Poster generation method and device, computer equipment and storage medium

Abstract

The invention provides a poster generation method and device, computer equipment and a storage medium. The method comprises the steps of obtaining product description information input by a user; according to the product description information, determining a product type, a target poster copywriting and a target poster element; selecting a corresponding target layout template according to the product type; and generating a target poster according to the target poster copywriting, the target poster element and the target layout template. According to the scheme, the automatic process from user input to complete poster generation is achieved, through automatic poster element recommendation, automatic copywriting generation and automatic layout adjustment, the user can easily complete poster design, the operation steps and design difficulty of the user are reduced, the poster manufacturing efficiency is greatly improved, the design cost is reduced, and the user experience is improved. The product propaganda requirement can be better met, and the method is particularly friendly to users without design experience such as peasant households and small commercial tenants.

Description

Technical Field

[0001] This application relates to the field of Internet technologies, and in particular, to a poster generation method, apparatus, computer device, and storage medium. Background Art

[0002] Currently, poster generation technology is mostly used in commercial advertising design, which usually requires relatively high design skills and operation complexity. Especially for merchants lacking design experience (such as farmers and small businesses), they face difficulties in generating promotional posters. Although traditional poster generation tools provide relatively rich design functions, they lack personalized adaptation for organic products, and the operation interface is complex, making it difficult to meet the needs of non-professional users such as farmers. Summary of the Invention

[0003] The purpose of this application aims to solve at least one of the above technical defects, especially problems such as the difficulty in meeting the needs of non-professional users such as farmers in the prior art.

[0004] In a first aspect, this application provides a poster generation method, including:

[0005] Obtain product description information input by a user;

[0006] Determine the product type, target poster copywriting, and target poster elements according to the product description information;

[0007] Select a corresponding target layout template according to the product type;

[0008] Generate a target poster according to the target poster copywriting, target poster elements, and target layout template.

[0009] In one embodiment, the generation process of the target poster copywriting includes:

[0010] Determine the original place of production according to the product description information;

[0011] According to the dialect corpus corresponding to the original place of production and the product description information, instruct a large language model to generate the target poster copywriting according to the words in the dialect corpus and the product description information.

[0012] In one embodiment, each product type corresponds to a template library, and the selection process of the target layout template includes:

[0013] Input the product description information into a visual attention prediction model, and output a heat map corresponding to the distribution of the user's visual focus;

[0014] For any candidate layout template in the template library corresponding to the product type, determine the first matching degree of the candidate layout template according to the position matching degree between each element area in the candidate layout template and the heat map and the preset weight value corresponding to each element area.

[0015] Select the target layout template with the highest first matching degree from the candidate layout templates.

[0016] In one embodiment, the selection process of the target poster element includes:

[0017] Determine the original place of origin according to the product description information;

[0018] For any type of element, input each candidate poster element belonging to the element type and the product description information into the multimodal matching model in sequence from the local element library corresponding to the original place of origin, and obtain the second matching degree between the candidate poster element and the product description information;

[0019] Take the candidate poster element with the highest second matching degree as the target poster element under the corresponding element type.

[0020] In one embodiment, generating a target poster according to the target poster copywriting, target poster elements and target layout template includes:

[0021] Generate a preview poster according to the target poster copywriting, target poster elements and target layout template;

[0022] In response to the zoom operation on the preview poster, record the number of operations in the corresponding zoom area;

[0023] Determine the area to be optimized where the number of operations is greater than the set threshold;

[0024] Enlarge the font or increase the resolution of the area to be optimized.

[0025] In one embodiment, after generating the target poster according to the target poster copywriting, target poster elements and target layout template, it further includes:

[0026] Obtain the screen resolution of the target display device;

[0027] Adjust the resolution of the target poster according to the screen resolution; the higher the screen resolution, the higher the corresponding resolution of the target poster.

[0028] In one embodiment, before generating the target poster according to the target poster copywriting, target poster elements and target layout template, it further includes:

[0029] In response to the user's modification operation, modify the target poster copywriting, target poster elements and / or target layout template.

