Pet insurance propaganda map drawing method and system, electronic equipment and storage medium

Abstract

The invention provides a pet insurance propaganda map making method and system, electronic equipment and a storage medium. The method comprises the steps of obtaining demographic data of a target customer and corresponding pet information; generating a propaganda map through a text-to-image generation model according to the text description in combination with the obtained demographic data of the target customer and the corresponding pet information; extracting a feature vector of the propaganda map; and performing fine tuning on the feature vector of the generated propaganda map by using a fine tuning model to obtain a fine-tuned propaganda map. According to the method, accurate propaganda map design can be performed according to the characteristics of the target customer, so that the propaganda effect and the conversion rate are improved, the method is also suitable for various pet insurance companies or related enterprises, can be used for making various pet insurance propaganda maps such as product introduction maps and preferential activity maps, can also be applied to other fields, and has wide application prospects. And more personalized and accurate marketing services are provided for different industries, such as medical care, financial management and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of image generation and processing, and in particular to a method, system, electronic device and storage medium for making a pet insurance promotional picture. Background Art

[0002] Pet insurance is an emerging insurance business with broad market prospects. However, since the target audience of pet insurance is relatively small, effective promotional means are needed to attract potential customers. Currently, most pet insurance promotional images on the market use traditional design methods, lacking innovation and personalization, and are difficult to attract the attention and interest of target customers.

[0003] Promotional images often need to be provided periodically and frequently, which requires the selected images to be high-frequency and attractive. However, the actual situation shows that it takes a lot of time and energy to produce a high-quality image, and only one can be produced at a time, and the production process also takes a long time. Summary of the invention

[0004] The present invention provides a method, system, electronic device and storage medium for making pet insurance promotional pictures, which realize the generation from text to image, thereby generating the required high-quality pictures faster and more efficiently.

[0005] The technical solution of the present invention is:

[0006] In a first aspect, the present application provides a method for producing a pet insurance promotional image, which is characterized by comprising:

[0007] Obtain demographic data of target customers and corresponding pet information;

[0008] Generate promotional images based on the text description combined with the acquired demographic data of target customers and the corresponding pet information through a text-to-image generation model;

[0009] Extract feature vectors of promotional images;

[0010] The fine-tuning model is used to fine-tune the feature vector of the generated promotional image to obtain a fine-tuned promotional image.

[0011] Furthermore, the demographic data of the target customers includes one or any two or more of the target customers' gender, age, income, and hobby information.

[0012] Furthermore, the pet information includes one, any two or more of the pet type, age, disease history, and vaccination history.

[0013] Furthermore, the text-to-image generation model is loaded using the Stable Diffusion large model.

[0014] Furthermore, the fine-tuning model is fine-tuned using LoRA.

[0015] Furthermore, the method of fine-tuning the generated promotional image using the fine-tuning model includes: in the attention layer of the text-to-image generation model, applying LoRA parameters to adjust the feature vector processed by the model.

[0016] In a second aspect, the present application also provides a pet insurance promotional picture production system, which is characterized by comprising:

[0017] Data and information acquisition module, used to obtain demographic data of target customers and corresponding pet information;

[0018] A text-to-image generation module is used to generate a promotional image based on the text description combined with the acquired demographic data of the target customers and the corresponding pet information through a text-to-image generation model;

[0019] Feature vector extraction module, used to extract feature vectors of promotional images;

[0020] The fine-tuning module uses the fine-tuning model to fine-tune the feature vector of the generated promotional image to obtain the fine-tuned promotional image.

[0021] Furthermore, the text-to-image generation module adopts a Stable Diffusion large model, and the fine-tuning module adopts a LoRA model for fine-tuning.

[0022] In a third aspect, the present application also provides a computer storage medium, characterized in that: the computer storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the above-mentioned method steps.

[0023] In a fourth aspect, the present application further provides an electronic device, characterized in that it comprises a processor and a memory, wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the above-mentioned method steps.

[0024] The beneficial effects brought about by the technical solutions provided by some embodiments of the present application include at least: the present application can accurately design promotional images according to the characteristics of target customers, thereby improving the promotional effect and conversion rate. In addition.

[0025] The present invention is also applicable to various pet insurance companies or related enterprises, and can be used to produce various pet insurance promotional pictures, such as product introduction pictures, promotional activity pictures, etc. At the same time, the present application can also be applied to other fields, such as health care, financial management, etc., to provide more personalized and accurate marketing services for different industries. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0027] Figure 1 Flow chart of the method of the embodiment of the invention

[0028] Figure 2 is a system block diagram of an embodiment of the present invention;

[0029] Figure 3 It is a schematic diagram of the structure of the electronic device of the present invention.

[0030] The specific implementation modes of the present invention are further described in detail below with reference to the accompanying drawings. DETAILED DESCRIPTION

[0031] Combine the following Figure 1 As shown, the technical solution provided by the present invention is described in more detail. The description of the exemplary implementation force is merely illustrative and is by no means intended to limit the present disclosure and its application or use. The present disclosure can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to make the present disclosure thorough and complete and to fully express the scope of the present disclosure to those skilled in the art. It should be noted that unless otherwise specifically stated, the relative arrangement of the parts and steps, the composition of the materials described in these embodiments. Numerical expressions and numerical values ​​should be interpreted as being merely exemplary and not as limitations.

