Commodity image generation method and system, computer equipment and storage medium

By analyzing user review data, product images are automatically generated, solving the problem of monotonous product image content in existing technologies, achieving more efficient and diversified image generation, and reducing costs.

CN119005145APending Publication Date: 2024-11-22ALIBABA (CHINA) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410947018.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-15
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

In existing technologies, product images are often monotonous, and the product images produced by professional design teams are limited by the materials provided by sellers, resulting in insufficient image content, low efficiency, and high costs.

Method used

By analyzing user review data, extracting image content elements and generating text title elements, and combining them with specified layout constraints, product images are generated. This allows for the automated generation of diverse product images using computer vision and large language models.

Benefits of technology

It improves the diversity and efficiency of product image generation, reduces reliance on professional design teams, lowers production costs, and generates images that better reflect the diverse needs of consumers.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119005145A_ABST
    Figure CN119005145A_ABST
Patent Text Reader

Abstract

The embodiment of the invention provides a commodity image generation method and system, computer equipment and a storage medium. The method is used for generating a commodity image according to target user comment data of a commodity; the target user comment data comprises image comment data and text comment data; the method comprises the steps of extracting image content elements from image comment data of target user comment data; generating a text title element according to the text comment data of the target user comment data; and combining the image content element and the text title element with specified layout constraint information to generate a commodity image. The commodity image is generated through the target user comment data of the user for the commodity, and the commodity image with various contents can be generated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments in this specification relate to the field of computer data processing, and specifically to a method, system, computer device, and storage medium for generating a product image. Background Art

[0002] With the rapid development of e-commerce, product images displayed on e-commerce platforms have become one of the key factors influencing consumers' purchasing decisions.

[0003] In the prior art, product images are usually designed and produced by a professional design team. Specifically, the professional design team usually shoots product materials based on the needs of the seller, and further performs post-production based on the photographed materials.

[0004] However, the content of the material obtained by photographing the product is relatively limited, so the content of the product image produced by the method in the prior art is relatively simple. Summary of the invention

[0005] Multiple embodiments of this specification provide a product image generation method, system, computer device and storage medium, which can generate product images with diverse contents.

[0006] One embodiment of the present specification provides a product image generation method for generating a product image based on target user comment data of a product; the target user comment data includes image comment data and text comment data; the method includes: extracting image content elements from the image comment data of the target user comment data; generating text title elements based on the text comment data of the target user comment data; and combining the image content elements and the text title elements with specified layout constraint information to generate a product image.

[0007] One embodiment of the present specification provides a product image generation device, which is used to generate a product image based on target user comment data of a product; the target user comment data includes image comment data and text comment data; the device includes: an image content element extraction module, which is used to extract image content elements from the image comment data of the target user comment data; a text title element generation module, which is used to generate text title elements based on the text comment data of the target user comment data; and a product image generation module, which is used to generate a product image by combining the image content elements and the text title elements with specified layout constraint information.

[0008] One embodiment of the present specification provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, a method in any of the above embodiments is implemented.

[0009] One embodiment of the present specification provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, a method in any of the above embodiments is implemented.

[0010] One embodiment of the present specification provides a computer program product, which is used to implement the method of any of the above embodiments.

[0011] The multiple implementation methods provided in this specification can generate corresponding product images through the target user comment data generated by the user's comments on the product. Since there are certain differences between each individual user, the user comment data has a high diversity, so that the generated product images also have a high diversity. Specifically, the target user comment data may include image comment data and text comment data, and image content elements can be extracted from the image comment data, and text title elements can be generated based on the text comment data. Thus, by combining the image content elements and the text title elements with the specified layout constraint information, a product image with diverse content can be generated. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A schematic diagram of a method for generating a product image provided in accordance with one embodiment of the present specification.

[0013] Figure 2 A schematic diagram of a method for generating a product image provided in accordance with one embodiment of the present specification.

[0014] Figure 3 A flowchart of a method for generating a product image provided in accordance with one embodiment of the present specification.

[0015] Figure 4 A schematic diagram of a device for generating a product image according to an embodiment of the present specification.

[0016] Figure 5 Schematic diagram of a computer device provided for one embodiment of the present specification DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments.

[0018] In the description of the embodiments of the present application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more features. In the description of the embodiments of the present application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0019] In the related art, product images need to be designed and produced by a professional design team. When producing product images, the professional design team will complete the production based on the wishes of the product seller. For example, the seller may hope that the product images designed by the professional design team can focus on the selling points of the product identified by the seller. Therefore, the product images produced by the professional design team can better reflect the seller's wishes.

