Intelligent poster generation method, device and equipment and computer storage medium

Abstract

The invention discloses an intelligent poster generation method, device and equipment and a computer storage medium, and relates to the technical field of data processing. The method comprises the steps of firstly obtaining a news text of a to-be-generated poster; inputting the news text into a classification model, performing feature extraction on the news text through the classification model to obtain a semantic vector of the news text, and calculating the similarity between the semantic vector and a preset semantic vector to obtain a target news category corresponding to the preset semantic vector with the highest similarity with the semantic vector; inputting the news text into a semantic extraction model corresponding to the target category, performing feature extraction of the target category on the news text through the semantic extraction model to obtain a semantic vector of the target category, and obtaining a semantic analysis result according to the semantic vector; and generating a corresponding poster according to the semantic analysis result and a preset template. According to the embodiment of the invention, the news text is automatically classified and extracted, and the poster is automatically generated according to the extraction result and the preset template, so that the poster generation efficiency is improved.

Description

Technical Field

[0001] This application belongs to the technical field of data processing, and particularly relates to a method, device, equipment and computer storage medium for intelligent generation of posters. Background Art

[0002] With the rapid development of digital media and Internet technologies, the news media industry has entered an era of information explosion, and the ways people obtain information have become more diverse. Among them, as an intuitive and vivid visual communication medium, posters have become an important means for news media to promote content.

[0003] Due to the timeliness of news, especially in the case of emergencies or hot topics, the media needs to quickly produce relevant posters to seize the opportunity of information dissemination. However, in the existing technology for poster production, posters are often designed manually. The producer needs to fully understand the news content and perform tasks such as material collection and layout design. Usually, it takes several hours or even days to complete a high-quality poster, and the efficiency of generating posters is low, making it difficult to meet the requirements of timeliness. Summary of the Invention

[0004] Embodiments of this application provide a method, device, equipment and computer storage medium for intelligent generation of posters to solve the problem of low efficiency in generating posters by existing methods.

[0005] In a first aspect, embodiments of this application provide a method for intelligent generation of posters, the method comprising:

[0006] Obtain the news text of the poster to be generated;

[0007] Input the news text into a classification model, extract features of the news text through the classification model to obtain a semantic vector of the news text, and calculate the similarity between the semantic vector and a preset semantic vector to obtain the target category of the news text, where the target category is the news category corresponding to the preset semantic vector with the highest similarity to the semantic vector;

[0008] Input the news text into a semantic extraction model corresponding to the target category, extract features of the target category from the news text through the semantic extraction model to obtain a semantic vector of the target category, and obtain a semantic analysis result based on the semantic vector;

[0009] Generate a corresponding poster according to the semantic analysis result and a preset template.

[0010] In a second aspect, an apparatus for intelligent generation of posters provided by embodiments of this application, the apparatus comprising:

[0011] An obtaining module, configured to obtain the news text of the poster to be generated;

[0012] A classification module, configured to input news text into a classification model, extract features of the news text through the classification model to obtain a semantic vector of the news text, and calculate the similarity between the semantic vector and a preset semantic vector to obtain the target category of the news text, where the target category is the news category corresponding to the preset semantic vector with the highest similarity to the semantic vector;

[0013] An extraction module, configured to input the news text into a semantic extraction model corresponding to the target category, extract features of the target category of the news text through the semantic extraction model to obtain a semantic vector of the target category, and obtain a semantic analysis result according to the semantic vector;

[0014] A generation module, configured to generate a corresponding poster according to the semantic analysis result and a preset template.

[0015] In a third aspect, an embodiment of the present application provides a terminal device, which includes: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the method for intelligent poster generation as in the first aspect is implemented.

[0016] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method for intelligent poster generation as in the first aspect is implemented.

[0017] In a fifth aspect, an embodiment of the present application provides a computer program product, and when the instructions in the computer program product are executed by a processor of an electronic device, the electronic device is enabled to execute the method for intelligent poster generation as in the first aspect.

