commercial banks can enhance their brand image by creating marketing posters. Professional design and compliant information disclosure can increase customers' trust in the bank's security, while achieving low-cost and efficient customer acquisition and improving the value conversion of traffic across all channels.

[0003] In existing technologies, poster generation requires manual communication of requirements, and after the design is completed, it also requires manual compliance review before the final poster is obtained. Therefore, poster generation suffers from low efficiency and consumes a large amount of manual review. Summary of the Invention

[0004] This application provides a poster generation method, apparatus, electronic device, and storage medium to improve poster generation efficiency, reduce review costs, and increase review efficiency.

[0005] In a first aspect, embodiments of this application provide a poster generation method, which includes:

[0006] Obtain the target poster elements through multiple rounds of questioning;

[0007] Using a trained poster generation model, a poster to be reviewed is generated based on the target poster elements.

[0008] The trained poster review model performs content review on the poster to be reviewed according to the review rules corresponding to the poster to be reviewed, and obtains the review result.

[0009] If the content review result is passed, the poster to be reviewed will be used as the target poster.

[0010] Secondly, embodiments of this application also provide a poster generation device, which includes:

[0011] The target poster element acquisition module is used to acquire target poster elements through multiple rounds of queries;

[0012] The poster generation module is used to generate posters to be reviewed based on the target poster elements using a trained poster generation model.

[0013] The poster review module is used to review the content of the posters to be reviewed by a trained poster review model according to the review rules corresponding to the posters to be reviewed, and obtain the review results.

[0014] The target poster determination module is used to select the poster to be reviewed as the target poster if the content review result is that the content review is passed.

[0015] Thirdly, embodiments of this application also provide an electronic device, which includes:

[0016] One or more processors;

[0017] Storage device for storing one or more programs;

[0018] When one or more programs are executed by one or more processors, the one or more processors implement any of the poster generation methods provided in the embodiments of this application.

[0019] Fourthly, embodiments of this application also provide a storage medium including computer-executable instructions, which, when executed by a computer processor, are used to perform any of the poster generation methods provided in embodiments of this application.

[0020] This application employs multiple rounds of inquiry to obtain target poster elements, accurately capturing user needs and improving the accuracy of subsequently generated posters awaiting review. A trained poster generation model generates posters awaiting review based on the target poster elements, automatically generating these posters and improving generation efficiency. A trained poster review model performs content review on the posters according to the corresponding review rules, obtaining review results and automatically reviewing them, improving review efficiency and reducing costs. If the review result is a pass, the poster awaiting review is used as the target poster, and a trained artificial intelligence model is used for both generation and review, quickly obtaining the target poster and further improving generation efficiency. Therefore, this technical solution solves the problems of low poster generation efficiency and the need for significant manual review, achieving the effects of improving poster generation and review efficiency while reducing review costs. Attached Figure Description

[0021] Figure 1 This is a flowchart of a poster generation method according to Embodiment 1 of this application;

[0022] Figure 2 This is a flowchart of a poster generation method according to Embodiment 2 of this application;

[0023] Figure 3 This is a flowchart of a poster generation method according to Embodiment 3 of this application;

[0024] Figure 4 This is a flowchart of a poster generation method according to Embodiment 4 of this application;

[0025] Figure 5 This is a schematic diagram of the structure of a poster generating device according to Embodiment 5 of this application;

[0026] Figure 6 This is a schematic diagram of the structure of an electronic device according to Embodiment Six of this application. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0028] It should be noted that the terms "first" and "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] Example 1

[0030] Figure 1 This is a flowchart of a poster generation method provided in Embodiment 1 of this application. This embodiment is applicable to situations where financial institutions automatically generate posters. The method can be executed by a poster generation device, which can be implemented in software and / or hardware and specifically configured in an electronic device.

[0031] See Figure 1 The poster generation method shown includes the following steps:

[0032] S110. Obtain the target poster elements through multiple rounds of questioning.

[0033] Target poster elements are the elements that generate a poster and are used to describe the poster's characteristics. For example, target poster elements may include content, color, visual style, and target audience. For a marketing poster, target poster elements may include the marketing campaign theme, target audience, marketing campaign, marketing time, color, and visual style.