[0032] The serial numbers assigned to the components in the present disclosure, such as "first", "second", etc., are only used to distinguish the objects described and do not have any order or technical meaning. The "connection" and "coupling" mentioned in the present invention, unless otherwise specified, include direct and indirect. In the description of the present invention, it should be understood that the directions or positional relationships indicated by the terms "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", etc. are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to the invention.

[0033] In the present disclosure, when a specific device is described as being located between a first device and a second device, there may or may not be an intermediate device between the specific device and the first device or the second device. When a specific device is described as being connected to other devices, the specific device may be directly connected to the other devices without an intermediate device, or may not be directly connected to the other devices but have an intermediate device.

[0034] All terms (including technical terms or scientific terms) used in the present disclosure have the same meanings as those understood by ordinary technicians in the field to which the present disclosure belongs, unless otherwise specifically defined. It should also be understood that terms defined in general dictionaries should be interpreted as having meanings consistent with their meanings in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined herein.

[0035] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0036] The dimensions of the various parts shown in the drawings are not drawn according to the actual proportional relationship. In the drawings, the same figure marks are attached to common structural elements or structural elements of the same type, and their repeated descriptions are appropriately omitted.

[0037] A promotional graphic is an image used to promote and advertise a product, service, event, or idea. It usually includes attractive visual elements such as pictures, illustrations, icons, and text to attract the audience's attention and convey relevant information. Promotional graphics can be used in various media platforms such as print ads, posters, flyers, social media, websites, etc.

[0038] When creating promotional graphics, you need to consider the following elements:

[0039] Target Audience: Understand your audience so you can provide them with relevant and engaging information.

[0040] Messaging: Make sure your graphic conveys a clear, concise message so that your audience can understand at a glance what you are promoting.

[0041] Visual appeal: Use high-quality images, illustrations, and icons, as well as appealing colors and fonts to make your promotional graphics stand out visually.

[0042] Layout and typesetting: Arrange the layout of the promotional image reasonably to ensure that the information is clearly structured and easy to read.

[0043] Brand consistency: Make sure your promotional graphics are consistent with your brand image, including colors, fonts, and design style.

[0044] Call to Action: Include a clear call to action in your promotional graphic that tells viewers what they should do, such as purchase a product, attend an event, or learn more.

[0045] Adaptability: Adjust the design of the promotional image according to different media platforms and size requirements to ensure that it can be used in various scenarios.

[0046] With the above elements, you can create an effective promotional graphic to help your product, service or event gain more attention and interest.

[0047] In combination with the above-mentioned general requirements for promotional images, this application provides a method for producing a pet insurance promotional image, including:

[0048] Obtain demographic data of target customers and corresponding pet information;

[0049] Generate promotional images based on the text description combined with the acquired demographic data of target customers and the corresponding pet information through a text-to-image generation model;

[0050] Extract feature vectors of promotional images;

[0051] The fine-tuning model is used to fine-tune the feature vector of the generated promotional image to obtain a fine-tuned promotional image.

[0052] In some embodiments, the demographic data of the target customer includes, but is not limited to, one or any two or more of the target customer's gender, age, income, occupation, and hobby information.

[0053] Specifically, the demographic data of target customers is a key factor in market analysis and marketing strategy, which helps us understand and define the characteristics of its potential customer base. This data can include but is not limited to the following aspects:

[0054] Gender: Knowing whether your target customers are male, female, or gender neutral helps you tailor your products or services to better meet the needs and preferences of a specific gender.

[0055] Age: Customers of different ages may have different needs and reactions to products or services. Age information helps companies identify the life cycle stage that their products are suitable for.

[0056] Income: The economic status of customers affects their purchasing power and consumption habits. Understanding the income level of target customers can help companies formulate pricing and promotion strategies.

[0057] Occupation: The customer's occupation will also affect their purchasing power and consumption habits. For example, some people in high-income occupations may be more willing to buy high-end products or services, while some people in low-income occupations may be more inclined to look for products or services with high cost performance. Understanding the target customers' occupations can help companies better understand their needs and preferences, thereby developing more effective promotional content and strategies.

[0058] Hobbies: Customers' interests and hobbies can reveal their lifestyles and consumption tendencies, which is particularly important for product design, advertising, and market segmentation.

[0059] Combining one or more of the above demographic data can more accurately position the market and generate products that meet the needs and expectations of target customers, thereby increasing market competitiveness and customer satisfaction.

[0060] In some embodiments, the pet information includes, but is not limited to, one, any two or more of the pet type, age, disease history, and vaccination history.

[0061] Pet information is various data and details about your pet. Here are some common categories of pet information and what they might contain:

[0062] Pet type: This includes the type of pet (such as dog, cat, bird, fish, reptile, etc.) and breed (such as Labrador, Siamese cat, goldfish, red-eared turtle, etc.). The breed of the pet is also important information, which is directly related to what kind of pet the image is generated.