[0018] An embodiment of the present application provides a method, device, device, and computer storage medium for intelligent poster generation. The method first obtains news text, inputs it into a classification model to extract features of the news text to obtain a semantic vector, and calculates the similarity between the semantic vector and a preset semantic vector to obtain the target category of the news text; it can quickly identify the category of the news text, without manual analysis one by one, saving labor and time costs, and reducing the possibility of human classification errors. Input the news text into a semantic extraction model corresponding to the target category to extract features to obtain a semantic analysis result. Using a dedicated semantic extraction model for different target categories can achieve automatic extraction of poster semantic elements and reduce the time for manual analysis and information extraction. Finally, a corresponding poster is generated according to the semantic analysis result and a preset template. The preset template provides the basic framework and layout rules of the poster. Combining with the semantic analysis result, the position of information in the poster can be quickly determined to obtain the corresponding poster, improving the efficiency of poster generation. Description of the Drawings

[0019] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required in the embodiments of the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

[0020] Figure 1 It is a schematic flowchart of the method for intelligent poster generation provided by the embodiments of the present application;

[0021] Figure 2 It is a schematic flowchart of an implementation manner for obtaining semantic vectors provided by the embodiments of the present application;

[0022] Figure 3 It is a schematic flowchart of an implementation manner for determining a preset template provided by the embodiments of the present application;

[0023] Figure 4 It is a schematic flowchart of an implementation manner for determining a preset semantic vector provided by the embodiments of the present application;

[0024] Figure 5 It is a schematic structural diagram of the device for intelligent poster generation provided by the embodiments of the present application;

[0025] Figure 6 It is a schematic structural diagram of the terminal device provided by the embodiments of the present application. Detailed Embodiments

[0026] The following will describe in detail the features and exemplary embodiments of various aspects of the present application. To make the purpose, technical solutions and advantages of the present application clearer, the following will further describe the present application in detail in combination with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than limiting the present application. For those skilled in the art, the present application can be implemented without some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by showing examples of the present application.

[0027] It should be noted that in this text, 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 terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising 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 "comprising..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the element.

[0028] When generating posters in the prior art, it mainly relies on the manual design of professional designers. Manual poster design involves multiple links from concept conception, material collection, layout design to finalization after modification. Each link requires a large amount of time and effort, and usually takes several hours or even days to complete a high-quality poster. Moreover, news has timeliness. Especially in the case of emergencies or hot topics, the media needs to quickly produce relevant posters to seize the opportunity of information dissemination. The traditional design process is difficult to respond in a timely manner, with low design efficiency, resulting in missing the best dissemination opportunity. To improve efficiency, some media organizations adopt fixed design templates. However, this templatized design method limits the play of creativity, resulting in posters lacking novelty. News content covers multiple fields such as politics, economy, culture, and sports, and the audience groups are diverse. The fixed design style cannot meet the aesthetic and information needs of different audiences, affecting the dissemination effect.

[0029] To solve the problems of the prior art, this application provides a method, device, equipment and computer storage medium for intelligent poster generation. The method first obtains a news text, inputs it into a classification model to extract features of the news text to obtain a semantic vector, and calculates the similarity between the semantic vector and a preset semantic vector to obtain the target category of the news text; it can quickly identify the category of the news text without manual analysis one by one, saving labor and time costs and reducing the possibility of human classification errors. Input the news text into the semantic extraction model corresponding to the target category to extract features to obtain a semantic analysis result. Using a dedicated semantic extraction model for different target categories can achieve automatic extraction of poster semantic elements and reduce the time for manual analysis and information extraction. Finally, generate a corresponding poster according to the semantic analysis result and a preset template. The preset template provides the basic framework and layout rules of the poster. Combining with the semantic analysis result, it can quickly determine the position of information in the poster to obtain the corresponding poster, improving the efficiency of poster generation.

[0030] The method for intelligent poster generation provided by the embodiments of the present application will be introduced below with reference to the accompanying drawings.

[0031] Figure 1 FIG. 4 shows a schematic flowchart of a method for intelligent poster generation provided by an embodiment of the present application. As Figure 1 shown, the method may include the following steps: S101 to S104.

[0032] S101, obtaining the news text of the poster to be generated.

[0033] Among them, the news text refers to the written material containing news events, information, descriptions, etc., and can be divided into multiple categories such as economy, sports, and military.

[0034] In some embodiments, the news text of the poster to be generated can obtain news text data through various methods such as user input, file upload, retrieval in a news database, and web crawler scraping.

[0035] The news text is the information basis for poster generation. By obtaining the news text, it provides the original data for generating news posters.

[0036] S102, inputting the news text into a classification model, extracting features of the news text through the classification model to obtain a semantic vector of the news text, and calculating the similarity between the semantic vector and a preset semantic vector to obtain the target category of the news text. The target category is the news category corresponding to the preset semantic vector with the highest similarity to the semantic vector.

[0037] Among them, the classification model is a model used to classify input data into predefined categories. The semantic vector is a high-dimensional vector representing the semantic features of the text, including the deep meaning and context information of the text. The preset semantic vector is predefined and calculated in advance, including multiple preset semantic vectors, and is a vector used to represent the semantic features of specific categories or concepts.