[0034] Users can input the relevant descriptions of the poster they need through the interactive interface using text or voice. The large model extracts the requirements from the user's input to obtain the target poster elements. When necessary target poster elements are missing, or when the user's input is unclear, the interactive interface can ask the user questions to guide them in inputting the corresponding target poster elements. If necessary, multiple rounds of questions can be conducted to accurately obtain all target poster elements, so that the generated poster meets the user's expected requirements.

[0035] S120. Using the trained poster generation model, generate a poster to be reviewed based on the target poster elements.

[0036] A trained poster generation model can be a text-to-image language model, used to transform text-described target poster elements into image-described posters. The model can be trained using historical poster requirements and corresponding posters. Specifically, it generates poster requirement text based on the target poster elements, generates composition prompts based on the poster requirement text, and inputs these prompts into the trained poster generation model to obtain the poster to be approved.

[0037] S130. Using the trained poster review model, the content of the poster to be reviewed is reviewed according to the review rules corresponding to the poster to be reviewed, and the review result is obtained.

[0038] A well-trained poster review model can serve as a large-scale video understanding model, enabling it to review the content of posters based on text descriptions and obtain review results. This model can be trained using historical review requirements and corresponding poster review results.

[0039] The review rules corresponding to the poster to be reviewed can be determined based on the content or type of the poster. The review rules and the poster to be reviewed are input into a trained poster review model, and the review results are output. The review results can include whether the content review passed, and if the content review fails, the reason for the failure is output. Review rules can include general rules and professional rules. General rules can be general review rules; for example, general rules can include terminology review, used to check whether the poster to be reviewed contains terms that do not conform to general specifications. Professional rules can be review rules specific to a particular field; for example, professional rules can include efficacy claim review, used to check whether the poster to be reviewed contains promotional content for corresponding efficacy claims.

[0040] S140. If the review result is that the content review is passed, the poster to be reviewed will be used as the target poster to be generated.

[0041] The target poster can be a poster that meets the review requirements and user needs. If the review result is that the content review is passed, meaning that the poster to be reviewed meets the relevant specifications, then the poster to be reviewed will be used to generate the target poster.

[0042] In an optional embodiment, if the review result is that the content review fails, the review result is returned to the trained poster generation model to instruct the trained poster generation model to make adaptive modifications to the poster to be reviewed, obtain an updated poster to be reviewed, and continue to review until the review result is that the content review passes.

[0043] If the review result is "content review failed," meaning the review result includes content that does not comply with the relevant review rules, the review result can include specific reasons for failure. Returning the review result to the trained poster generation model allows the model to adaptively modify the poster to be reviewed based on the reasons for failure, resulting in an updated poster. After obtaining the updated poster, it is input into the trained poster review model for further content review to obtain the review result. If the review result is "content review failed," the above steps are repeated until the review result is "content review passed."

[0044] By automatically modifying posters that fail content review through a trained poster generation model, the compliance and generation efficiency of the target posters can be guaranteed. At the same time, the review results are returned to the trained poster generation model, which can perform incremental training, learn relevant knowledge, avoid the same errors when generating posters for review in the next time, and improve the compliance of the output of the trained poster generation model.

[0045] It should be noted that all information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this disclosure are information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data comply with the relevant laws, regulations and standards of the relevant regions.

[0046] In the financial sector, especially in banking, marketing posters are a crucial frontline tool for attracting customers, promoting products / services, and enhancing brand image. Creating marketing posters helps strengthen brand image; professional design and compliant information disclosure can increase customer trust in the bank, while simultaneously achieving low-cost, high-efficiency customer acquisition and improving the value conversion of traffic across all channels. However, in an environment of fierce market competition and frequent shifts in trending topics, quickly capturing market opportunities, creating marketing plans, and ensuring the legality and compliance of marketing efforts have become urgent needs for commercial banks in their pursuit of market share.