[0063] Pet Name: Every pet should have a unique name to facilitate identification and communication.

[0064] Gender: Your pet's gender may affect its appearance, behavior, and health.

[0065] Weight: A pet's weight can reflect its size and health. Being overweight or underweight may be a sign of health problems.

[0066] Coat color: For some pets, such as dogs and cats, coat color is also an important characteristic.

[0067] Personality: A pet's personality can show how cute they are and also affect people's perception of them.

[0068] Eating habits: A pet's eating habits can affect its health, so it is important to understand what the pet likes to eat and what it doesn't like to eat, which indirectly reflects its appearance and other characteristics.

[0069] Training status: For some pets, such as dogs, understanding their training status can help owners better manage their behavior.

[0070] Special needs: Some pets may have special care needs, such as requiring regular bathing, or requiring a special diet.

[0071] Microchip Information: If the pet is microchipped, this information should also be recorded.

[0072] Breeding records: For pets used for breeding, their breeding records are also important information.

[0073] Age: The age of the pet, usually in years, but sometimes in months or days (especially for young pets).

[0074] Medical History: Past and current illnesses or health problems your pet has, including diagnosis, treatment course, and outcomes.

[0075] Vaccination history: All vaccines and preventive treatments received by the pet, including vaccination date, vaccine type (such as rabies, triple vaccine, etc.) and manufacturer.

[0076] Diet and Nutrition: Your pet’s daily diet, including food type (dry food, wet food, raw food, etc.), food brand, feeding frequency, and amount.

[0077] Behavior and Training: The pet's behavioral traits, training level, known behavioral problems or special habits.

[0078] Living environment: The living environment of the pet, including indoor, outdoor, cage size, temperature control, etc.

[0079] Daily care: Daily care needs of pets, such as combing hair, cutting nails, cleaning ears and teeth, etc.

[0080] Medical records: Your pet’s medical records, including regular checkups, lab test results, x-rays or ultrasounds, etc.

[0081] Medications and Treatments: Medications, supplements, and treatments your pet is currently taking or has taken in the past.

[0082] Microchipping and Identification: Whether the pet is microchipped, and any other forms of identification (such as collar tags, tattoos, etc.).

[0083] Emergency Contact: Information about someone who can be contacted in an emergency, including the pet owner, family member, or emergency contact.

[0084] Insurance information: If the pet is insured, the insurance company, policy number and coverage.

[0085] Special Needs: Any special needs your pet may have, such as disabilities, chronic disease management, special dietary requirements, etc.

[0086] The above information can directly or indirectly reflect the external characteristics of the pet.

[0087] The above information can be obtained and saved in the following ways:

[0088] Design a questionnaire: Create a questionnaire that includes questions about demographic characteristics (such as age, gender, education level, income level, occupation, etc.) and pet information (such as pet type, number, age, health status, etc.).

[0089] Choose a data collection method: You can collect data through online surveys, telephone interviews, face-to-face interviews, or mail surveys. Each method has its advantages and disadvantages, and the most appropriate method should be selected based on the characteristics of the target customer group and the resources available.

[0090] Determine the sample: Select a representative sample to ensure that the sample reflects the characteristics of the target customer group. Random sampling, stratified sampling, or other sampling techniques can be used.

[0091] Collect data: Administer questionnaires or other data collection methods, ensuring data quality and accuracy.

[0092] Data analysis: Organize and analyze the collected data. Statistical software can be used to help process data and generate reports.

[0093] Comply with privacy regulations: When collecting and processing personal data, relevant data protection and privacy regulations must be followed to ensure the information security and privacy rights of participants.

[0094] Report and apply results: Organize the analysis results into reports.

[0095] The above factors can be used as considerations for generating promotional images. According to the different uses of promotional images, corresponding factors can be selected to generate corresponding promotional effect images.

[0096] It should be noted that the above target customer information is the specific group of people that the promotional image produced this time is intended to promote. According to the text description, the text-to-image generation model selects information matching the text description from the target customer information and pet information to generate the promotional image.

[0097] In some embodiments, to generate a promotional image, it is necessary to select target customer information and pet information based on the text description. When creating a promotional image through a text-to-image generation model, it is important to note that:

[0098] The text-to-image generation model captures the content, style, and focus of the text description, and determines the key information mentioned in the text, such as pet types, characteristics, and advantages of services or products. Based on the text description, the characteristics of the target customer group are determined, such as age, gender, interests, lifestyle, etc. Then, pet information is collected. If specific pets are mentioned in the text description, pictures and related information of these pets are collected. If there are no specific pets, some general and attractive pet pictures can be selected.

[0099] The text-to-image generation model also designs the layout. According to the text description and the needs of the target customers, the layout of the promotional image is designed, such as effectively combining pet pictures and text information to attract the attention of the target customers. In addition, colors and fonts are selected. According to the text description and the preferences of the target customers, appropriate color schemes and fonts can be selected. The colors and fonts should be able to convey the emotions and information in the text description. Then, based on the generated image, necessary adjustments and improvements are made to ensure that all elements in the image can accurately reflect the content of the text description and are attractive to the target customers. This process may require multiple iterations to ensure that the final promotional image can effectively convey the information in the text description and attract target customers.