[0020] However, the content of product images produced by professional design teams is relatively simple.

[0021] On the one hand, since some sellers may have more targeted intentions, the contents of multiple product images produced by the professional design team may all express the seller's same intention, making the contents of different product images relatively similar. For example, for some decorations, sellers may think that the texture of the decorations is more distinctive, so the product images produced by the professional design team may all focus on presenting the texture of the decorations.

[0022] On the other hand, the materials that professional design teams can use to create product images are relatively limited. For example, for some clothing, the number of models hired by sellers may be limited. Therefore, due to the limitation of the number of models, the image materials obtained by professional design teams through photographing models wearing clothing are not rich.

[0023] In addition, the product images produced by professional design teams in related technologies also face the problems of low efficiency and high cost.

[0024] Through in-depth analysis, we found that the content of product images produced by relevant technologies is mainly determined by the sellers of the products, which is one of the important reasons why the content of product images is relatively simple. Limited by the personal preferences of sellers and the lack of rich materials provided to professional design teams by sellers due to cost constraints, the content of product images is not rich enough.

[0025] To this end, the implementation methods of this specification take into account the high diversity of consumer preferences and automatically generate product images that meet the wishes of different consumers from the perspective of consumers, which can improve the richness of the content in the product images to a certain extent.

[0026] Specifically, in the implementation of this specification, a product image can be constructed based on user comment data. The user comment data is data generated by different users commenting on the product. Since different users may have different characteristics of the products they are interested in, the user comment data generated by different users have great differences, so the product image generated by the user comment data also has high diversity.

[0027] Furthermore, the target user comment data targeted by the implementation of this specification may include image comment data and text comment data. Therefore, the image content element can be extracted from the image comment data. According to the text comment data of the target user comment data, a text title element can be generated. Then, the image content element and the text title element are combined with the specified layout constraint information to generate a product image, and a product image with diverse content can be obtained.

[0028] The embodiment of this specification provides a commodity image generation system. The commodity image generation system may include a client and a server. The client may interact with the user. The client may be an electronic device with network access capability. Specifically, for example, the client may be a desktop computer, a tablet computer, a laptop computer, a smart phone, a digital assistant, a smart wearable device, a shopping guide terminal, a television, a smart speaker, a microphone, etc. Among them, smart wearable devices include but are not limited to smart bracelets, smart watches, smart glasses, smart helmets, smart necklaces, etc. Alternatively, the client may also be software that can be run in an electronic device. The server may be an electronic device with certain computing and processing capabilities. It may have a network communication module, a processor, and a memory, etc. Of course, the server may also refer to software running in an electronic device. The server may also be a distributed server, which may be a system with multiple processors, memories, network communication modules, etc. operating in collaboration. Alternatively, the server may also be a server cluster formed by several servers. Alternatively, with the development of science and technology, the server may also be a new technical means that can realize the corresponding functions of the embodiment of the specification. For example, it may be a new form of "server" based on quantum computing.

[0029] See also Figure 1 and Figure 2 One embodiment of the present specification provides an application scenario example of a commodity image generation system.

[0030] The product image generation system may include a server and a client. The server may be used to run the product image generation method. The client may be used for the user to interact with the product image generation system. The user of the product image generation system may be a staff of an e-commerce platform. The staff may design a new layout template in the product image generation system. The staff hopes to use the new layout template to generate product images. The new layout template may be used as part of the specified layout constraint information.

[0031] The staff can operate on the client to instruct the product image generation system to generate a product image for the specified layout constraint information. The specified layout constraint information can have a first product category tag. The first product category tag can indicate the product category to which the product suitable for generating the product image with the specified layout constraint information belongs. For example, the first product category tag can indicate spring clothing.

[0032] Accordingly, the server can filter out the first user comment data in the user comment data set according to the first product category label. The user comment data set can be maintained on the server, including multiple user comment data. Multiple user comment data can correspond to product category labels respectively. The product category label can indicate the product category to which the product targeted by the corresponding user comment data belongs. That is, the user commented on the product of which product category. Thus, the server can filter out the first user comment data with the same or similar label as the first product category in the user comment data set. Specifically, the first user comment data may refer to the comments made by consumers on "spring clothing".

[0033] Exemplarily, there are two first user comment data, including first user comment data 1 and first user comment data 2.