[0047] The current marketing poster production process suffers from the following shortcomings: First, design efficiency is low. Repeated communication and adjustments are required before the final poster draft is confirmed, resulting in high communication costs. Second, production efficiency is low. To meet the marketing needs of different audiences and targets and quickly adapt to market demands, higher requirements are placed on poster generation efficiency, which manual generation cannot satisfy. Third, compliance review costs are high. Marketing plan design drafts require manual review for compliance, increasing the marketing preparation cycle and potentially leading to missed market opportunities. The entire production process, including requirements communication, poster design, and poster review, consumes a significant amount of manpower and is inefficient.

[0048] The technical solution of this embodiment obtains target poster elements through multiple rounds of queries, accurately understands user needs, and improves the accuracy of subsequently generated posters awaiting review. A trained poster generation model generates posters awaiting review based on the target poster elements, automatically generating posters for review, thus improving poster generation efficiency. A trained poster review model performs content review on posters awaiting review according to the review rules corresponding to them, obtaining review results. The trained poster review model automatically reviews posters awaiting review, improving review efficiency and reducing review costs. If the review result is a pass for content review, the poster awaiting review is used as the target poster. A trained artificial intelligence model is then used to generate and review the poster, quickly obtaining the target poster and improving poster generation efficiency. Therefore, the technical solution of this application solves the problems of low poster generation efficiency and the need for significant manual review, achieving the effects of improving poster generation and review efficiency while reducing review costs.

[0049] Example 2

[0050] Figure 2 This is a flowchart of a poster generation method provided in Embodiment 2 of this application. The technical solution of this embodiment is further refined based on the above technical solution.

[0051] Furthermore, the phrase "obtain target poster elements through multiple rounds of questioning" is refined to: "obtain the poster requirements input by the user; extract user demand elements from the poster requirements through the element extraction model; determine whether the user demand elements include all necessary poster elements; if so, use the user demand elements as target poster elements; otherwise, prompt the user with the missing necessary poster elements through at least one round of questioning until all necessary poster elements are extracted," in order to accurately obtain the target poster elements.

[0052] See Figure 2 The poster generation method shown includes:

[0053] S210. Obtain the poster requirements input by the user.

[0054] Users can input their requirements for the poster through the interactive interface of the feature extraction model, allowing the model to extract the user's desired elements from the poster requirements. For example, users can input their poster requirements via text or voice, and the interactive interface will then retrieve these inputs.

[0055] S220. Extract user demand elements from the poster requirements using a large-scale element extraction model.

[0056] The feature extraction model is a pre-trained large-scale model used to extract user demand elements. This model can be pre-trained using historical user demand documents labeled with user demand elements.

[0057] Input the poster requirements into the feature extraction model, and output the user demand elements.

[0058] In one optional embodiment, the user demand elements are extracted from the poster requirements using a large-scale element extraction model, including: if there is a new concept in the poster requirements, a retrieval plugin is invoked to perform a relevant retrieval of the new concept to obtain an auxiliary understanding document; the large-scale element extraction model is further trained using the auxiliary understanding document to obtain an updated large-scale element extraction model; and the user demand elements are extracted from the poster requirements using the updated large-scale element extraction model.

[0059] New concepts can be nouns or adjectives that have not appeared before. In the design field, new design style descriptions, color descriptions, and trending buzzwords frequently emerge. For example, in color descriptions, there are terms like "macaron" and "dopamine"; and in style descriptions, there's "new Chinese style." These are all terms that have appeared at a specific point in time, and these terms are considered new concepts. Therefore, to accurately extract user needs, when a new concept appears, a search plugin is used to perform relevant searches for the new concept, obtaining documentation to aid understanding.

[0060] The retrieval plugin can be an engine plugin for internet retrieval, used to retrieve explanatory texts related to new concepts and generate supplementary understanding documents. These documents can include retrieved concept explanations and example diagrams, used to supplement the training of the feature extraction model.

[0061] By supplementing the training of the feature extraction model with document comprehension aids, an updated feature extraction model is obtained. The updated feature extraction model can determine the user demand elements corresponding to the new concept. Therefore, the poster requirements are input into the updated feature extraction model, and the output is the user demand elements, thereby improving the accuracy of the updated feature extraction model in understanding the new concept.