[0100] In order to generate promotional images, it is necessary to extract key information from the text description, including the characteristics of the target customers, the types and characteristics of pets, etc. For example, it is necessary to:

[0101] Analyze the text description to determine the target customer's age, gender, occupation, hobbies, etc. For example, if the text description mentions "young mother", then we can infer that the target customer is a young woman who may be interested in childcare, home decoration, etc.

[0102] Analyze the text description to determine the type and characteristics of the pet. For example, if the text description mentions "cute puppy", then it can be inferred that the pet is a dog and has cute characteristics.

[0103] Choose appropriate image elements based on the target customers and pet information. For example, for a young mother, we can choose warm family scenes, baby supplies and other elements; for a cute puppy, we can choose lively backgrounds, dog toys and other elements.

[0104] Using the text-to-image generation model, the selected elements are combined into a promotional image. In this process, the layout, color, size, etc. of the elements can be adjusted as needed to make the promotional image more attractive.

[0105] To achieve the above-mentioned purpose and method, in some embodiments, the text-to-image generation model can be loaded by using, but not limited to, the Stable Diffusion model. The Stable Diffusion model is a generative model based on deep learning, which is used to generate high-quality images. By learning a large amount of training data, it can generate images with high realism and diversity. The core idea of ​​the Stable Diffusion model is to perform a diffusion process in the latent space to generate new samples.

[0106] The main advantages of the Stable Diffusion model are its stability and controllability. Compared with other generative models, such as generative adversarial networks (GANs), the Stable Diffusion model is more stable during training and is not prone to problems such as mode collapse. In addition, the Stable Diffusion model can also control the style and content of the generated image by adjusting the parameters in the latent space, thereby achieving more diverse image generation. A typical application of the Stable Diffusion model is image generation. By sampling different points in the latent space, images with different styles and contents can be generated. In addition, the Stable Diffusion model can also be used for tasks such as image restoration and image super-resolution. This embodiment uses the open source Stable Diffusion 1.5 version large model on the Internet to generate corresponding feature vectors.

[0107] The main components of the Stable Diffusion model include:

[0108] Encoder: Converts the input image into a low-dimensional latent vector representation.

[0109] Decoder: Converts the latent vector representation back to a high-dimensional image.

[0110] Diffusion Process: Simulates the image generation process by gradually adding noise and denoising.

[0111] In the training phase, the Stable Diffusion large model learns the distribution of data by minimizing the reconstruction error. In the generation phase, given a random latent vector, the model can generate an image corresponding to it. The advantages of the Stable Diffusion large model include:

[0112] The generated images are of high quality and can capture the details and complex structures in the training data. The model is stable and is not prone to overfitting or mode collapse. The diversity and style of the generated images can be controlled by adjusting the distribution of the latent vectors.

[0113] In some embodiments, the promotional image generated by the text-to-image generation model needs to be fine-tuned, and the feature vector of the promotional image needs to be extracted during fine-tuning; the feature vector of the generated promotional image is fine-tuned using the fine-tuning model, and finally a fine-tuned promotional image is obtained. The promotional image obtained in this way is the one with the best effect.

[0114] In the current marketing and advertising fields, customized and personalized promotional images are increasingly valued. Using fine-tuning models to optimize promotional images can better attract the attention of the target group and improve advertising effectiveness. The fine-tuning model is a machine learning-based technology that adapts the trained model parameters to new specific tasks or data sets by slightly adjusting them. This embodiment may involve adjusting the style, color, layout, etc. of the image to meet specific visual requirements or brand standards. The fine-tuning model allows innovation in details while maintaining overall design consistency. For example, unique color matching or layout can be achieved through fine-tuning, making it stand out from many promotional images. Innovative designs are more likely to attract consumers' attention, improve brand recognition, and enhance the effect of promotional images.

[0115] In some embodiments, the fine-tuning model is fine-tuned using LoRA.

[0116] In summary, LoRA is an effective fine-tuning method that can improve fine-tuning performance while reducing the number of model parameters. This is very helpful for scenarios with large-scale language models and limited computing resources.

[0117] In some embodiments, the method of fine-tuning the generated promotional image using a fine-tuning model includes: in the attention layer of the text-to-image generation model, applying LoRA parameters to adjust the feature vector processed by the model.

[0118] LoRA (Low-Rank Adaptation of Large Language Models) is a lightweight fine-tuning method that uses a small number of images to train a small model, which is then combined with the base model and affects the model results by inserting layers. Stable Diffusion requires at least 24G of video memory for comprehensive fine-tuning. But with LoRA, it can be done on a single 12GB GPU. Freeze the pre-trained model parameters and insert trainable layers in each Transformer block. There is no need to fully adjust all the parameters of the model, and the model size is small.