[0034] The first user review data 1 may include image review data describing a photo and a text review data. The image review data may represent an image used to review a product. The first user review data 1 may be a user review of shirt A. The image review data of the first user review data 1 describes two girls wearing shirt A. The content of the text review data of the first user review data 1 is: "This shirt is made of good material. My friend and I like it very much."

[0035] The first user comment data 2 includes image comment data describing two photos and a text comment data. The first user comment data 2 may be a user comment on T-shirt B. The image comment data of the first user comment data 2 describes that there are two boys wearing T-shirts B in the photo 1. The image comment data of the first user comment data 2 describes that there are two T-shirts B in the photo 2. The text comment data of the first user comment data 2 reads: "This T-shirt has a good style. My brother and I like it very much."

[0036] The specified layout constraint information may also have constraint conditions for image comment data and / or text comment data. Exemplarily, the constraint conditions for the specified layout constraint information may include limiting that image comment data in the user comment data needs to describe two or more characters, and limiting that the content of text comment data in the user comment data needs to be greater than 10 Chinese characters.

[0037] After obtaining the first user comment data, the server can filter out the first user comment data whose text comment data contains more than 10 Chinese characters. In addition, the server can identify whether there are two or more people in the picture described by the image comment data of the first user comment data based on computer vision technology. If the first user comment data meets the above constraints, it can be confirmed as the target user comment data. For example, the first user comment data 1 and the first user comment data 2 both meet the above constraints and can be used as the target user comment data.

[0038] Furthermore, the server may generate a product image according to the first user comment data 1 and the first user comment data 2 in combination with the specified layout constraint information.

[0039] Specifically, the specified layout template included in the specified layout constraint information may have multiple element filling areas. The specified layout constraint information includes the filling specifications corresponding to the element filling areas. The filling specifications may define the types of elements filled in the corresponding element filling areas and other contents. Exemplarily, the specified layout template may include three element filling areas. Element filling area 1 may be defined as filling text title elements to form a title in the product image. Element filling area 2 and element filling area 3 may be defined as filling image content elements. The filled image content elements need to describe a person wearing clothing of the product category "Spring Clothing". And, the image content elements filled in element filling area 2 and element filling area 3 need to describe different characters.

[0040] Therefore, for the element filling area filled with image content elements, the server can extract character information from the image comment data based on computer vision technology as image content elements. Specifically, the server can identify one or more of the model image, full-body or half-body image, specific human posture image, product image or image color value based on one or more of the means of human body detection, human key point detection combined with posture judgment rules, product detection, color recognition, etc. Then, use one or more of the means of image cropping, clipping or stroking to process the identified content. Finally, use layer filling or material replacement and other means to fill the extracted character information, that is, the image content element, into the corresponding element filling area.

[0041] For the element filling area filled with the text title element, the server can construct a title generation prompt instruction through the text comment data, and then call the large language model through the title generation prompt instruction to instruct it to understand and summarize the text comment data and generate a text title element. After the server obtains the text title element fed back by the large language model, it can fill it into the element filling area corresponding to the title in the specified layout constraint information.

[0042] Finally, the server can feed back the generated product images to the merchant's client for use by the merchant. For example, the image generated by the server for the first user comment data 1 can include two girls wearing spring clothes and a title. The title content can be: Spring Comfortable Outfits. The image generated for the first user comment data 2 can include two boys wearing spring clothes and a title. The title content can be: Spring Trend Styles.

[0043] See also Figure 3 One embodiment of the present specification provides a method for generating a product image. The method for generating a product image can be applied to a server. The method for generating a product image can be used to generate a product image according to target user review data of a product. The method for generating a product image can include the following steps.

[0044] Step S110: extracting image content elements from the image comment data of the target user comment data.

[0045] Goods can refer to any commodity. For example, goods can be clothing, electronic products, or furniture. Of course, goods can also be services provided to users. For example, online consultation services or cleaning services.

[0046] Product images may refer to images used to display products. Specifically, product images may focus on displaying the features of the product. Alternatively, product images may also be used to display visual content that attracts target consumers to purchase the product.

[0047] User review data may refer to data generated after a user comments on a certain product. For example, a user may be a consumer who purchases a product on an e-commerce platform. A user may post a review on an e-commerce platform after purchasing a product. The review content may include a product description or product experience entered by the user in text. Of course, the review content may also include product pictures, videos or other audio and video images taken by the user.