[0062] By calling the retrieval plugin to perform relevant searches for new concepts, auxiliary understanding documents are obtained, enabling timely and accurate learning of new concepts. These documents are then used to supplement and train the element extraction model, resulting in an updated model capable of accurately understanding new concepts. This updated model then extracts user demand elements from poster requirements, ensuring the accuracy of user demand element extraction and addressing trending online terms and new terms and concepts in related fields.

[0063] In an optional embodiment, if a new concept exists in the poster requirements, the retrieval plugin is invoked to perform a relevant retrieval of the new concept. After obtaining the auxiliary understanding document, the method further includes: using the auxiliary understanding document to supplement the training of the poster generation model to obtain an updated poster generation model.

[0064] When a poster requires a new concept, the poster generation model is unlikely to possess the relevant knowledge of that concept. Therefore, to ensure that the poster generation model has the relevant knowledge of the new concept, it is supplemented with training through document comprehension to obtain an updated poster generation model. This ensures that the updated poster generation model has the relevant knowledge of the new concept and can accurately map it to the generated poster, thus guaranteeing the accuracy of the generated poster.

[0065] By assisting in understanding the document, the poster generation model is further trained to obtain an updated poster generation model. This ensures that the updated poster generation model possesses relevant knowledge of the new concepts and guarantees the accuracy of the updated poster generation model in expressing the new concepts.

[0066] S230. Determine whether the user needs elements include all necessary poster elements.

[0067] Essential poster elements can be any poster elements that are absolutely necessary. Different poster types may include different essential poster elements. Professional technicians can determine all the essential poster elements for each type of poster based on experience; this application does not specifically limit this. For example, for marketing posters, essential poster elements may include the marketing campaign theme, target audience, marketing activity, and marketing time. Based on the poster type, all corresponding essential poster elements are determined. Then, user requirement elements are matched one by one with all essential poster elements. If any essential poster elements remain, it can be determined that user requirement elements do not include all essential poster elements. If no essential poster elements remain, meaning all essential poster elements match their corresponding user requirement elements, it can be determined that user requirement elements include all essential poster elements.

[0068] S240. If so, then the user demand element will be used as the target poster element.

[0069] If so, meaning the user needs elements include all necessary poster elements, then the user needs elements are used as the target poster elements. The target poster elements can be the input to the trained poster generation model, used to generate the poster to be reviewed.

[0070] S250. Otherwise, prompt the user with the missing necessary poster elements through at least one round of questioning until all necessary poster elements are extracted.

[0071] Otherwise, if the user's needs do not include all necessary poster elements, the user can be prompted through the interactive interface to identify the missing elements until all necessary elements are retrieved. For example, if the target audience is missing, you could ask, "Which user groups are you targeting with your marketing campaign?" If the user's answer is incorrect, you can continue to guide them. For instance, if the target audience is missing and the user answers "unregistered customers," you could ask, "Do you target middle-aged, elderly, or youth groups?" The questions can be asked in multiple rounds from various angles until all angles have been explored or the desired answer has been obtained.

[0072] S260. Using the trained poster generation model, generate a poster to be reviewed based on the target poster elements.

[0073] S270. Using the trained poster review model, the content of the poster to be reviewed is reviewed according to the review rules corresponding to the poster to be reviewed, and the review result is obtained.

[0074] S280. If the review result is that the content review is passed, then the poster to be reviewed will be used as the target poster.

[0075] The technical solution of this embodiment obtains the poster requirements input by the user; extracts the user's required elements from the poster requirements through a large element extraction model; determines whether the user's required elements include all necessary poster elements; if so, the user's required elements are used as target poster elements; otherwise, the user is prompted with at least one round of queries to identify the missing necessary poster elements until all necessary poster elements are extracted, ensuring that new concepts can be identified and that the user can be guided to supplement the necessary poster elements in a timely manner, thus ensuring the comprehensiveness and accuracy of the necessary poster elements.

[0076] Example 3

[0077] Figure 3 This is a flowchart of a poster generation method provided in Embodiment 3 of this application. The technical solution of this embodiment is further refined based on the above technical solution.