[0119] Fine-tuning a model is a method of making small adjustments to a pre-trained model to better suit a specific task. In text-to-image generation models, fine-tuning can be achieved by adjusting the attention layer. LoRA (Local Re-parametrization) is a technique for fine-tuning Transformer models that adjusts the model by adding a learnable low-rank matrix to the attention layer.

[0120] Specifically, you can follow the steps below to fine-tune the text-to-image generation model using LoRA:

[0121] First, choose a pre-trained text-to-image generation model, such as GAN, VAE, or diffusion model.

[0122] In the attention layer of the model, the original attention weight matrix is ​​decomposed into the product of two low-rank matrices. These two low-rank matrices can be learnable parameters or fixed random matrices.

[0123] During the training process, these two low-rank matrices are multiplied with the original attention weight matrix to obtain a new attention weight matrix. In this way, the model can fine-tune the attention layer by learning these two low-rank matrices while maintaining the original structure.

[0124] Use the new promotional image data to train the model. During the training process, the model will learn how to adjust the attention weight according to the input text information, so as to generate promotional images that better meet the requirements.

[0125] During the testing phase, the trained model is used to generate corresponding promotional images for the given text.

[0126] In this way, the text-to-image generation model can be fine-tuned without changing the model structure, thereby improving the quality of the generated promotional images.

[0127] As an example, the LoRA module usually consists of two low-rank matrices, A and B, which have a small size and can effectively reduce the number of parameters and reduce the risk of overfitting. This module is inserted into the normal forward propagation path of the selected layer. It can be between the input and output of the layer or before and after the activation function of the layer. For example, the pre-trained layer has an output feature vector of Z. The LoRA module can be connected like this:

[0128] Input feature vector X

[0129] Original layer weight W

[0130] Original layer output Z = X*W

[0131] LoRA module A*B

[0132] LoRA adjusted output Z'=Z+A*B

[0133] The next layer or activation function.

[0134] In summary, the feature vector X produces the output Z through the original layer weights W. Then, the LoRA module adjusts Z through its low-rank matrices A and B to generate Z'.

[0135] This adjustment is lightweight because it only involves the product of two small matrices and can capture important interactions between the input and output eigenvectors.

[0136] During fine-tuning, the weights W of the original layer are usually frozen, while the parameters A and B of the LoRA module are trained to adapt to the new data. In this way, the model can quickly adapt to new tasks without requiring a lot of computing resources.

[0137] The above embodiment uses Stable Diffusion + LoRA fine-tuning, which has faster training speed and calculation, and a smaller training model. Since the original model has been frozen and a new layer has been injected for training, the weights of the new layer can be saved as a single file. Excellent results can be obtained when the number of trained weights is several orders of magnitude less than the size of the original model. The accuracy of the model on specific tasks can be improved. The effect of the idea can be quickly generated, and these results can provide creators with new creative inspiration, open up new design ideas and directions, and better achieve their own design goals.

[0138] In summary, Stable Diffusion is a data augmentation method for text generation that creates new training samples by adding noise to the training data. This method can improve the generalization ability of the model and make it perform better when faced with unseen inputs. The main advantage of Stable Diffusion is that it can be applied to any type of text data and can be easily combined with other natural language processing techniques. LoRA fine-tuning is a transfer learning technique that fine-tunes a pre-trained language model by adding a small number of trainable parameters. These parameters are usually low-rank approximations, which means that their number is much smaller than the number of parameters of the original model. The main advantage of LoRA fine-tuning is that it can improve the performance of the model without significantly increasing the computational cost. Combining these two techniques, the training data can be augmented using Stable Diffusion first, and then the augmented data can be trained using LoRA fine-tuning. This method may further improve the generalization ability and performance of the model.

[0139] Therefore, the above embodiment can also accurately design promotional pictures according to the characteristics of target customers, thereby improving the promotion effect and conversion rate. In addition, it is also possible to continuously optimize and adjust the promotion strategy to adapt to market changes and user needs. It can also be applied to various pet insurance companies or related companies, and can be used to produce various pet insurance promotional pictures, such as product introduction pictures, promotional activity pictures, etc. At the same time, it can also be applied to other fields, such as health care, financial management, etc., to provide more personalized and accurate marketing services for different industries.

[0140] The pet insurance promotional picture making system of this embodiment is an online tool specially designed for pet insurance companies or related institutions to create attractive and professional promotional materials.

[0141] In addition, the system may also include a preview module, an export and sharing module, a cloud storage module, a tutorial and help module, a permission management module, a data analysis module, and the like. The preview module is used to preview the final promotional image effect after the design is completed to ensure that the user's needs are met. The export and sharing module is used to allow users to export the promotional image into common image formats (such as JPG, PNG, etc.) or directly share it on social media platforms after the design is completed. The cloud storage module allows users to save the design in the cloud for future access and re-editing. The tutorial and help module is used to provide usage tutorials and online help so that users can get started quickly. The permission management module is used for corporate users. The system may also include permission management functions to control the access and editing permissions of different users. The data analysis module is used to provide data analysis functions to help users understand key indicators such as the click-through rate and conversion rate of the promotional image.