[0048] The target user comment data may be a type of user comment data, and may include image comment data and text comment data.

[0049] Image review data can represent image data such as pictures and videos used as product reviews. For example, after a user purchases a product on an e-commerce platform, he or she can post photos and videos of the purchased product in the review to evaluate the style, color and other information of the product.

[0050] Text review data can represent the text content of product reviews. For example, users can express their feelings about the price, function, or user experience of a product through text when commenting.

[0051] In some cases, the contents of the image review data and the text review data may cooperate with each other to express the user's evaluation of the product.

[0052] The image content element may represent a visual element extracted from the image comment data. The visual element may be a specific object that appears in the picture or video represented by the image comment data. For example, the visual element may be a commodity or a person that appears in the picture or video represented by the image comment data. Accordingly, the image content element may be represented by a pixel set formed by pixels used to describe the corresponding commodity or person in the image comment data. Alternatively, a target pixel set may be obtained after performing data processing such as image enhancement on the pixel set. The image content element may be represented by a target pixel set.

[0053] In some embodiments, the image content element may include a person, the product, or a person using the product. Accordingly, the server may identify a target area where the image content element is located in the image comment data; and construct the image content element based on the target area data describing the target area obtained from the image comment data.

[0054] The target area may be an area in the image comment data that describes an image content element. Accordingly, the target area data may represent a pixel set formed by pixels in the image comment data that describe the target area. For example, the commodity may be a school bag. The image represented by the image comment data may contain a target area that displays the school bag. The target area data may be a pixel set formed by pixels in the image comment data that describe the school bag. In some embodiments, the target area data may be used as an image content element.

[0055] The server can identify the target area in any way. Specifically, the server can identify the target area where the image content element is located in the image comment data by computer vision technology. For example, the product can be clothing. The image content element can represent a person wearing the clothing. The server can segment the target area where the person wearing the clothing is located in the image comment data by image segmentation technology to obtain target area data. The image content element can be formed based on the target area data.

[0056] In some embodiments, when identifying the target area where the image content element is located, the server may use different methods to identify the target area according to the characteristics of the specified layout constraint information. For example, the specified layout constraint information may be a layout template constructed for spring clothing. The specified layout constraint information may want to display a person wearing clothing. Therefore, when the server identifies the target area in the image comment data, it may use one or more of the methods such as human body detection, human key point detection combined with posture judgment rules to identify the person. The product and image color value and other information are identified by means of product detection, color recognition, etc. In this way, the image content elements representing the person wearing clothing can be extracted more accurately. For some other layout templates, for example, some layout templates are layout templates designed for products with more prominent geometric shapes. Accordingly, the server can also identify the image content elements by using an image recognition algorithm that is good at identifying geometric features.

[0057] After obtaining the target area data, the server may directly use it as an image content element. Alternatively, the server may also perform post-processing on the target area data to construct an image content element. Specifically, the server may process the target area data to perform image cropping, cutting out, or stroking on the target area to obtain an image content element.

[0058] In some embodiments, the specified layout constraint information may have certain constraints on the image content elements. Accordingly, the server may also extract the image content elements from the image comment data according to these constraints. For example, the specified layout constraint information limits the image content element to a half-body image. The server can identify the half-body image in the image comment data. Or, if the target area identified by the server in the image comment data is a full-body photo of a person. The server can further process the target area data to crop the target area to form a half-body image of the person as the image content element. Of course, the specified layout constraint information may limit the image content elements that need to describe two or more people. The server needs to extract the image content elements that describe two people in the image comment data.

[0059] In some embodiments, the server may also provide the image comment data and / or text comment data in the target user comment data to the multimodal big model, and generate image content elements through the multimodal big model according to the requirements of the image content elements of the specified layout constraint information.

[0060] Step S120: Generate a text title element according to the text comment data of the target user comment data.

[0061] The text title element may represent content used to form a title in a product image. The title of a product image may be used to express the content that the product image is intended to highlight. For example, a product image may include two models wearing clothing. The product image may want to highlight the style of the clothing. Therefore, the title of the product image may be "Fashionable Style Clothing". Accordingly, the text title element may refer to "Fashionable Style Clothing".

[0062] Since the text comment data in the target user comment data is formed by user input, different users may have different expression habits, ways of expression, and languages. In order to better understand different users' concerns about the product and generate the title of the product image, that is, the text title element, the text comment data can be provided to the large language model, and the large language model can be instructed to generate the text title element.