[0038] In some embodiments, the classification model is a pre-trained model for zero-shot learning. When a new news category needs to be added, there is no need to retrain the model with news texts of the new news category. Only the preset semantic vector of the new news category needs to be added, enabling the model to recognize and classify categories not seen in the training stage, improving the flexibility and adaptability of the model, and reducing the cost and time of model training.

[0039] In some embodiments, the classification model can be a model such as the Bidirectional Encoder Representations from Transformers (BERT), the A Robustly Optimized BERT Pretraining Approach (RoBERTa), and the A Lite BERT (ALBERT).

[0040] In some embodiments, the similarity between the semantic vector and the preset semantic vector can be calculated by methods such as dot product operation, cosine similarity calculation, Euclidean distance calculation, or Manhattan distance calculation.

[0041] Classifying news texts to clarify the subject or field to which the news texts belong, being able to understand the semantic features of the texts, and mapping them to the corresponding category space, so as to subsequently perform targeted semantic extraction and poster design on the news texts according to different categories.

[0042] S103. Input the news text into the semantic extraction model corresponding to the target category, extract the features of the target category from the news text through the semantic extraction model to obtain the semantic vector of the target category, and obtain the semantic analysis result according to the semantic vector.

[0043] Among them, the semantic extraction model is a model used to extract semantic features and key information from texts, and the semantic analysis result is a summary of the semantic features of the text determined according to the semantic vector.

[0044] In some embodiments, the semantic extraction model is a model trained with the training text of the target category, and each news category corresponds to a semantic extraction model respectively.

[0045] Inputting the news text into the semantic extraction model corresponding to the target category and optimizing the model for different categories can extract the features of the text under this category, generate the semantic vector of the target category, significantly reduce the extraction calculation amount of irrelevant features, and thus improve the accuracy and pertinence of the poster.

[0046] S104. Generate the corresponding poster according to the semantic analysis result and the preset template.

[0047] Among them, the preset template is a pre-designed poster template, including the position layout and design style of the poster elements, and the text content of the poster elements is included in the semantic analysis result.

[0048] In some embodiments, there are multiple preset templates, corresponding to different news categories or different semantic analysis results respectively.

[0049] By extracting key information and semantic features from the semantic analysis results and combining with a preset poster template, a poster that matches the news content can be generated, presenting the news text in a visual way. At the same time, the poster is automatically generated according to the semantic analysis results and the preset template, improving the efficiency of poster generation.

[0050] Obtain the news text and input it into a classification model to extract semantic vectors from the news text, and calculate the similarity between the semantic vectors and the preset semantic vectors to obtain the target category of the news text; it can quickly identify the category of the news text without manual analysis one by one, saving labor and time costs and reducing the possibility of human classification errors. Input the news text into the semantic extraction model corresponding to the target category to extract semantic analysis results. Using a dedicated semantic extraction model for different target categories can achieve the automatic extraction of poster semantic elements and reduce the time for manual analysis and information extraction. Finally, generate the corresponding poster according to the semantic analysis results and the preset template. The preset template provides the basic framework and layout rules of the poster. Combining with the semantic analysis results, the position of information in the poster can be quickly determined to obtain the corresponding poster, improving the efficiency of poster generation.

[0051] In some embodiments, before inputting the news text into the classification model, the method further includes:

[0052] Perform a word segmentation operation on the news text to obtain multiple lexical units, splitting the continuous news text into multiple independent lexical units, so that the text is transformed into basic language units that are easier to analyze and understand;

[0053] Remove the preset words from the lexical units to obtain target lexical units, where the preset words are predefined stop words, and stop words are words with little semantic contribution to the text, such as common function words and auxiliary words. Removing these words can reduce noise, thereby retaining words that make important contributions to semantics and improving the efficiency and accuracy of subsequent processing;

[0054] Perform part-of-speech analysis on the target lexical units to obtain part-of-speech analysis results, and determine the grammatical structure of the news text according to the part-of-speech analysis results, providing richer information for subsequent semantic analysis and classification;

[0055] Input the news text including the grammatical structure into the classification model. The news text including the grammatical structure provides richer semantic and grammatical information, which helps the classification model to understand and classify the text more accurately.

[0056] After tokenizing the news text, removing predefined vocabulary, and performing part-of-speech analysis, the text with grammatical structure information is input into a classification model. Utilizing the additional grammatical structure information, the classification model can more deeply understand the semantics and logical relationships of the news text, thereby more accurately extracting features and determining the news category. Compared with only inputting the original text, the performance of the classification model and the accuracy of classification are improved.