[0142] Through such a system, pet insurance companies can quickly and efficiently produce professional-level promotional materials to enhance their brand image and attract more customers.

[0143] In a second aspect, the present application also provides a pet insurance promotional image production system, which can be an online platform or software designed to help insurance companies, advertising agencies or independent marketers quickly and efficiently create attractive pet insurance promotional materials. It includes:

[0144] Data and information acquisition module, used to obtain demographic data of target customers and corresponding pet information;

[0145] A text-to-image generation module is used to generate a promotional image based on the text description combined with the acquired demographic data of the target customers and the corresponding pet information through a text-to-image generation model;

[0146] Feature vector extraction module, used to extract feature vectors of promotional images;

[0147] The fine-tuning module uses the fine-tuning model to fine-tune the feature vector of the generated promotional image to obtain the fine-tuned promotional image.

[0148] In some embodiments, the data and information acquisition module, the main task of which is to collect demographic data of target customers and information about their pets. This data may include demographic information such as age, gender, occupation, income level, and information such as the type, age, and health status of the pet. This module may include acquiring data from databases, files, networks, and other places. In some embodiments, this module also needs to process various data formats and may require some preprocessing, such as cleaning data, processing missing values, etc.

[0149] In some embodiments, the text-to-image generation module: This module generates a promotional image through a text-to-image generation model based on the input text description and the demographic data and pet information of the target customer obtained from the data and information acquisition module. For example, if the text description is "a cute puppy running on the grass", then the generated promotional image may be a picture of a puppy running on the grass. The text-to-image generation model will analyze the text "a cute puppy running on the grass" and then create a picture based on this description. This picture will contain the following elements:

[0150] A puppy: The model needs to recognize that the subject is a dog and understand the adjective "cute" so that the cute features of the puppy can be reflected when generating images, such as through details such as the puppy's expression, posture, or fur color.

[0151] Grass: The background should be grass, which means there will be green plants as ground cover in the image and possibly grass of varying lengths and colors.

[0152] Running: The action description indicates that the puppy is moving, so the puppy in the image is not static, but appears to be running, with dynamic effects such as spread limbs and fluttering ears.

[0153] Based on this information, the text-to-image generation model synthesizes an image that reflects the scene described in the text as accurately as possible. The model is able to generate high-quality, realistic images, making the conversion from text to image more natural and accurate.

[0154] In some embodiments, the feature vector extraction module extracts a feature vector of the generated promotional image. A feature vector is a mathematical representation that can capture key features of an image, such as color, shape, texture, etc.

[0155] The choice of the above methods depends on the specific application scenario and requirements. For example, if you need to focus on the color information of the image, you can choose the color histogram; if you need to focus on the shape information of the image, you can choose the HOG feature; if you need to focus on the global information of the image, you can choose the GIST feature; if you need to focus on the local information of the image, you can choose the SIFT feature; if you need to automatically learn the features of the image, you can choose the CNN feature.

[0156] In some embodiments, a fine-tuning module: This module uses a fine-tuning model to fine-tune the feature vector of the generated promotional image to obtain a promotional image that better meets the needs of target customers. The fine-tuning model may adjust the color, shape, texture and other features of the promotional image based on demographic information such as the age, gender, occupation, income level and other demographic information of the target customers, as well as their pet information.

[0157] In some embodiments, the text-to-image generation module uses a Stable Diffusion large model, the feature vector extraction module can be implemented by the Stable Diffusion large model, or by a separate feature vector extraction module, and the fine-tuning module uses a LoRA model for fine-tuning. In general, LoRA (Local Re-parametrization of Transformer Architectures) is a method for fine-tuning large language models, which reduces the number of model parameters by introducing local re-parameterization in the Transformer architecture, thereby improving the efficiency and performance of fine-tuning.

[0158] The steps for fine-tuning using LoRA are as follows:

[0159] First, choose a pre-trained large language model, such as BERT, GPT, etc.

[0160] Then, the LoRA layer is inserted into the Transformer layer of the original model. The LoRA layer includes a low-rank decomposition matrix and a trainable vector. In this way, the number of parameters of the original model will be greatly reduced, thereby reducing the computational complexity of fine-tuning.

[0161] Next, the model is fine-tuned using the dataset of the target task. During fine-tuning, only the parameters of the LoRA layer are updated, while the parameters of the original model remain unchanged. This ensures that the performance of the model on the target task is improved while maintaining the knowledge of the original model.

[0162] Finally, the performance of the fine-tuned model on the target task is evaluated. If the performance meets the requirements, the fine-tuned model can be deployed to the actual application.

[0163] The above-mentioned data and information acquisition module for acquiring the demographic data of target customers and the corresponding pet information, the text-to-image generation module for generating a promotional image through a text-to-image generation model based on the text description combined with the acquired demographic data of target customers and the corresponding pet information; the feature vector extraction module for extracting the feature vector of the promotional image and the fine-tuning module for fine-tuning the feature vector of the generated promotional image using the fine-tuning model to obtain the fine-tuned promotional image can be deployed on the same hardware receipt, for example, on the same computer.