[0030] In a second aspect, the present application provides a poster generation device, including:

[0031] A data acquisition module for acquiring product description information input by a user;

[0032] A material determination module, configured to determine a product type, a target poster copy, and target poster elements according to product description information;

[0033] A template determination module, configured to select a corresponding target layout template according to the product type;

[0034] A poster generation module, configured to generate a target poster according to the target poster copy, the target poster elements, and the target layout template.

[0035] Thirdly, the present application provides a computer device, including one or more processors and a memory. When computer-readable instructions stored in the memory are executed by the one or more processors, the steps of the poster generation method in any of the above embodiments are performed.

[0036] Fourthly, the present application provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors are caused to perform the steps of the poster generation method in any of the above embodiments.

[0037] As can be seen from the above technical solutions, the embodiments of the present application have the following advantages:

[0038] This solution realizes an automated process from user input to generating a complete poster. By automatically recommending poster elements, automatically generating copy, and automatically adjusting the layout, it enables users to easily complete poster design, reduces the user's operation steps and design difficulty, greatly improves the poster production efficiency, reduces the design cost, can better meet the product promotion needs, and is particularly friendly to users without design experience such as farmers and small merchants. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] 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 required for description in the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained according to these drawings without creative efforts.

[0040] Figure 1 A flowchart of a poster generation method provided in an embodiment of the present application;

[0041] Figure 2 A flowchart of generating a target poster copy in an embodiment of the present application;

[0042] Figure 3 A flowchart of selecting a target layout template in an embodiment of the present application;

[0043] Figure 4 Schematic diagram of the process of selecting target layout elements in an embodiment of the present application;

[0044] Figure 5 Schematic diagram of the process of optimizing a preview poster in an embodiment of the present application;

[0045] Figure 6 Schematic diagram of the process of optimizing resolution in an embodiment of the present application;

[0046] Figure 7 Internal structure diagram of a computer device provided in an embodiment of the present application. Detailed implementation manners

[0047] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. The embodiments described in the specification are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0048] The embodiment of the present application provides a poster generation method, including steps S102 to S108.

[0049] S102, obtain the product description information input by the user.

[0050] It can be understood that the product description information here refers to the textual description content provided by the user for the product associated with the poster they expect to generate through a specific input interface. This information can cover various aspects of the product, such as key points like the product category, variety, origin, characteristics, etc., and is an important basis for the subsequent poster generation process. The user input interface can be implemented through a common text input box component. Specifically, a clearly marked input box is set on the web page or application interface, and the user can input the product description information here. After the user finishes inputting and submits, the front-end code can send the input text information to the back-end server through the HTTP protocol. At the back-end, a service can be built using Python's Flask or Django framework to receive and process this information. In addition, the user can also describe the product information by voice. Specifically, a voice recognition engine is used to convert the voice into text, which can provide a more convenient input method for the user, especially in scenarios where the user is not convenient to input manually or has a low level of education.

[0051] S104, determine the product type, target poster copywriting, and target poster elements according to the product description information.

[0052] It can be understood that the product type here refers to the classification of products based on factors such as the core features and the field to which the product belongs, such as technology products, health products, fashion products, etc. The target poster copy is the promotional text content generated around the product features, advantages, etc. for display on the poster. The target poster elements include visual elements such as graphics, icons, colors, fonts, etc. related to the product.

[0053] Specifically, natural language processing technology and knowledge graphs and other technologies can be used to deeply analyze the product description information obtained in step S102. The text classification algorithm in natural language processing technology can be used to determine the product type. By using a large number of pre-constructed text datasets with labeled product types, a classification model is trained, such as a support vector machine (SVM) classifier or a convolutional neural network (CNN) classification model based on deep learning. When the product description information is input, the model determines the product type to which it belongs according to the text features. For the generation of the target poster copy, a template-based method or generative adversarial network (GAN) technology can be used. The generative adversarial network generates novel copy that conforms to the product features through the adversarial training of the generator and the discriminator.

[0054] There are multiple different types of elements required when producing a poster, and at least one target poster element is selected for each type of element. The target poster elements can be screened from the element library according to the product description information. Specifically, when faced with the product description information, the system first performs text preprocessing on the description information, for example, using lexical analysis and syntactic analysis technologies in natural language processing to extract keywords and key phrases in the product description information. Then, the extracted text content is text-matched with the labels carried by the elements in the element library. The matching algorithm can adopt common text similarity calculation methods such as the cosine similarity algorithm, and by calculating the cosine value of the vector angle between the description information text and the label text, the similarity between the two is judged. Of course, the target poster elements can also be directly obtained by inputting the product description information into a text-to-image model.