[0057] In some embodiments, the method consists of a content classification module, a knowledge extraction and semantic analysis module, an adaptive layout and dynamic content optimization module, and a poster generation module. The content classification module is used to classify the obtained news text; the knowledge extraction and semantic analysis module is used to process the classified news text to obtain the semantic analysis result of the news text; the adaptive layout and dynamic content optimization module is used to generate a preset template for the poster and adjust the preset template of the poster; the poster generation module is used to generate a poster according to the semantic analysis result and the poster preset template. Each functional module (content classification, knowledge extraction, layout optimization, poster generation) is developed independently, defining a unified interface specification to support the separate replacement and upgrade of the modules. This not only improves the maintainability and scalability of the system but also facilitates the integration of new functions and the rapid iteration of technologies, ensuring that the system can continuously adapt to the changing business requirements and technological developments.

[0058] In some embodiments, as Figure 2 shown, feature extraction of the news text by the classification model to obtain the semantic vector of the news text may include: S201 to S204.

[0059] S201, feature extraction of the news text by the classification model to obtain multiple word vectors of the news text.

[0060] S202, calculate the correlation value between each first word vector and each second word vector respectively, and assign a weight value to each second word vector corresponding to each first word vector according to the correlation value. The first word vector is any word vector, and the second word vector is any word vector other than the first word vector among the multiple word vectors.

[0061] Among them, the greater the correlation value between the first word vector and the second word vector, the greater the weight value assigned to the second word vector.

[0062] In some embodiments, assigning a weight value to each second word vector corresponding to each first word vector according to the correlation value may include:

[0063] Perform a normalization operation on the correlation value of the first word vector and the corresponding second word vector to obtain the normalized correlation value;

[0064] Use the normalized correlation value as the weight value and assign it to the corresponding second word vector.

[0065] S203. Determine the target semantic vector corresponding to each first word vector according to the second word vector and the weight value of the second word vector.

[0066] Among them, the target semantic vector is calculated by weighted summation according to all the second word vectors corresponding to the first word vector and the weight value of the second word vector. The target semantic vector synthesizes the semantic association information between all the second word vectors corresponding to the first word vector and the first word vector. Since it includes multiple semantic word vectors, multiple target semantic vectors are also correspondingly generated.

[0067] S204. Determine the semantic vector of the news text according to the target semantic vector.

[0068] In some embodiments, when determining the semantic vector of the news text according to the target semantic vector, the target semantic vectors can be aggregated into the semantic vector of the news text by taking the average value, weighted average value, self-attention mechanism or other aggregation methods. Among them, when taking the weighted average value, the weight of the target semantic vector can be determined by the frequency of the vocabulary corresponding to the first word vector corresponding to the target semantic vector appearing in the news text.

[0069] By converting the news text into word vectors and further considering the correlation relationship between the word vectors to generate semantic vectors, it is possible to capture the semantic information in the text more deeply and comprehensively. Compared with simply directly processing the text or only using a single word vector representation, this method can better understand the semantic meaning of the vocabulary in the context and their interactions with each other, thus providing a more accurate semantic basis for subsequent classification tasks.

[0070] In some embodiments, the semantic analysis result includes multiple poster elements, and the preset template includes the positions of the poster elements. Generating the corresponding poster according to the semantic analysis result and the preset template may include:

[0071] Determine the positions in the preset template corresponding to the poster elements of the semantic analysis result;

[0072] Insert the semantic analysis result into the corresponding positions in the preset template to obtain the corresponding poster.

[0073] By combining the preset template and semantic analysis technology, posters can be generated efficiently. The automated process reduces manual intervention and improves work efficiency. The preset template can be customized according to different requirements and scenarios to meet various poster design requirements.

[0074] In some embodiments, the preset template further includes the positions corresponding to the visual elements. Before obtaining the corresponding poster, the method may further include:

[0075] Determine the corresponding target visual elements according to the semantic analysis result;

[0076] Insert the target visual element into the corresponding position of the preset template.

[0077] The visual element is a component that constitutes the visual effect of the poster, and may include content such as pictures, icons, color schemes, etc. related to the key news information. For example, in a sports news poster, the visual element can be a picture of the game scene, a portrait icon of the athlete, a specific icon representing different sports events, etc.; in science and technology news, it may be a picture of a high-tech product, an icon representing technological innovation, etc. For the color scheme, science and technology news may use cool colors to reflect the sense of technology, and culture and art news may use warm colors to convey the sense of art.