[0164] The method for generating a promotional image of the above system includes obtaining demographic data of target customers and corresponding pet information;

[0165] Generate promotional images based on the text description combined with the acquired demographic data of target customers and the corresponding pet information through a text-to-image generation model;

[0166] Extract feature vectors of promotional images;

[0167] The fine-tuning model is used to fine-tune the feature vector of the generated promotional image to obtain a fine-tuned promotional image.

[0168] In some embodiments, the demographic data of the target customer includes, but is not limited to, one or any two or more of the target customer's gender, age, income, occupation, and hobby information.

[0169] In some embodiments, the pet information includes one, any two or more of the pet type, age, disease history, and vaccination history.

[0170] In some embodiments, the Stable Diffusion large model is loaded in the text-to-image generation model.

[0171] In some embodiments, the working principle of the system includes fine-tuning using a LoRA fine-tuning model.

[0172] In some embodiments, the fine-tuning module uses the fine-tuning model to fine-tune the generated promotional image, specifically including applying LoRA parameters in the attention layer of the text-to-image generation model to adjust the feature vector processed by the model.

[0173] For more specific methods, please refer to the description of the method, which will not be repeated here.

[0174] The above-mentioned embodiment can accurately design promotional images according to the characteristics of target customers, thereby improving the promotion effect and conversion rate.

[0175] The above embodiments are also applicable to various pet insurance companies or related enterprises, and can be used to produce various pet insurance promotional pictures, such as product introduction pictures, promotional activity pictures, etc. At the same time, this application can also be applied to other fields, such as health care, financial management, etc., to provide more personalized and accurate marketing services for different industries. The following are some possible application scenarios and industries:

[0176] Healthcare:

[0177] Produce introductory graphics and promotional materials for health insurance products.

[0178] Design promotional graphics for specific diseases or treatments, such as heart disease, diabetes, etc.

[0179] Make posters or brochures to popularize health knowledge.

[0180] Financial Management:

[0181] Create introductory diagrams for various investment products, such as funds, stocks, bonds, etc.

[0182] Design promotional graphics for financial solutions targeting different customer groups.

[0183] Produce charts and illustrations for financial market analysis reports.

[0184] educate:

[0185] Create course introductions and promotional graphics, such as online courses, training programs, etc.

[0186] Design posters and promotional materials for academic conferences or seminars.

[0187] Create interpretive maps of educational policies or research findings.

[0188] travel:

[0189] Produce introductory maps and brochures for tourist destinations.

[0190] Design a poster for a travel package or special offer.

[0191] Create cultural and historical background maps for tourist attractions.

[0192] Retail and e-commerce:

[0193] Produce product introduction pictures and promotional advertisements.

[0194] Design promotional graphics for festivals or special events, such as Singles Day, Black Friday, etc. Create illustrations for user guides and product manuals.

[0195] Technology and Innovation:

[0196] Create presentation graphics for new technologies or products, such as artificial intelligence, blockchain, etc. Design promotional materials for technology conferences or exhibitions.

[0197] Produce charts and illustrations for technology trend reports.

[0198] Environmental protection and sustainable development:

[0199] Produce introductory pictures and brochures for environmental protection projects.

[0200] Design a poster for an environmental event or initiative.

[0201] Produce graphics and illustrations for sustainability reports.

[0202] Non-profit organizations and charities:

[0203] Produce introductory pictures and promotional materials for public welfare projects.

[0204] Design a poster for a fundraising event or volunteer recruitment.

[0205] Create graphs and illustrations for research reports on social issues.

[0206] Government and Public Services:

[0207] Produce policy promotion graphics and public service introduction graphics.

[0208] Design a poster for a public safety or health education campaign.

[0209] Produce reports and diagrams for urban planning or infrastructure development.

[0210] Culture and Arts:

[0211] Produce brochures and flyers for art exhibitions.

[0212] Design a poster for a cultural festival or concert.

[0213] Produce interpretive maps of works of art or cultural heritage.

[0214] These application scenarios are just the tip of the iceberg. With the development of digital marketing and visual communication technology, the application potential of the above-mentioned embodiments will be further expanded, providing more abundant and efficient marketing and promotion services for all walks of life.

[0215] In a third aspect, the present application further provides a computer storage medium, wherein the computer storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor and executing the above-mentioned method steps.

[0216] The present application also provides a computer storage medium, which stores multiple instructions, which are suitable for being loaded by a processor and executing the above method steps. Computer storage medium is a physical device for storing data, such as a hard disk, an optical disk, a USB flash drive, etc. These storage media can store various types of data, including text, images, audio, video, and program code. The storage medium stores some programs or scripts, which contain a series of instructions suitable for being loaded by a processor and executing the above method steps, and these instructions can be read, parsed, and executed by the computer's processor.

[0217] For example, an executable file (such as an .exe file) stored on a USB flash drive contains a series of instructions. When this file is copied to a computer and run, the computer's processor will operate according to the instructions in the file to achieve a specific function or task. The specific execution process can be found in the detailed description of the system and method, which will not be repeated here.