[0055] S106, select the corresponding target layout template according to the product type.

[0056] It can be understood that the target layout template is a pre-designed layout structure for poster pages of different product types, including design schemes such as the position distribution and size ratio of elements like text and pictures on the poster. This step is based on determining the product type in step S104. The system internally maintains a layout template library, and each template corresponds to a specific product type. After determining the product type, the system retrieves and selects the corresponding target layout template from the template library according to the identifier of the product type. This process is similar to an index-based query operation in a database, with the product type as the index to quickly locate the matching layout template. For example, the poster layout of electronic products may focus more on the display of product pictures, placing the product picture in the center of the poster and occupying a large area, with text descriptions distributed around the picture; while the poster layout of food products may emphasize the bright color combination and the combination of food pictures and promotional information.

[0057] S108. Generate a target poster based on the target poster copywriting, target poster elements, and target layout template.

[0058] It can be understood that the target poster is the final complete poster work that meets the user's product promotion needs. This step is a process of integrating and rendering the target poster copywriting and target poster elements determined in the previous steps according to the structure and style specified by the target layout template. Specifically, the target layout template stipulates the central position and occupied area size of each target poster element and the copywriting part. Thus, the pictures in the target poster elements can be pasted to the specified positions in the template, and the target poster copywriting can also be pasted to the specified positions after generating an image using the fonts specified in the target poster elements. Finally, the poster can be saved in common image formats such as JPEG or PNG.

[0059] This solution realizes an automated process from user input to generating a complete poster. By automatically recommending poster elements, automatically generating copywriting, and automatically adjusting the layout, it enables users to easily complete poster design, reduces the user's operation steps and design difficulty, greatly improves the poster production efficiency, reduces the design cost, and can better meet the product promotion needs, which is particularly friendly to users without design experience such as farmers and small merchants.

[0060] In one embodiment, please refer to Figure 2 , the generation process of the target poster copywriting includes steps S202 to S206.

[0061] S202. Determine the original place of production according to the product description information.

[0062] It can be understood that in this poster copywriting generation process, the original place of origin refers to the source place of the product. From a principle perspective, determining the original place of origin requires the use of text analysis technology to deeply analyze the product description information. The product description information usually contains rich clues, which may directly mention the name of the place of origin or indirectly imply the place of origin through relevant features. For example, if the product description is "cheese made from pure milk source in xx mountainous area", by using text analysis technology to identify the keyword "xx mountainous area" in it, the original place of origin can be determined. This process often involves the named entity recognition (NER) technology in natural language processing, which can identify entities with specific meanings from the text, such as place names, organization names, etc. When determining the original place of origin, the NER technology can extract the place of origin related words in the text. At the same time, knowledge graph technology may also be used. The knowledge graph shows entities and the relationships between entities in a structured form. When the text analysis extracts possible place of origin related words, they can be verified and supplemented in the knowledge graph to ensure the accurate determination of the original place of origin. For example, if only "a certain well-known wine" is mentioned in the text, through the association relationship between wine and the place of origin in the knowledge graph, combined with other feature descriptions of the product, its possible original place of origin can be inferred.

[0063] S204. According to the dialect corpus corresponding to the original place of origin and the product description information, instruct the large language model to generate the target poster copywriting according to the words in the dialect corpus and the product description information.