[0078] Traditional poster design often requires designers to spend time retrieving relevant visual elements. By the system determining relevant visual elements according to the semantic analysis results and inserting them into the positions of the preset template, the time and workload of design decisions are greatly reduced. At the same time, it ensures that the visual elements selected for the poster are closely related to the news content. For example, if the semantic analysis result is sports event news, pictures of the game scene or athlete image icons can be selected as visual elements, making the poster more intuitively display the news core, improving the accuracy and effectiveness of information transmission, and enhancing the attractiveness to the audience.

[0079] In some embodiments, after inserting the semantic analysis result into the corresponding position of the preset template to obtain the corresponding poster, the method may further include:

[0080] Evaluate the initial layout of the poster elements in the poster according to the preset evaluation indicators to obtain the evaluation scores of each preset evaluation indicator;

[0081] Determine the comprehensive evaluation value of the poster according to the preset weights corresponding to the preset evaluation indicators and the evaluation scores;

[0082] Iteratively adjust the poster layout based on the initial layout of the poster and determine the comprehensive evaluation value until the comprehensive evaluation value reaches the preset threshold or the number of iterations reaches the preset number, and output the target layout;

[0083] Update the poster according to the target layout to obtain the updated poster.

[0084] Since different semantic analysis results have different poster elements and different contents, the layouts of the posters obtained by directly inserting the semantic analysis results into the corresponding positions of the preset template may be quite different. Therefore, it is necessary to iteratively optimize the generated posters and adjust the layouts of the posters to ensure that the visual effects and information transmission efficiency of the posters reach the optimal.

[0085] In some embodiments, the poster layout may include the positions of the poster elements, the styles of the poster elements, the proportions between the poster elements, white spaces, alignment methods, etc.

[0086] In some embodiments, an optimization algorithm can be used to adjust the poster layout, where the optimization algorithm can include genetic algorithms, simulated annealing algorithms, particle swarm algorithms, etc.

[0087] In some embodiments, such as Figure 3 shown, before generating the corresponding poster according to the semantic analysis result and the preset template, the method can further include: S301 to S303.

[0088] S301, obtain the poster background image.

[0089] The poster background image is the background image used for poster design, and can be obtained by means such as user upload, selection from a preset picture library, or web crawler scraping.

[0090] S302, perform target area recognition on the poster background image to obtain the target area.

[0091] The target area is the area in the background image suitable for placing poster elements, and can be a blank area or an area with less pattern. Through target area recognition, it can be ensured that the poster elements are placed in the most visually appropriate area, enhancing the overall layout and visual effect of the poster.

[0092] S303, set the positions of each poster element in the target area to obtain the preset template.

[0093] The poster elements are the various parts that make up the poster, such as elements like titles, subtitles, pictures, and icons.

[0094] To improve efficiency, some media organizations adopt fixed design templates. However, this templated design method limits the play of creativity and results in posters lacking novelty. News content covers multiple fields such as politics, economy, culture, and sports, and the audience is diverse. The fixed design style cannot meet the aesthetic and information needs of different audiences, affecting the communication effect.

[0095] Through target area recognition, it can be ensured that the poster elements are placed in the most visually appropriate area in the background image, which helps to enhance the overall layout and visual effect of the poster. By generating a preset template from the poster background image, the personalization and customization of the poster can be achieved. Different background images can adapt to different news themes and styles, dynamically adjusting the positions of the poster elements to make the poster more targeted and personalized.

[0096] In some embodiments, setting the positions of each poster element in the target area can include:

[0097] Divide the target area into multiple grids according to a preset ratio;

[0098] Set the poster elements at the preset grid positions.

[0099] By dividing the target area into multiple grids and placing the poster elements at the preset grid positions, the visual balance and symmetry of the elements can be ensured. Ensure that different poster elements can be placed in the visually focused areas, and at the same time can adapt to different poster sizes, ensuring that the preset templates of the posters can maintain good visual effects on different target areas.

[0100] In some embodiments, such as Figure 4 shown, before calculating the similarity between the semantic vector and the preset semantic vector to obtain the target category of the news text, the method may further include: S401 to S404.

[0101] S401, obtain the training news text corresponding to the target category;

[0102] Among them, the training news text is the description text for the new news category. For example, if the "environmental protection" category is to be added, a description text can be written: "Environmental protection news usually involves themes such as environmental protection, sustainable development, climate change, and green energy.

[0103] S402, perform feature extraction on the training news text to obtain a text feature vector;

[0104] S403, determine the attribute vector corresponding to the preset annotation information according to the preset annotation information of the training news text;

[0105] Among them, the attribute vector is used to represent the category or label information described in the text.