[0063] Specifically, the server may construct a title generation prompt instruction through the text comment data; and call the large language model according to the title generation prompt instruction to instruct the large language model to generate the text title element based on the text comment data.

[0064] The title generation prompt instruction may represent an instruction (Prompt) provided to the large language model. Among them, the title generation prompt instruction may include text comment data in the target user comment data. For example, the text comment data may be "The material of this clothing is very good." Accordingly, the constructed title generation prompt instruction may be "Please understand the following content, which is the comments of consumers when purchasing goods. Please extract the user's central theme from it, so as to help me build a title to intuitively reflect the following content. Content: The material of this clothing is very good." Accordingly, the large language model may feedback "Comfortable material wear". Therefore, "Comfortable material wear" can be used as a text title element.

[0065] In some embodiments, the specified layout constraint information may have certain constraints on the text title element. For example, the specified layout constraint information may require that the number of words in the text title element needs to be within 3 to 10 words. Accordingly, the server can provide the constraints of the specified layout constraint information when calling the large language model to generate the text title element. For example, the prompt instruction can be adjusted to: "Please understand the following content. The following content is the comments of consumers when purchasing goods. Please extract the user's central theme from it to help me build a title. Please note that the number of words in the title should be controlled between 3 and 10 words, and can intuitively reflect the following content. Content: The material of this clothing is very good."

[0066] Of course, the server may also perform post-processing on the text title element fed back by the large language model to satisfy the constraints of the specified layout constraint information.

[0067] In some embodiments, the large language model may be a generative model for generating text generated through large-scale corpus training. Of course, the large language model may also be a multimodal large model. Accordingly, the text comment data and the image comment data may also be provided to the multimodal large model to instruct the multimodal large model to generate a text title element by combining the text comment data and the image comment data.

[0068] Step S130: Combining the image content element and the text title element with the specified layout constraint information to generate a product image.

[0069] In this embodiment, the specified layout constraint information may include a specified layout template. The specified layout template may be used to indicate the layout of the text title element and the image content element of the product image in the product image.

[0070] Specifically, the specified layout template may include a plurality of element filling areas for the image content element or the text title element to fill. The element filling area may represent a pit position in the specified layout template provided for the image content element or the text title element to fill or replace. The positions of different element filling areas in the specified layout template may be different.

[0071] The specified layout constraint information includes a fill specification corresponding to the element fill area. The fill specification can limit the content filled into the element fill area. For example, the fill specification can limit whether the element filled into the element fill area is a text title element or an image content element. Alternatively, the fill specification can limit the content of the text title element and / or image content element filled into the element fill area. For example, the fill specification may limit the number of characters of the filled text title element to be within the range of 3 to 5 Chinese characters. Alternatively, the fill specification may limit the filled image content element to be a half-length photo of a person, etc.

[0072] In some implementations, the fill specifications of all element fill areas in the specified layout constraint information may form a constraint condition for specifying a layout template.

[0073] The server can fill the image content element or the text title element into the corresponding element filling area according to the filling specification to form the product image. For example, the server extracts two image content elements, namely a shirt and a T-shirt. And the server generates a text title element. The specified layout template may have three element filling areas, which are used to fill clothing 1, clothing 2 and title respectively. Accordingly, the server can fill the image content elements representing the shirt and T-shirt, and the text title elements into the corresponding element filling areas.

[0074] In the implementation of this specification, since the user review data is data provided by consumers, it can directly reflect consumers' concerns about the product. In addition, since the amount of user review data is large, different consumers may have different opinions. Therefore, the content of the product image generated by the user review data is not limited to the selling points unilaterally identified by the seller, and is more diverse than related technologies.

[0075] In addition, in the embodiments of this specification, the method for generating product images is automatically run by a computer device, which improves the efficiency of generating product images to a certain extent compared to the related art that requires a professional design team to produce. Due to the increase in efficiency, a larger number of product images can be generated per unit time compared to the related art. There are still some differences between these product images, which indirectly increases the content diversity of the product images.

[0076] In some embodiments, the specified layout constraint information has a first product category label. The first product category label indicates the product category to which the product to which the specified layout constraint information is applied to generate the product image belongs. The product image generation method may further include: based on the first product category label, filtering out user comment data generated by users commenting on the product with the first product category label in the user comment data set as the target user comment data; wherein the user comment data set includes a plurality of user comment data.

[0077] In some cases, the specified layout constraint information may be designed and formed according to a unified design specification. However, it takes a lot of time to design independent specified layout constraint information for each product.