[0078] Furthermore, the process of "using a trained poster review model to review the content of the poster to be reviewed according to the review rules corresponding to the poster to be reviewed, and obtaining the review result" is further refined into: "According to the type of the poster to be reviewed, obtain the review rules corresponding to the poster to be reviewed; call the large language model to generate review prompts according to the review rules; input the review prompts and the poster to be reviewed into the trained poster review model to obtain the review result", so as to automatically review the poster to be reviewed.

[0079] See Figure 3 The poster generation method shown includes:

[0080] S310. Obtain the target poster elements through multiple rounds of questioning.

[0081] S320. Using the trained poster generation model, generate a poster to be reviewed based on the target poster elements.

[0082] S330. Based on the type of the poster to be reviewed, obtain the review rules corresponding to the poster to be reviewed.

[0083] Posters awaiting review can be categorized into various types based on their intended use, which helps determine the corresponding review rules. Different types of posters may have different review rules. For example, poster types may include marketing, advertising, and recruitment, but this application does not specifically limit this. A corresponding rule file can be obtained based on the type of poster to be reviewed. After preprocessing the rule file, a large language model is invoked to generate at least one review rule based on the review rules. For example, for marketing posters, the corresponding rule file may include marketing and advertising compliance management measures and inspection guidelines. For example, preprocessing may include operations such as removing special characters, removing irrelevant information, removing duplicate descriptions, and word segmentation.

[0084] Rule files may be updated periodically. By generating audit rules after obtaining the rule files, it can be ensured that the audit rules can be updated in a timely manner after the rule files are updated, thus ensuring the real-time nature of the audit rules.

[0085] S340. Call the large language model to generate review prompts based on the review rules.

[0086] The review prompts can be generated by the poster review model. The large language model is invoked to automatically generate review prompts for the compliance check of posters awaiting review based on each review rule, ensuring that the posters comply with regulatory requirements. For example, regarding the rule of legal compliance, the review prompts could be as follows: As an expert familiar with the compliance review of commercial bank marketing content, you are reviewing the image content of marketing plans. Please determine whether the marketing plan images are compliant and filter out those that do not meet the requirements. When reviewing marketing images, please follow these rules: 1. The content of the marketing images must be legal and compliant, and must not violate public order and good morals; 2. The output should be returned in JSON format, including whether the marketing plan images are compliant and the reasons for failure.

[0087] Large language models can be pre-trained based on historical review rules and corresponding review prompts, improving the accuracy of review prompts generated by the trained large language model. This improves the efficiency and accuracy of review prompt generation by relying on the efficiency and accuracy of the large language model.

[0088] S350. Input the review prompts and the poster to be reviewed into the trained poster review model to obtain the review results.

[0089] Input the review prompts and the poster to be reviewed into the trained poster review model, and output the review results. This poster review model is a large-scale video understanding model capable of automatically reviewing images in the poster based on the review prompts. The review results include whether the content passed review, and the reasons for failure if the content failed review.

[0090] The poster review model is pre-trained based on historical posters, historical review rules, and corresponding review results. This pre-trained model ensures accuracy and improves the accuracy of review results. The efficiency and accuracy of the pre-trained poster review model enhance the efficiency and accuracy of review result generation.

[0091] S360. If the content review result is "passed", then the poster to be reviewed will be used as the target poster.

[0092] The technical solution of this embodiment obtains the review rules corresponding to the poster to be reviewed based on the type of the poster to be reviewed, and obtains the review rules corresponding to the poster type in real time. This allows for timely acquisition of new review rules after they are updated, ensuring the real-time nature of the review rules. A large language model is invoked to generate review prompts based on the review rules. The large model can quickly generate accurate review prompts, improving the efficiency and accuracy of prompt generation. The review prompts and the poster to be reviewed are input into a trained poster review model to obtain the review results, automatically completing the compliance review of the poster to be reviewed, reducing manual costs and improving the efficiency and accuracy of the review.

[0093] Example 4

[0094] Figure 4 This is a flowchart of a poster generation method provided in Embodiment 4 of this application. The technical solution of this embodiment is further refined based on the above technical solution.