[0106] In some embodiments, the attribute vector is represented in the one-hot encoding manner. Each attribute corresponds to a dimension in the vector. If the text is for this data row, the corresponding dimension value is 1, otherwise it is 0. Through the attribute vector, the discrete attribute information of the text or data can be converted into a numerical representation that can be understood and processed by the machine learning model.

[0107] S404, determine the preset semantic vector of the category according to the text feature vector and the attribute vector.

[0108] By combining feature extraction and the attribute vector, the semantic features of the new category can be captured more accurately; by adding new preset semantic vectors, the text of the new category can be accurately classified without retraining the entire model and without a large amount of labeled data, quickly adapting to the new category, and improving the flexibility and adaptability of the model.

[0109] By using preset annotation information and feature extraction, the dependence on a large amount of annotated data can be reduced. This method utilizes the existing annotation information and feature vectors to quickly generate new preset semantic vectors, reducing the costs of data annotation and model training.

[0110] In some embodiments, after generating the corresponding poster according to the semantic analysis result and the preset template, the method may further include:

[0111] Obtaining an adjustment instruction of a user, where the adjustment instruction includes adjustment information of the user for the poster;

[0112] Adjusting the poster according to the adjustment information to obtain an adjusted target poster.

[0113] Different users may have different expectations and preferences for the display effect of the poster. By obtaining the adjustment instruction of the user and modifying the poster accordingly, the personalized ideas of the user can be fully considered, making the final poster more in line with the specific needs of the user and making the poster generation process more flexible.

[0114] In some embodiments, the adjustment instruction of the user is input through an interaction interface, allowing the user to preview the generated poster and make necessary fine-tuning. The user can adjust the text content, font style, color combination, element position, replace pictures, etc. of the poster by means of dragging or clicking interactions, etc., to customize the poster layout. The system reflects the user's modifications in real time, updates the layout and re-renders the preview image. After the user is satisfied with the generated poster, the user can download or share the poster with one click.

[0115] In some embodiments, after adjusting the user's poster to obtain a target poster, a new poster preset template can be generated according to the target poster and stored in the preset template library.

[0116] In some embodiments, after adjusting the poster according to the adjustment information to obtain an adjusted target poster, the method may further include:

[0117] Updating the preset template corresponding to the target poster according to the target poster.

[0118] The adjustments made by the user to the poster often reflect the personalized needs in actual applications or a better way to present specific news content. By feeding back these adjustments into the preset template, the template can be continuously evolved to better adapt to different types of news and the diverse needs of users, improving the applicability and practicality of the template in subsequent poster production. As the preset template is continuously updated according to the user's adjustments, the subsequent generated posters are more likely to be close to the user's expectations in the initial state, reducing the need for the user to repeatedly adjust. The newly generated posters can reach a higher quality standard faster, improving the production efficiency of news posters.

[0119] The target poster adjusted by the user can not only adjust the preset template corresponding to the target poster, but also update and adjust the news of the same category or posters with similar themes as the target poster. At the same time, it is also possible to summarize and analyze the adjustment information of the user for posters of the same category or similar themes, obtain common adjustment information, and update the preset template corresponding to posters of the same category or similar themes through the common adjustment information.

[0120] Figure 5 Fig. 500 shows a device for intelligent generation of posters provided by an embodiment of the present application, as Figure 5 shown, the device may include:

[0121] An acquisition module 501, configured to acquire news text of a poster to be generated;

[0122] A classification module 502, configured to input the news text into a classification model, extract features of the news text through the classification model to obtain a semantic vector of the news text, and calculate a similarity between the semantic vector and a preset semantic vector to obtain a target category of the news text, where the target category is a news category corresponding to the preset semantic vector with the highest similarity to the semantic vector;

[0123] An extraction module 503, configured to input the news text into a semantic extraction model corresponding to the target category, extract features of the target category of the news text through the semantic extraction model to obtain a semantic vector of the target category, and obtain a semantic analysis result according to the semantic vector;

[0124] A generation module 504, configured to generate a corresponding poster according to the semantic analysis result and a preset template.

[0125] In some embodiments, the device 500 for intelligent generation of posters may further include:

[0126] The extraction module 503 is further configured to extract features of the news text through the classification model to obtain multiple word vectors of the news text;

[0127] A calculation module, configured to calculate an association value between each first word vector and each second word vector respectively, and assign a weight value to each second word vector corresponding to each first word vector according to the association value, where the first word vector is any word vector, and the second word vector is any word vector other than the first word vector among the multiple word vectors;

[0128] A determination module, configured to determine a target semantic vector corresponding to each first word vector according to the second word vector and the weight value of the second word vector;

[0129] The determination module is further configured to determine a semantic vector of the news text according to the target semantic vector.