[0218] The present application also provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the above method steps.

[0219] See also Figure 3 , is a schematic diagram of the main hardware structure of the electronic device provided in the embodiment of the present application. Figure 3 As shown,

[0220] The customer voice demand classification device 300 may include: at least one processor 301 , at least one network interface 304 , a user interface 303 , a memory 305 , and at least one communication bus 302 .

[0221] The communication bus 302 is used to realize the connection and communication between these components.

[0222] The user interface 303 may include a touch screen and a camera.

[0223] The network interface 304 may optionally include but is not limited to a standard wired interface, a wireless interface (such as a WI-FI interface, Bluetooth, a traffic card, etc.).

[0224] Among them, the processor 301 may include one or more processing cores. The processor 301 uses various interfaces and lines to connect the various parts of the entire multi-tenant mode data acquisition system 300, and executes various functions and processes data of the multi-tenant mode data acquisition system 300 by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Optionally, the processor 301 can be implemented in at least one hardware form of digital signal processing (DSP), field programmable gate array (FPGA), and programmable logic array (PLA). The processor 301 can integrate one or more combinations of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; and the modem is used to handle wireless communication. It is understandable that the above-mentioned modem may not be integrated into the processor 301, but may be implemented by a separate chip.

[0225] Among them, the memory 305 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may optionally be at least one storage device located away from the aforementioned processor 301. As a computer storage medium, the memory 305 may include an operating system, a network communication module, a user interface module and an application. In Figure 3 In the system 300 shown, the user interface 303 is mainly used to provide an input interface for the user and obtain the data input by the user; and the processor 301 can be used to call the application stored in the memory 305 and specifically execute the following steps: Figure 1 The specific process can be referred to Figure 1 As shown, no further description is given here.

[0226] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium along with the main application program. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be cloud storage, a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0227] The features or feature combinations mentioned above in the specification, drawings and claims, as long as they are meaningful within the scope of the present invention and do not contradict each other, can be used in any combination or alone. The advantages and features described for the system provided by the present invention are applicable to the method and computer-readable storage medium provided by the present invention in a corresponding manner, and vice versa.

[0228] In summary, this application eliminates the problem that traditional small models have limited ability to understand long texts and cannot effectively learn more contextual information. Secondly, the language recognition learning model based on multiple large models can have a more thorough understanding of diverse vocabulary and grammatical rules. In addition, it can better handle ambiguity and have a clearer understanding of more contextual information. By introducing language recognition learning models of multiple large models, the accuracy and stability of classification can be improved; the scoring comparison method can fully consider the prediction results of each model, reduce the influence of subjectivity, and improve the objectivity and stability of the prediction.

[0229] It should also be noted that the terms "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, product or apparatus that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, product or apparatus. In the absence of further restrictions, the elements defined by the sentence "comprises one..." do not exclude the presence of other identical elements in the process, method, product or apparatus that includes the elements.

[0230] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for producing a pet insurance promotional picture, characterized in that: include: Obtain demographic data of target customers and corresponding pet information; Generate promotional images based on the text description combined with the acquired demographic data of target customers and the corresponding pet information through a text-to-image generation model; Extract feature vectors of promotional images; The fine-tuning model is used to fine-tune the feature vector of the generated promotional image to obtain a fine-tuned promotional image.

2. The method for producing a pet insurance promotional picture according to claim 1, characterized in that: The demographic data of the target customers include one or any two or more of the target customers' gender, age, income, and hobby information.

3. The method for producing a pet insurance promotional picture according to claim 1 is characterized in that: The pet information includes one, any two or more of the pet type, age, disease history, and vaccination history.

4. The method for producing a pet insurance promotional picture according to claim 1 is characterized in that: The text-to-image generation model is loaded using the Stable Diffusion large model.

5. The method for producing a pet insurance promotional picture according to claim 1 is characterized in that: The fine-tuning model is fine-tuned using LoRA.

6. The method for producing a pet insurance promotional picture according to claim 1: characterized in that: The method for fine-tuning the generated promotional image using the fine-tuning model includes: in the attention layer of the text-to-image generation model, applying LoRA parameters to adjust the feature vector processed by the model.

7. A pet insurance promotional picture making system, characterized in that: include: Data and information acquisition module, used to obtain demographic data of target customers and corresponding pet information; A text-to-image generation module is used to generate a promotional image based on the text description combined with the acquired demographic data of the target customers and the corresponding pet information through a text-to-image generation model; Feature vector extraction module, used to extract feature vectors of promotional images; The fine-tuning module uses the fine-tuning model to fine-tune the feature vector of the generated promotional image to obtain the fine-tuned promotional image.

8. The pet insurance promotional picture making system according to claim 7, characterized in that: The text-to-image generation module adopts the Stable Diffusion large model, and the fine-tuning module adopts the LoRA model for fine-tuning.

9. A computer storage medium, characterized in that: The computer storage medium stores a plurality of instructions, which are suitable for being loaded by a processor and executing the method steps according to any one of claims 1-6.

10. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the method steps according to any one of claims 1 to 6.

Images