[0078] Considering that products of the same product category may have certain similarities. Therefore, the layout of the title text elements and the image content elements in the product images of products of the same product category may be similar. In order to improve the diversity of product images of different product categories and reduce the cost and time spent on designing layout templates, specified layout constraint information may be formulated for product categories. Therefore, the specified layout constraint information may have a product category tag, which may indicate the product category to which the product belongs that is suitable for applying the corresponding specified layout constraint information to generate the product image.

[0079] The first product category label may indicate a product category label of the specified layout constraint information.

[0080] The user comment data set may be a collection of user comment data. The user comment data may also have a product category label, indicating the product category to which the product targeted by the user comment data belongs. Thus, the server may, for each layout template, traverse the user comment data that matches it and generate a product image corresponding to the user comment data. Specifically, the server may, based on the first product category label, filter out the user comment data that matches the first product category label in the user comment data set as the target user comment data. Furthermore, the server may generate a product image by combining the image content elements and text title elements obtained in the target user comment data with the specified layout constraint information.

[0081] In some embodiments, the commodity includes a target commodity. The target commodity has a second commodity category label; the second commodity category label indicates the commodity category to which the target commodity belongs. The commodity image generation method may further include: obtaining user comment data generated by a user commenting on the target commodity as the target user comment data; matching the specified layout constraint information in a layout template library through the second commodity category label; wherein the layout template library includes a plurality of layout templates with commodity category labels; the commodity category label indicates the commodity category to which the commodity to which the corresponding layout template is applied to generate the commodity image belongs.

[0082] In some cases, the server may also traverse the layout templates that match each product category or each product to generate a product image.

[0083] The second product category label may indicate the product category to which the target product belongs.

[0084] The layout template library may be a data set maintained on a server. The layout template library may include multiple layout templates. At least some of the layout templates may correspond to different commodity category labels.

[0085] Thus, for the target product, the server can match the corresponding specified layout constraint information in the layout template library through the second product category tag of the target product. Further, based on the user comment data generated by the user's comments on the target product, the server can obtain the image content element and the text title element therefrom, and generate the product image in combination with the specified layout constraint information.

[0086] In some embodiments, the specified layout constraint information has constraint conditions for image comment data and / or text comment data. The product image generation method may also include: selecting user comment data that meets the constraint conditions from the user comment data set as the target user comment data; wherein the user comment data set includes multiple user comment data.

[0087] In some cases, in order to better reflect the characteristics of a certain product or a certain product category, the specified layout constraint information has certain constraints on the content or arrangement of the image content element and / or the text title element. However, even the user comment data of the product that matches the specified layout constraint information may not meet the constraint conditions of the specified layout constraint information. Therefore, when the server filters in the user data set, it can filter out the user comment data that meets the constraint conditions as the target user comment data.

[0088] For example, the product may be gloves. The constraint condition may be that the image content element needs to include full-body photos of two people wearing gloves. Then, some user review data that only includes partial parts of the person can be filtered out.

[0089] In some embodiments, the constraints on the image comment data may include at least one of the following: limiting the number of images represented by the image comment data, limiting the type of specified objects included in the image, limiting the number of specified objects included in the image, limiting the part of the specified object displayed in the image, or limiting the posture of the specified object in the image when the specified object is a person.

[0090] When a user reviews a product, they may upload one or more images. Therefore, the constraint may include the number of images represented by the image review data.

[0091] The specified object may refer to any object. Specifically, the specified object may be a person, an animal, a plant, or an industrial product. Correspondingly, the type of the specified object may indicate which type of person, animal, plant, or industrial product the specified object is. For example, the constraint condition may limit the specified object to be a human being. That is, there must be a person in the image represented by the image comment data.

[0092] The number of specified objects may be the number of specified objects in the image represented by the image comment data. For example, the constraint may be that the number of people in a single image must be more than two. Alternatively, the constraint may be that the image described by the image comment data includes at least two different people.

[0093] Alternatively, the constraint condition may also limit the image described by the image comment data to include a certain part of the specified object. For example, when the specified object is a person, the constraint condition may limit the image to include the person's limbs or the person's upper body.

[0094] Of course, the constraint condition may also limit the posture of a specified object in the image described by the image comment data. For example, when the specified object is a person, the constraint condition may limit the person in the image to be in a standing posture. Alternatively, the constraint condition may limit the person in the image to have his eyes closed.