[0095] Furthermore, the phrase "generating a poster to be reviewed based on the target poster elements using a trained poster generation model" is further refined into: "calling a large language model to generate composition prompts based on the target poster elements; inputting the composition prompts into the trained poster generation model to obtain the poster to be reviewed," thus automatically generating the poster to be reviewed.

[0096] See Figure 4 The poster generation method shown includes:

[0097] S410. Obtain the target poster elements through multiple rounds of questioning.

[0098] S420: Call the large language model to generate composition prompts based on the target poster elements.

[0099] Composition cues can be used in poster generation models. The target poster elements are input into a large language model, and the output is the composition cues. The large language model can be pre-trained based on historical target poster elements and corresponding composition cues, improving the accuracy of the generated composition cues. This leverages the efficiency and accuracy of the large language model to enhance the efficiency and accuracy of composition cue generation.

[0100] For example, a first language model can be called to generate a text scheme based on the target poster elements, and then a second language model can be called to generate compositional cues based on the text scheme. The first and second language models can be different or the same. The first language model can be pre-trained based on the target poster elements and the corresponding text scheme. The second language model can also be pre-trained based on the text scheme and the corresponding compositional cues.

[0101] For example, the text scheme generated by the first major language model based on the target poster elements can be as follows:

[0102] As a professional marketing planner and poster design expert, skilled in designing marketing posters that meet client requirements based on marketing objectives, please design n (e.g., 5) text-based design schemes for marketing posters based on the marketing campaign theme, target audience, campaign activities, and campaign timeline. When designing, please adhere to the following rules: 1. Consider relevant elements of the marketing campaign, including variables such as the campaign theme, target audience, campaign activities, and campaign timeline; 2. Utilize creativity to design n different styles of marketing posters; 3. Return the output in JSON format, including the subject and actions, scene and environment, visual style, poster size ratio, style control, and brand color scheme.

[0103] The image-based cue words generated by the second language model based on the text scheme can be as follows:

[0104] {Subject and Action}, {Scene and Environment}, {Visual Style}, {Size and Proportion}, {Style Control}, {Brand Color Scheme}. For example, a confidently smiling young woman is using a mobile phone, with a holographic banking interface floating above it, surrounded by digital particles and gold financial charts. The background is a modern minimalist style with a blue-gold gradient. Style: 3D rendering, hyper-realistic details, cinematic lighting and soft shadows. Composition: Centralized subject and shallow depth of field. Aspect Ratio: 16:9, clean and professional aesthetics, no text, brand red color scheme.

[0105] S430. Input the composition prompts into the trained poster generation model to obtain the poster to be reviewed.

[0106] Inputting composition prompts into a trained poster generation model outputs a poster awaiting review. Relying on the efficiency and accuracy of the poster generation model, posters awaiting review can be obtained efficiently. The poster generation model can be pre-trained using elements from multiple copyrighted historical target posters and their corresponding historical posters. The trained poster generation model possesses the ability to generate target posters from target poster elements, improving the quality of generated posters awaiting review and reducing manual labor costs.

[0107] S440. Using the trained poster review model, the content of the poster to be reviewed is reviewed according to the review rules corresponding to the poster to be reviewed, and the review result is obtained.

[0108] S450. If the review result is that the content review is passed, then the poster to be reviewed will be used as the target poster.

[0109] The technical solution of this embodiment generates composition prompts based on the target poster elements by calling a large language model. The large language model generates composition prompts quickly and accurately without the need for manual writing of prompts, reducing labor costs and improving the accuracy of composition prompts. The composition prompts are then input into a trained poster generation model to obtain a poster to be reviewed. The poster generation model's ability to generate images from text can be relied upon to quickly generate posters to be reviewed, improving poster generation efficiency.

[0110] Example 5

[0111] Figure 5 The diagram shown is a structural schematic of a poster generation device according to Embodiment 5 of this application. This embodiment is applicable to situations where financial institutions automatically generate posters. The specific structure of the poster generation device is as follows:

[0112] The target poster element acquisition module 510 is used to acquire target poster elements through multiple rounds of queries;

[0113] The poster generation module 520 is used to generate a poster to be reviewed based on the target poster elements using a trained poster generation model.