[0064] It can be understood that a dialect corpus is a database that collects and organizes dialect vocabulary, grammatical structures, and expressions in a specific region. Dialect corpora in different regions have unique language features, reflecting local cultures, living habits, etc. This step establishes a connection between the original place of origin and the dialect corpus, and uses product description information and the dialect corpus to instruct the large language model to work. First, according to the determined original place of origin, the corresponding dialect corpus is located from the pre-constructed set of dialect corpora. For example, if the original place of origin is Sichuan, the system will call the Sichuan dialect corpus. The principle of this step is to establish a connection between the original place of origin and the corresponding dialect corpus, combine the product description information with the dialect corpus, and provide accurate instructions for the large language model to generate the target poster copy. After the original place of origin of the product is determined through step S202, the system locates the matching dialect corpus from the pre-constructed set of dialect corpora. For example, if the original place of origin is Sichuan, the system will call the Sichuan dialect corpus. Although large language models, which are pre-trained on a vast amount of general text data, possess powerful language understanding and generation capabilities, they lack knowledge of specific regional cultural characteristics. At this time, the dialect corpus intervenes as a knowledge base. After the product description information is input into the system, the large language model uses retrieval techniques in natural language processing, taking key information such as words and phrases in the product description as the retrieval basis, and quickly matches in the dialect corpus. For example, for fruits with the origin in Sichuan, in the Sichuan dialect corpus, a characteristic description like "perfectly wonderful" can be retrieved to express the product quality. After receiving this input that combines the knowledge of the dialect corpus and the product description information, the large language model, based on the general language patterns and semantic understanding capabilities acquired through its pre-training, combines the regional characteristic knowledge provided by the dialect corpus to generate target poster copy that not only fits the product characteristics but also is rich in the cultural characteristics of the original place of origin.

[0065] In one embodiment, please refer to Figure 3 , each product type corresponds to a template library, and the process of selecting the target layout template includes steps S302 to S306.

[0066] S302, input the product description information into the visual attention prediction model, and output a heat map corresponding to the distribution of the user's visual focus.

[0067] It can be understood that the "visual attention prediction model" is a model constructed based on deep learning technology. Its core function is to simulate the attention patterns of the human visual system to different visual elements, and predict the distribution of the user's visual focus when observing relevant content by analyzing the input information. In this embodiment, the model receives product description information as input. Although the product description information is in text form, the model can achieve prediction through the cross - application of natural language processing and computer vision technologies. For example, the model first uses natural language processing technology to perform pre - processing operations such as word segmentation, part - of - speech tagging, and semantic understanding on the product description information, and extracts key product features, attributes, and other information. Then, based on a large number of pre - trained data pairs of images and corresponding description information, the text information is mapped to the visual space concept. The model learns the importance patterns of different product features in visual presentation through architectures such as convolutional neural networks (CNNs). When the product description information is input, the model predicts the possible positions of the user's visual focus when viewing a poster related to the product according to the learned patterns, and outputs the result in the form of a heat map. In the heat map, the darker the color area, the more concentrated the user's visual focus, and the lighter the color, the lower the degree of attention.

[0068] A mature deep - learning framework such as TensorFlow or PyTorch can be selected to build the visual attention prediction model. First, collect a large amount of product description information and corresponding user eye - movement tracking data (these data record the visual focus trajectories of users when viewing relevant posters) to construct a training dataset. During the training process, the product description information is processed by the natural language processing module and then input into the model. The predicted heat map output by the model is compared with the heat map generated from the actual eye - movement tracking data, and the model parameters are continuously adjusted through the back - propagation algorithm to improve the prediction accuracy.

[0069] S304. For any candidate layout template in the template library corresponding to the product type, determine the first matching degree of the candidate layout template according to the position matching degree between each element area in the candidate layout template and the heat map, and the preset weight value corresponding to each element area.

[0070] It can be understood that the template library is a pre-built database that stores a series of layout templates for different product types. Each template contains the design schemes such as the position distribution and size ratio of elements such as text and pictures on the poster. The candidate layout templates are the candidate templates in the template library that have not yet been determined as the final target layout template. The element area refers to the specific position range used to place different types of elements (such as product image area, title text area, description text area, etc.) in the candidate layout template. The preset weight value is pre-set according to the product type and the importance of the element in the poster, and is used to measure the relative importance of different element areas in the matching calculation. For example, for electronic products, the product image display area may be assigned a higher preset weight value because users tend to pay more attention to the appearance of the product.