[0095] In some implementations, the constraint conditions for the text comment data include at least one of the following: limiting the number of specified characters in the text comment data or limiting the number of words and / or sentences in the text comment data.

[0096] For the text comment data, the constraint condition may be the number of characters in the text comment data, the number of words in the text comment data, or the number of sentences in the text comment data.

[0097] In some embodiments, the step of generating a product image by combining the image content element and the text title element with specified layout constraint information includes: constructing an image generation prompt instruction through the image content element, the text title element and the specified layout constraint information; calling an image generation big model according to the image generation prompt instruction to instruct the image generation big model to generate a product image based on the image content element and the text title element.

[0098] In this embodiment, the image generation model can be a multimodal model. The image generation model can generate a product image that meets the requirements of the image generation prompt instruction based on the input image generation prompt instruction. The image generation prompt instruction is constructed based on the image content element, the text title element, and the specified layout constraint information, so that the image generation prompt instruction can be used to instruct the image generation model to generate a product image that meets the specified layout constraint information based on the image content element and the text title element.

[0099] In the process of generating product images, the image generation model can utilize image content elements and text title elements, so that the final product image can combine the specified layout constraint information with the image generation model's own content generation capability, so that the text content, image content, and layout in the final product image can be more adaptable. In addition, because the image generation model's own content generation capability is utilized, there will be some slight changes between the generated multiple product images on the basis of meeting the specified layout constraint information. As a result, the multiple product images generated according to the implementation method of this specification can have greater applicability.

[0100] In some implementations, when constructing the image generation prompt instruction, only the constraint information included in the specified layout constraint information may be used, without using the specified layout template. Of course, when constructing the image generation prompt instruction, the entire content of the specified layout constraint information may also be used.

[0101] See also Figure 4 One embodiment of the present specification also provides a product image generation system. The product image generation system is used to generate a product image according to target user comment data of the product. The target user comment data includes image comment data and text comment data. The product image generation system may include: an image content element extraction module, a text title element generation module and a product image generation module.

[0102] The image content element extraction module is used to extract image content elements from the image comment data of the target user comment data.

[0103] The text title element generating module is used to generate a text title element according to the text comment data of the target user comment data.

[0104] The product image generation module is used to generate a product image by combining the image content element and the text title element with the specified layout constraint information.

[0105] The specific functions and effects achieved by the commodity image generation system can be explained with reference to other embodiments of this specification, and will not be repeated here. Each module in the commodity image generation system can be implemented in whole or in part by software, hardware, and a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0106] The embodiments of the present specification also provide a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a computer, the computer executes the commodity image generation method in any of the above embodiments.

[0107] The embodiments of the present specification also provide a computer program product including instructions, which, when executed by a computer, enables the computer to execute the product image generation method in any of the above embodiments.

[0108] See also Figure 5 The present embodiment can provide a computer device, which includes: a memory, and one or more processors communicatively connected to the memory; the memory stores instructions that can be executed by one or more processors, and the instructions are executed by the one or more processors to enable the one or more processors to implement the product image generation method in any of the above embodiments.

[0109] In some embodiments, the computer device may include a processor, a non-volatile storage medium, an internal memory, a communication interface, a display device, and an input device connected by a system bus. The non-volatile storage medium may store an operating system and related computer programs.

[0110] The user information or user account information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, etc.) involved in multiple implementation methods of this specification are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0111] It should be understood that the specific examples herein are only intended to help those skilled in the art to better understand the embodiments of the present specification, rather than to limit the scope of the present invention.

[0112] It can be understood that in the various implementations of this specification, the size of the sequence number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the implementation methods of this specification.

[0113] It can be understood that the various embodiments described in this specification can be implemented individually or in combination, and the embodiments of this specification are not limited to this.

[0114] Unless otherwise specified, all technical and scientific terms used in the embodiments of this specification have the same meaning as those generally understood by those skilled in the art of the technical field of this specification. The terms used in this specification are only for the purpose of describing specific embodiments and are not intended to limit the scope of this specification. The term "and / or" used in this specification includes any and all combinations of one or more related listed items. The singular forms of "a", "above", and "the" used in the embodiments of this specification and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0115] It can be understood that the processor of the embodiment of this specification can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method implementation can be completed by the hardware integrated logic circuit or software instructions in the processor. The above processor can be a general processor, a digital signal processor (Digital Signal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiment of this specification can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of this specification can be directly embodied as a hardware decoding processor to perform, or the hardware and software modules in the decoding processor are combined and executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0116] It is understood that the memory in the embodiments of this specification may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memory. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (programmable ROM, PROM), an erasable programmable read-only memory (erasablePROM, EPROM), an electrically erasable programmable read-only memory (EEPROM) or a flash memory. The volatile memory may be a random access memory (RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0117] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this specification.