[0114] The poster review module 530 is used to review the content of the poster to be reviewed by a trained poster review model according to the review rules corresponding to the poster to be reviewed, and obtain the review result.

[0115] The target poster determination module 540 is used to select the poster to be reviewed as the target poster if the review result is that the content review is passed.

[0116] The technical solution of this embodiment obtains target poster elements through multiple rounds of queries, accurately understands user needs, and improves the accuracy of subsequently generated posters awaiting review. A trained poster generation model generates posters awaiting review based on the target poster elements, automatically generating posters for review, thus improving poster generation efficiency. A trained poster review model performs content review on posters awaiting review according to the review rules corresponding to them, obtaining review results. The trained poster review model automatically reviews posters awaiting review, improving review efficiency and reducing review costs. If the review result is a pass for content review, the poster awaiting review is used as the target poster. A trained artificial intelligence model is then used to generate and review the poster, quickly obtaining the target poster and improving poster generation efficiency. Therefore, the technical solution of this application solves the problems of low poster generation efficiency and the need for significant manual review, achieving the effects of improving poster generation and review efficiency while reducing review costs.

[0117] Optional, the target poster element acquisition module 510 includes:

[0118] The poster requirement acquisition unit is used to acquire the poster requirements input by the user.

[0119] The User Requirement Element Extraction Unit is used to extract user requirement elements from the poster requirements through the element extraction model.

[0120] The necessary poster element determination unit is used to determine whether the user's needs include all necessary poster elements.

[0121] The target poster element determination unit is used to determine the user demand elements as target poster elements if the condition is met.

[0122] The necessary poster element query unit is used to prompt the user for missing necessary poster elements through at least one round of queries until all necessary poster elements are extracted.

[0123] Optional, the user requirement element extraction unit includes:

[0124] The new concept retrieval subunit is used to call the retrieval plugin to perform relevant searches for new concepts if the poster requirements contain new concepts, and obtain documents to aid understanding.

[0125] The feature extraction model training subunit is used to supplement the feature extraction model by assisting in document understanding, so as to obtain an updated feature extraction model.

[0126] The User Requirement Element Extraction Subunit is used to extract user requirement elements from the poster requirements using the updated element extraction model.

[0127] Optionally, the user demand element extraction unit may also include:

[0128] The poster generation model training subunit is used to supplement the training of the poster generation model by assisting in document understanding, so as to obtain an updated poster generation model.

[0129] Optional, the poster review module 530 includes:

[0130] The review rules acquisition unit is used to obtain the review rules corresponding to the poster to be reviewed based on the type of the poster to be reviewed;

[0131] The review prompt word generation unit is used to call the large language model and generate review prompt words according to the review rules.

[0132] The review result determination unit is used to input the review prompts and the poster to be reviewed into the trained poster review model to obtain the review result.

[0133] Optional, the poster generation module 520 to be reviewed includes:

[0134] The composition prompt generation unit is used to call the large language model to generate composition prompts based on the target poster elements.

[0135] The poster generation unit is used to input composition prompts into the trained poster generation model to obtain the poster to be reviewed.

[0136] The poster generation apparatus provided in this application embodiment can execute the poster generation method provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects for executing the poster generation method.

[0137] According to embodiments of the present invention, the present invention also provides an electronic device, a readable storage medium, and a computer program product.

[0138] Example 6

[0139] Figure 6 This is a schematic diagram of the structure of an electronic device provided in Embodiment Six of this application, as shown below. Figure 6 As shown, the electronic device includes a processor 610, a memory 620, an input device 630, and an output device 640; the number of processors 610 in the electronic device can be one or more. Figure 6 Taking a processor 610 as an example; the processor 610, memory 620, input device 630, and output device 640 in the electronic device can be connected via a bus or other means. Figure 6 Taking the example of a connection between China and Israel via a bus.

[0140] The memory 620, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the poster generation method in this application embodiment (e.g., the target poster element acquisition module 510, the poster generation module 520 to be reviewed, the poster review module 530 to be reviewed, and the target poster determination module 540). The processor 610 executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory 620, thereby implementing the above-described poster generation method.