[0071] This step evaluates the suitability of the template by calculating the degree of match between the layout template to be selected and the heat distribution map generated in step S302. First, the spatial position of each element area in the layout template to be selected is compared with the heat distribution map, and the degree of position matching can be determined by the area intersection ratio. For example, the position matching score of the product image element area is 8 points (out of 10 points), the preset weight value is 0.5, the position matching score of the title text area is 6 points, the preset weight value is 0.3, the position matching score of the description text area is 4 points, and the preset weight value is 0.2, then the first matching degree of the layout template to be selected is 8×0.5 + 6×0.3 + 4×0.2 = 6.6 points. In this way, the position matching of different element areas and their importance in the poster are comprehensively considered, so that the first matching degree can more accurately reflect the degree of fit between the layout template to be selected and the distribution of the user's visual focus.

[0072] S306: Select the layout template with the highest first matching degree from the candidate layout templates as the target layout template.

[0073] It can be understood that this step is the final decision link in the entire target layout template selection process. Its principle is based on the first matching degree of each candidate layout template calculated in the previous step S304. By comparing the first matching degrees of all candidate layout templates, the template with the highest value is selected as the target layout template. Because the first matching degree comprehensively considers the position matching degree between the candidate layout template and the user's visual focus distribution and the importance of each element area, selecting the template with the highest first matching degree can maximize the user's visual preference when viewing the poster, making the layout of the poster more in line with the user's attention pattern for product information, thereby improving the information transmission efficiency and attractiveness of the poster.

[0074] In one embodiment, see Figure 4, the selection process of the target poster element includes steps S402 to S406.

[0075] S402, determine the original place of origin according to the product description information.

[0076] For this step, reference can be made to step S202.

[0077] S404, for any type of element, sequentially input each candidate poster element belonging to the type of element from the local element library corresponding to the original place of origin and the product description information into the multi-modal matching model to obtain the second matching degree between the candidate poster element and the product description information.

[0078] It can be understood that the local element library is constructed for a specific original place of origin and is a database that collects various representative elements of that place. These elements cover various types such as graphics, icons, and colors and can reflect the local unique culture and regional characteristics. The candidate poster element is a candidate element in the local element library that has not been determined whether it will be used as the target poster element for a certain type of element (such as graphic elements, color elements, etc.). The multi-modal matching model is a model that integrates the processing capabilities of multiple data modalities (such as text, images, etc.), and its core function is to measure the matching degree between different modality data.

[0079] This step evaluates the fit between the candidate poster element and the product description information through the multi-modal matching model. After determining the original place of origin, the system extracts the candidate poster elements belonging to a certain type of element from the corresponding local element library. These candidate poster elements and the product description information are input into the multi-modal matching model together. If the candidate poster element is a graphic, the model first uses computer vision technology to extract the features of the graphic, such as edge detection, color histogram extraction, etc., to obtain the visual features of the graphic. At the same time, natural language processing technology is used to perform semantic analysis on the product description information to extract key product features, attributes, and other information. Then, the model calculates the second matching degree between the candidate poster element and the product description information in the multi-modal through a specific matching algorithm, such as cosine similarity calculation, deep learning-based feature vector matching, etc. This matching degree reflects the degree of tight association between the candidate poster element and the product description information in terms of semantics, vision, etc.

[0080] The multi-modal matching model can be built using a mature deep learning framework. For the local element library, a database management system such as MySQL can be used for storage and management to facilitate quick retrieval of candidate poster elements. In the model training stage, a large number of product description information from different origins and corresponding various poster elements are collected, and the matching relationships between them are labeled to construct a training dataset. During the training process, the poster elements and product description information are input into the model, and the matching degree results output by the model are compared with the labeled true matching relationships. The model parameters are continuously adjusted through the backpropagation algorithm to improve the accuracy of matching degree calculation.

[0081] S406, use the candidate poster element with the second highest matching degree as the target poster element under the corresponding element type.

[0082] It can be understood that this step is the final decision-making link in the process of selecting the target poster element. Its principle is based on the second matching degree of each candidate poster element calculated in step S404. By comparing the second matching degrees of all candidate poster elements belonging to the same element type, the candidate poster element with the highest value is selected as the target poster element under this element type. Because the second matching degree comprehensively considers the matching degree between the candidate poster element and the product description information in multiple modalities, selecting the candidate poster element with the highest second matching degree can maximize the fit between the poster element and the product description information, highlight the characteristics of the product and the cultural connotations of the original place of origin, thereby improving the overall quality and attractiveness of the poster.