[0130] In some embodiments, the apparatus 500 for intelligent poster generation may further include:

[0131] A determination module, further configured to determine the positions in a preset template corresponding to the poster elements of the semantic analysis result;

[0132] An insertion module, configured to insert the semantic analysis result into the corresponding positions of the preset template to obtain a corresponding poster.

[0133] In some embodiments, the apparatus 500 for intelligent poster generation may further include:

[0134] An acquisition module 501, further configured to acquire a poster background image;

[0135] An identification module, configured to perform target area identification on the poster background image to obtain a target area;

[0136] A setting module, configured to set the positions of the respective poster elements in the target area to obtain a preset template.

[0137] In some embodiments, the acquisition module 501 is further configured to acquire training news texts corresponding to a target category;

[0138] An extraction module 503, further configured to perform feature extraction on the training news texts to obtain text feature vectors;

[0139] A determination module, further configured to determine an attribute vector corresponding to the preset annotation information according to the preset annotation information of the training news texts;

[0140] A determination module, further configured to determine a preset semantic vector of the category according to the text feature vectors and the attribute vectors.

[0141] In some embodiments, the apparatus 500 for intelligent poster generation may further include:

[0142] An acquisition module 501, further configured to acquire an adjustment instruction of a user, where the adjustment instruction includes adjustment information of the user for the poster;

[0143] An adjustment module, configured to adjust the poster according to the adjustment information to obtain an adjusted target poster.

[0144] Figure 5 Each module in the shown apparatus may implement Figure 1 each step in, and achieve the corresponding technical effects, which will not be elaborated herein for the sake of brevity.

[0145] Figure 6 FIG. shows a schematic hardware structure diagram of a terminal device provided in an embodiment of the present application.

[0146] The terminal device may include a processor 601 and a memory 602 storing computer program instructions.

[0147] Specifically, the above-mentioned processor 601 may include a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or may be configured as one or more integrated circuits for implementing the embodiments of the present application.

[0148] The memory 602 may include a mass storage for data or instructions. By way of example and not limitation, the memory 602 may include a Hard Disk Drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. In one example, the memory 602 may include removable or non-removable (or fixed) media, or the memory 602 is a non-volatile solid-state memory. The memory 602 may be internal or external to the integrated gateway disaster recovery device.

[0149] In one example, the memory 602 may include a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk storage media device, an optical storage media device, a flash memory device, an electrical, optical, or other physical / tangible memory storage device. Thus, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method for intelligent poster generation according to the present disclosure.

[0150] The processor 601 reads and executes the computer program instructions stored in the memory 602 to implement Figure 1 the method for intelligent poster generation in the illustrated embodiments.

[0151] In one example, the terminal device may further include a communication interface 603 and a bus 604. Among them, as Figure 6 shown, the processor 601, the memory 602, and the communication interface 603 are connected through the bus 604 to complete communication with each other.

[0152] The communication interface 603 is mainly used to implement communication between the various modules, devices, units, and / or devices in the embodiments of the present application.

[0153] The bus 604 includes hardware, software, or both, and couples components of the terminal device to each other. By way of example and not limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, the bus 604 may include one or more buses. Although embodiments of the present application describe and illustrate specific buses, the present application contemplates any suitable bus or interconnect.

[0154] In addition, in combination with the method for intelligent poster generation in the above embodiments, an embodiment of the present application can be implemented by providing a computer storage medium. Computer program instructions are stored on the computer storage medium; when the computer program instructions are executed by a processor, any one of the methods for intelligent poster generation in the above embodiments is implemented.

[0155] An embodiment of the present application also provides a computer program product, including a computer program, which when executed by a processor implements any one of the methods for intelligent poster generation in the above embodiments.

[0156] It should be clear that the present application is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated, and those skilled in the art can make various changes, modifications, and additions, or change the order between steps after understanding the spirit of the present application.

[0157] The functional blocks shown in the above-described structural block diagrams can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, and so on. When implemented in software, the elements of the present application are programs or text segments used to perform the required tasks. The program or text segment can be stored in a machine-readable medium or transmitted via a data signal carried in a carrier wave over a transmission medium or a communication link. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, read-only memory (ROM), flash memory, erasable read-only memory (EROM), floppy disks, compact disc read-only memory (CD-ROM), optical discs, hard disks, fiber optic media, radio frequency (RF) links, and so on. The text segment can be downloaded via a computer network such as the Internet, an intranet, and so on.