[0141] The memory 620 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function; the data storage area may store data created based on terminal usage. Furthermore, the memory 620 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 620 may further include memory remotely located relative to the processor 610, which can be connected to the electronic device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0142] Input device 630 can be used to receive input character information and generate key signal inputs related to user settings and function control of the electronic device. Output device 640 may include display devices such as a display screen.

[0143] Example 7

[0144] Embodiment 7 of this application also provides a storage medium containing computer-executable instructions. When executed by a computer processor, the computer-executable instructions are used to execute a poster generation method. The method includes: obtaining target poster elements through multiple rounds of querying; generating a poster to be reviewed based on the target poster elements using a trained poster generation model; performing content review on the poster to be reviewed according to the review rules corresponding to the poster to be reviewed using a trained poster review model, and obtaining a review result; if the review result is that the content review is passed, then the poster to be reviewed is used as the target poster.

[0145] Of course, the computer-executable instructions provided in the embodiments of this application are not limited to the method operations described above, but can also perform related operations in the poster generation method provided in any embodiment of this application.

[0146] Based on the above description of the implementation methods, those skilled in the art can clearly understand that this application can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0147] It is worth noting that in the above-described embodiments of the poster generation device, the various units and modules are divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of this application.

[0148] Note that the above are merely preferred embodiments and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application, the scope of which is determined by the scope of the appended claims.

Claims

1. A poster generation method, characterized in that, include: Obtain the target poster elements through multiple rounds of questioning; Using a trained poster generation model, a poster to be reviewed is generated based on the target poster elements. By using a trained poster review model, the content of the poster to be reviewed is reviewed according to the review rules corresponding to the poster to be reviewed, and the review result is obtained. If the review result is that the content review is passed, then the poster to be reviewed will be used as the target poster.

2. The method according to claim 1, characterized in that, The process of obtaining target poster elements through multiple rounds of questioning includes: Obtain the poster requirements input by the user; The user demand elements are extracted from the poster requirements using a large-scale element extraction model. Determine whether the user needs elements include all necessary poster elements; If so, then the user demand elements will be used as target poster elements; Otherwise, prompt the user with the missing essential poster elements through at least one round of questioning until all essential poster elements are extracted.

3. The method according to claim 2, characterized in that, The step of extracting user demand elements from the poster requirements using a large-scale element extraction model includes: If the poster requirements contain new concepts, the search plugin will be invoked to perform relevant searches for the new concepts and obtain documents to aid understanding. By using the aided document comprehension, the feature extraction model is further trained to obtain an updated feature extraction model. The updated feature extraction model is used to extract user demand elements from the poster requirements.

4. The method according to claim 3, characterized in that, If a new concept exists in the poster requirements, the search plugin is invoked to perform a relevant search for the new concept and obtain documents to aid understanding. This process also includes: The poster generation model is further trained using the aided document comprehension tool to obtain an updated poster generation model.

5. The method according to claim 1, characterized in that, The pre-trained poster review model performs content review on the poster to be reviewed according to the review rules corresponding to the poster to be reviewed, and obtains the review result, including: Based on the type of the poster to be reviewed, obtain the review rules corresponding to the poster to be reviewed; The large language model is invoked to generate approval prompts based on the aforementioned approval rules. The review prompt and the poster to be reviewed are input into the trained poster review model to obtain the review result.

6. The method according to claim 1, characterized in that, The step of generating a poster for review based on the target poster elements using a trained poster generation model includes: The large language model is invoked to generate composition prompts based on the target poster elements; Input the composition prompts into the trained poster generation model to obtain the poster to be reviewed.

7. A poster generating device, characterized in that, include: The target poster element acquisition module is used to acquire target poster elements through multiple rounds of queries; The poster generation module is used to generate a poster to be reviewed based on the target poster elements using a trained poster generation model. The poster review module is used to review the content of the poster to be reviewed by a trained poster review model according to the review rules corresponding to the poster to be reviewed, and obtain the review result. The target poster determination module is used to designate the poster to be reviewed as the target poster if the review result is that the content review is passed.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the poster generation method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the poster generation method as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the poster generation method according to any one of claims 1-6.

Images