[0083] In one embodiment, please refer to FIG. 5. Generating a target poster according to the target poster copywriting, target poster elements, and target layout template includes steps S502 to S508.

[0084] S502, generate a preview poster according to the target poster copywriting, target poster elements, and target layout template.

[0085] It can be understood that the preview poster is a preliminary integrated product generated according to the above-mentioned target poster copywriting, target poster elements, and target layout template, and is an intermediate product for users to view the general effect of the poster in advance. It can be implemented by means of the API of an image editing software.

[0086] S504, in response to the zoom-in operation on the preview poster, record the number of operations on the corresponding zoomed-in area.

[0087] It can be understood that the zoom-in operation refers to the behavior of a user magnifying a certain area of the preview poster through a specific operation (such as rolling the mouse wheel, pinching to zoom on the touch screen, etc.) when viewing the preview poster. The number of operations is the cumulative number of zoom-in operations performed by the user on a specific zoomed-in area.

[0088] The system needs to monitor the interaction operations between the user and the preview poster in real time. When a zoom-in operation event is detected, the system first determines the coordinate range of the area targeted by the zoom-in operation in the preview poster. This can be calculated by obtaining the coordinates of the mouse click position or the touch point on the touch screen and combining with the zoom ratio of the zoom-in operation. Then, the system maintains a record structure internally to store the coordinate range of each zoomed-in area and the corresponding number of operations. Whenever a zoom-in operation occurs for a certain area, the system checks whether the area already exists in the record structure. If it exists, the number of operations for the corresponding area is incremented by 1; if it does not exist, a new record is added to the record structure, recording the coordinate range of the area and initializing the number of operations to 1. In this way, the system can accurately record the number of operations for each zoomed-in area for subsequent analysis of the user's attention to different areas.

[0089] S506, determine the areas with the number of operations greater than the set threshold as the areas to be optimized.

[0090] It can be understood that the set threshold is a pre-set value, which is used as a criterion for judging whether a certain zoomed-in area is an area to be optimized. The areas to be optimized refer to the areas in the preview poster where the number of zoom-in operations by the user exceeds the set threshold and thus are considered to need further optimization to improve the display effect.

[0091] By traversing the record structure that records the number of operations of the zoomed-in areas, compare the number of operations of each area with the set threshold. If the number of operations of a certain area is greater than the set threshold, it means that the user shows a high degree of attention to this area and may want to view the content of this area more clearly. Therefore, such areas are determined as the areas to be optimized for subsequent targeted optimization to meet the user's needs for information display in this area. For example, if the set threshold is 5 times and the number of operations of a certain area reaches 6 times, then this area is determined as the area to be optimized.

[0092] S508, perform font enlargement or resolution improvement on the areas to be optimized.

[0093] It can be understood that the processing of the areas to be optimized aims to improve the clarity and readability of the content in these areas. Font enlargement is for the text content within the areas to be optimized, increasing the size of the text so that users can read it more clearly. Resolution improvement is to increase the resolution of the images in the areas to be optimized, making the image details clearer.

[0094] After determining the area to be optimized, the system first determines the main content type within the area to be optimized. If the main content is text, the system uses image editing technology to find the text layer within the area to be optimized and adjusts the font size of the text according to a preset magnification ratio. For example, the font size is increased from the original size 12 to size 16. If the main content is an image, the system uses an image super-resolution algorithm, such as the super-resolution convolutional neural network (SRCNN) algorithm based on deep learning, to process the image in this area and improve the image resolution. This algorithm reconstructs the low-resolution image in the area to be optimized by learning the mapping relationship between a large number of low-resolution images and their corresponding high-resolution images, generating a higher-resolution image, thereby enhancing the clarity of the image.

[0095] In one embodiment, after generating the target poster according to the target poster copywriting, target poster elements, and target layout template, refer to Figure 6 , and it further includes steps S602 to S604.

[0096] S602, obtain the screen resolution of the target display device.

[0097] It can be understood that the target display device refers to an electronic device finally used to display the target poster, such as a computer monitor, mobile phone screen, tablet computer display screen, electronic billboard, etc. "Screen resolution" is an important indicator to measure the fineness of screen display. It represents the number of pixels that the screen can display in the horizontal and vertical directions, usually presented in the form of "horizontal pixel number × vertical pixel number", such as the common 1920×1080, 2560×1440, etc.