[0158] It should also be noted that in the exemplary embodiments mentioned in the present application, some methods or systems are described based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be executed in the order mentioned in the embodiments, can be different from the order in the embodiments, or several steps can be executed simultaneously.

[0159] Aspects of the present disclosure have been described above with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present disclosure. It should be understood that each block in the flowcharts and / or block diagrams, and the combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the functions / actions specified in one or more blocks of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It should also be understood that each block in the block diagram and / or flowchart, and the combinations of blocks in the block diagram and / or flowchart, can also be implemented by dedicated hardware performing the specified functions or actions, or by a combination of dedicated hardware and computer instructions.

[0160] The above are only specific embodiments of the present application. Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules, and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be elaborated herein. It should be understood that the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should be covered within the protection scope of the present application.

Claims

1. A method for intelligently generating posters, characterized in that: include: Get the news text of the poster to be generated; The news text is input into a classification model, and features of the news text are extracted by the classification model to obtain a semantic vector of the news text, and similarity between the semantic vector and a preset semantic vector is calculated to obtain a target category of the news text, where the target category is a news category corresponding to a preset semantic vector having the highest similarity to the semantic vector; Inputting the news text into the semantic extraction model corresponding to the target category, extracting features of the target category from the news text using the semantic extraction model to obtain a semantic vector of the target category, and obtaining a semantic analysis result based on the semantic vector; A corresponding poster is generated according to the semantic analysis result and a preset template.

2. The method for intelligently generating posters according to claim 1, characterized in that: The step of extracting features from the news text by the classification model to obtain a semantic vector of the news text includes: Extracting features of the news text by using the classification model to obtain multiple word vectors of the news text; Respectively calculating the association value between each first word vector and each second word vector, and assigning a weight value to each second word vector corresponding to each first word vector according to the association value, wherein the first word vector is any word vector, and the second word vector is any word vector among the multiple word vectors except the first word vector; Determine a target semantic vector corresponding to each first word vector according to the second word vector and the weight value of the second word vector; The semantic vector of the news text is determined according to the target semantic vector.

3. The method for intelligently generating posters according to claim 1, characterized in that: The semantic analysis result includes a plurality of poster elements, the preset template includes positions of the poster elements, and generating a corresponding poster according to the semantic analysis result and the preset template includes: Determine the corresponding position of the poster element of the semantic analysis result in the preset template; The semantic analysis result is inserted into the corresponding position of the preset template to obtain the corresponding poster.

4. The method for intelligently generating posters according to claim 1, characterized in that: Before generating a corresponding poster according to the semantic analysis result and a preset template, the method further includes: Get the poster background image; Performing target area recognition on the poster background image to obtain a target area; The position of each poster element is set in the target area to obtain a preset template.

5. The method for intelligently generating posters according to claim 1, characterized in that: Before calculating the similarity between the semantic vector and the preset semantic vector to obtain the target category of the news text, the method further includes: Obtain training news text corresponding to the target category; Performing feature extraction on the training news text to obtain a text feature vector; Determining an attribute vector corresponding to the preset annotation information according to the preset annotation information of the training news text; A preset semantic vector of the category is determined according to the text feature vector and the attribute vector.

6. The method for intelligently generating posters according to claim 1, characterized in that: After generating a corresponding poster according to the semantic analysis result and the preset template, the method further includes: Obtaining a user's adjustment instruction, wherein the adjustment instruction includes the user's adjustment information for the poster; The poster is adjusted according to the adjustment information to obtain an adjusted target poster.

7. A device for intelligently generating posters, characterized in that: include: An acquisition module, used to acquire the news text of the poster to be generated; A classification module, used for inputting the news text into a classification model, performing feature extraction on the news text through the classification model to obtain a semantic vector of the news text, and calculating the similarity between the semantic vector and a preset semantic vector to obtain a target category of the news text, wherein the target category is a news category corresponding to a preset semantic vector having the highest similarity to the semantic vector; An extraction module, used for inputting the news text into a semantic extraction model corresponding to the target category, extracting features of the target category from the news text through the semantic extraction model to obtain a semantic vector of the target category, and obtaining a semantic analysis result based on the semantic vector; A generation module is used to generate a corresponding poster according to the semantic analysis result and a preset template.

8. A terminal device, characterized in that: The device comprises: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the method for intelligently generating posters as described in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the method for intelligently generating posters as described in any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the method for intelligently generating posters as described in any one of claims 1 to 6.

Images