[0118] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method implementation methods and will not be repeated here.

[0119] In the several embodiments provided in this specification, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device implementation described above is only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0120] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0121] In addition, each functional unit in each embodiment of the present specification may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0122] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this specification, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of each implementation method of this specification. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0123] The above are only specific implementations of this specification, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in this specification, which should be included in the protection scope of this specification. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for generating a product image, characterized in that: Used to generate product images based on target user review data of the product; The target user comment data includes image comment data and text comment data; the method includes: extracting image content elements from the image comment data of the target user comment data; Generate a text title element according to the text comment data of the target user comment data; The image content element and the text title element are combined with the specified layout constraint information to generate a product image.

2. The method according to claim 1, characterized in that The specified layout constraint information has a first commodity category tag; the first commodity category tag indicates the commodity category to which the commodity suitable for applying the specified layout constraint information to generate a commodity image belongs; the method includes: Based on the first product category label, user comment data generated by users commenting on products with the first product category label is filtered out from the user comment data set as the target user comment data; wherein the user comment data set includes multiple user comment data.

3. The method according to claim 1, characterized in that The commodities include a target commodity; the target commodity has a second commodity category label; The second product category label indicates the product category to which the target product belongs; the method further includes: Acquire user review data generated by users commenting on the target product as the target user review data; The specified layout constraint information is matched in a layout template library through the second product category label; wherein the layout template library includes a plurality of layout templates with product category labels; the product category label indicates the product category to which the product that is suitable for applying the corresponding layout template to generate a product image belongs.

4. The method according to claim 1, characterized in that: The specified layout constraint information has constraint conditions for image comment data and / or text comment data; the method further includes: The user comment data satisfying the constraint condition is screened out from the user comment data set as the target user comment data; wherein the user comment data set includes a plurality of user comment data.

5. The method according to claim 4, characterized in that The constraint condition for the image comment data includes at least one of the following: limiting the number of images represented by the image comment data, limiting the type of the specified object included in the image, limiting the number of the specified objects included in the image, limiting the part of the specified object shown in the image, or limiting the posture of the specified object in the image when the specified object is a person; Alternatively, the constraint condition for the text comment data includes at least one of the following: limiting the number of specified characters in the text comment data or limiting the number of words and / or sentences in the text comment data.

6. The method according to claim 1, characterized in that The specified layout constraint information includes a specified layout template; the specified layout template includes a plurality of element filling areas for the image content element or the text title element to fill; the element filling areas have corresponding filling specifications; The step of combining the image content element and the text title element with the specified layout constraint information to generate a product image includes: The image content element or the text title element is filled into the corresponding element filling area according to the filling specification to form the product image.

7. The method according to claim 1, characterized in that The image content elements include a person, the product, or a person using the product; the step of extracting the image content elements from the image comment data of the target user comment data includes: identifying a target area where the image content element is located in the image comment data; The image content element is constructed based on the target area data describing the target area acquired from the image comment data.

8. The method according to claim 1, characterized in that The step of generating a text title element according to the text comment data of the target user comment data comprises: Constructing a title generation prompt instruction through the text comment data; The large language model is called according to the title generation prompt instruction to instruct the large language model to generate the text title element based on the text comment data.

9. The method according to claim 1, characterized in that: The step of combining the image content element and the text title element with the specified layout constraint information to generate a product image includes: Constructing an image generation prompt instruction through the image content element, the text title element and the specified layout constraint information; The image generation model is called according to the image generation prompt instruction to instruct the image generation model to generate a product image based on the image content element and the text title element.

10. A commodity image generation system, characterized in that: Used to generate product images based on target user review data of the product; The target user comment data includes image comment data and text comment data; the system includes: An image content element extraction module, used to extract image content elements from the image comment data of the target user comment data; A text title element generating module, used for generating a text title element according to the text comment data of the target user comment data; The product image generation module is used to generate a product image by combining the image content element and the text title element with the specified layout constraint information.

11. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 9 when executing the computer program.

12. A computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the method according to any one of claims 1 to 9 when executed by a processor.