[0098] The system needs to establish a communication connection with the target display device to obtain its screen resolution information. For different types of devices, the methods for obtaining the screen resolution are different. In a computer operating system, the operating system provides corresponding API functions to query the screen resolution. For example, the ratio of physical pixels to CSS pixels can be obtained through window.devicePixelRatio.

[0099] S604, adjust the resolution of the target poster according to the screen resolution. The higher the screen resolution, the higher the corresponding resolution of the target poster.

[0100] It can be understood that this step is based on obtaining the screen resolution of the target display device in step S602. The system adjusts the resolution of the target poster according to the correspondence between the screen resolution and the target poster resolution. Usually, an image scaling algorithm is used to achieve this adjustment. When the screen resolution is high, in order to make full use of the display capabilities of the screen and enable the poster to be clearly displayed on a high-resolution screen, it is necessary to increase the resolution of the target poster. For example, if the screen resolution of the target display device is 2560×1440 and the original target poster resolution is 1920×1080, the system can use an image interpolation algorithm, such as bilinear interpolation algorithm or bicubic interpolation algorithm, to insert new pixels between the pixels of the original poster image, increasing the number of pixels in the image, thereby increasing the resolution of the poster. On the contrary, when the screen resolution is low, it may be necessary to reduce the resolution of the target poster to avoid problems such as blurring or distortion when the poster is displayed on a low-resolution screen. In this case, a downsampling algorithm can be used to remove some pixels to reduce the resolution. In this way, it is possible to determine whether to generate a high-definition poster according to the performance of the user's device to meet the needs of different devices.

[0101] In one embodiment, before generating the target poster according to the target poster copywriting, target poster elements, and target layout template, it further includes: in response to a user's modification operation, modifying the target poster copywriting, target poster elements, and / or target layout template.

[0102] It can be understood that in the poster generation process, the "user's modification operation" refers to the behavior of the user expressing their adjustment requirements through specific interaction methods based on the review of the initially generated target poster copywriting, target poster elements, or target layout template. These operations aim to make the poster more in line with the user's expectations. When the user performs modification operations on the display interface (such as the poster editing page on the web or the poster generation interface of the application), for example, modifying the target poster copywriting in the text input box, changing the target poster elements through dragging or replacement operations, or switching the target layout template in the layout selection area, the system captures these operation events in real time. For the modification of the target poster copywriting, the system obtains the new text content input by the user through the event listening function of the text input box and overwrites the original copywriting information. If the target poster elements are modified, the system determines the element to be modified according to the user's selection operation (such as clicking to select a certain graphic element), and then updates the relevant attributes (position coordinates, element type, etc.) of the element according to the user's subsequent operations (such as dragging to change the position, replacing with a new element). In terms of modifying the target layout template, the system listens for the click event of the user on the layout template switching button. When the user selects a new layout template, the system replaces the original layout template information with the newly selected template information. In this way, the system can adjust the target poster copywriting, target poster elements, and target layout template in real time according to the user's modification operations, laying a foundation for generating a target poster that better meets the user's needs. Additionally, in some embodiments, the size and position of each target poster element can be adjusted on the preview poster interface.

[0103] The present application provides a poster generation device, including a data acquisition module, a material determination module, a template determination module, and a poster generation module. The data acquisition module is used to acquire the product description information input by the user. The material determination module is used to determine the product type, target poster copywriting, and target poster elements according to the product description information. The template determination module is used to select the corresponding target layout template according to the product type. The poster generation module is used to generate a target poster according to the target poster copywriting, target poster elements, and target layout template.

[0104] For the specific limitations of the poster generation device, reference can be made to the limitations of the poster generation method in the above text, which will not be elaborated here. Each module in the above poster generation device can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor in the computer device in the form of hardware or be independent of it, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules. It should be noted that the division of modules in the embodiments of the present application is illustrative, only a logical function division, and there may be other division methods in actual implementation.

[0105] The present application provides a computer device, including one or more processors, and a memory. Computer-readable instructions are stored in the memory. When the computer-readable instructions are executed by the one or more processors, the steps of the poster generation method in any of the above embodiments are executed.

[0106] Schematically, as Figure 7 shown, Figure 7 is a schematic internal structure diagram of a computer device provided by an embodiment of the present application. Referring to Figure 7 , the computer device 700 includes a processing component 702, which further includes one or more processors, and memory resources represented by a memory 701 for storing instructions executable by the processing component 702, such as application programs. The application programs stored in the memory 701 may include one or more than one module, each corresponding to a set of instructions. In addition, the processing component 702 is configured to execute instructions to perform the steps of the poster generation method in any of the above embodiments.

[0107] The computer device 700 may further include a power supply component 703 configured to perform power management of the computer device 700, a wired or wireless model interface 704 configured to connect the computer device 700 to a model, and an input / output (I / O) interface 705.

[0108] The present application provides a storage medium, in which computer-readable instructions are stored. When the computer-readable instructions are executed by one or more processors, the one or more processors are caused to execute the steps of the poster generation method in any of the above embodiments.

[0109] Finally, it should also be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0110] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0111] The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for generating a poster, characterized in that, including: Obtain the product description information input by the user; Determine the product type, target poster copywriting, and target poster elements according to the product description information; Select the corresponding target layout template according to the product type; Generate a target poster according to the target poster copywriting, the target poster elements, and the target layout template.

2. The poster generation method according to claim 1, wherein The generation process of the target poster copywriting includes: Determine the original place of origin according to the product description information; According to the dialect corpus corresponding to the original place of origin and the product description information, instruct the large language model to generate the target poster copywriting according to the words in the dialect corpus and the product description information.

3. The poster generation method according to claim 1, wherein Each product type corresponds to a template library, and the selection process of the target layout template includes: Input the product description information into the visual attention prediction model, and output a heat map corresponding to the user's visual focus distribution; For any candidate layout template in the template library corresponding to the product type, determine the first matching degree of the candidate layout template according to the position matching degree between each element area in the candidate layout template and the heat map and the preset weight value corresponding to each element area; Select the one with the highest first matching degree from the candidate layout templates as the target layout template.

4. The poster generation method according to claim 1, wherein The selection process of the target poster elements includes: Determine the original place of origin according to the product description information; For any type of element, input each candidate poster element belonging to the type of element and the product description information from the local element library corresponding to the original place of origin into the multimodal matching model in turn, and obtain the second matching degree between the candidate poster element and the product description information; Use the candidate poster element with the highest second matching degree as the target poster element corresponding to the type of element.

5. The poster generation method according to claim 1, wherein The generation of the target poster according to the target poster copywriting, the target poster elements, and the target layout template includes: Generate a preview poster according to the target poster copywriting, the target poster elements, and the target layout template; In response to the zoom-in operation on the preview poster, record the number of operations on the corresponding zoomed-in area; Determine the area to be optimized where the number of operations is greater than the set threshold; Enlarge the font or increase the resolution of the area to be optimized.

6. The poster generation method according to claim 1, characterized in that, After generating the target poster according to the target poster copywriting, the target poster elements, and the target layout template, it further includes: Obtain the screen resolution of the target display device; Adjust the resolution of the target poster according to the screen resolution; the higher the screen resolution, the higher the corresponding resolution of the target poster.

7. The poster generation method according to claim 1, wherein Before generating the target poster according to the target poster copywriting, the target poster elements, and the target layout template, it further includes: In response to the user's modification operation, modify the target poster copywriting, the target poster elements, and / or the target layout template.

8. A poster generating device, characterized in that, including: A data acquisition module for obtaining the product description information input by the user; A material determination module for determining the product type, target poster copywriting, and target poster elements according to the product description information; A template determination module, configured to select a corresponding target layout template according to the product type; A poster generation module, configured to generate a target poster according to the target poster copywriting, the target poster elements, and the target layout template.

9. A computer device, characterized in that, Comprising one or more processors, and a memory, wherein computer-readable instructions are stored in the memory, and when the computer-readable instructions are executed by the one or more processors, the steps of the poster generation method according to any one of claims 1-7 are executed.

10. A storage medium, characterized in that, Computer-readable instructions are stored in the storage medium, and when the computer-readable instructions are executed by one or more processors, one or more processors are caused to execute the steps of the poster generation method according to any one of claims 1-7.

Images