Image processing method and device, electronic equipment and storage medium

Abstract

The invention provides an image processing method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence, in particular to the fields of large models, generative models, multi-modal large models, image processing and the like. According to the specific implementation scheme, the method comprises the following steps: determining a target structure template from a plurality of preset structure templates according to the number of images meeting a preset condition in a target image set; according to a grouping mode for the target structure template, performing grouping processing on the plurality of images in the target image set to obtain at least one target subset; for each target subset in the at least one target subset, generating a subset copywriting according to the image in the target subset and the description text of the image; and generating image-text data according to the target structure template, the images in the target image set and the subset copywriting.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to the fields of large models, generative models, multimodal large models, image processing, etc. More specifically, the present disclosure provides an image processing method, device, electronic device, storage medium and computer program product. Background Art

[0002] Users often take photos in life, work, travel and other scenarios. Photos in electronic devices such as mobile phones are usually displayed in full in the order of shooting time, which fails to highlight the key points and affects the user's browsing experience. Summary of the invention

[0003] The present disclosure provides an image processing method, an apparatus, an electronic device, a storage medium, and a computer program product.

[0004] According to one aspect of the present disclosure, there is provided an image processing method, comprising: determining a target structure template from a plurality of predetermined structure templates according to the number of images in a target image set that meet predetermined conditions; performing grouping processing on a plurality of images in the target image set according to a grouping pattern used for the target structure template to obtain at least one target subset; generating a subset copy for each target subset in the at least one target subset according to images in the target subset and description texts of the images; and generating graphic and text data according to the target structure template, images in the target image set and the subset copy.

[0005] According to another aspect of the present disclosure, an image processing device is provided, comprising: a target structure template determination module, a grouping module, a subset copywriting generation module, and a graphic and text generation module. The target structure template determination module is used to determine a target structure template from a plurality of predetermined structure templates according to the number of images in a target image set that meet predetermined conditions. The grouping module is used to perform grouping processing on a plurality of images in a target image set according to a grouping pattern for the target structure template to obtain at least one target subset. The subset copywriting generation module is used to generate a subset copywriting for each target subset in at least one target subset according to the images in the target subset and the description text of the images. The graphic and text generation module is used to generate graphic and text data according to the target structure template, the images in the target image set, and the subset copywriting.

[0006] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method provided by the present disclosure.

[0007] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable a computer to execute the method provided by the present disclosure.

[0008] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the method provided by the present disclosure is implemented.

[0009] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0011] Figure 1 is a system architecture diagram of an image processing method and device according to an embodiment of the present disclosure;

[0012] Figure 2 is a schematic flow chart of an image processing method according to an embodiment of the present disclosure;

[0013] Figure 3 is a schematic diagram of an application scenario of the image processing method according to an embodiment of the present disclosure;

[0014] Figure 4 is a schematic diagram of the principle of an image processing method according to an embodiment of the present disclosure;

[0015] Figure 5 is an example diagram of graphic data according to an embodiment of the present disclosure;

[0016] Figure 6 is a schematic diagram of the front-end display effect according to an embodiment of the present disclosure;

[0017] Figure 7 is a schematic structural block diagram of an image processing device according to an embodiment of the present disclosure; and

[0018] Figure 8 It is a structural block diagram of an electronic device used to implement the image processing method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0019] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0020] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0021] In the technical solution disclosed in the present invention, the user's authorization or consent is obtained before obtaining or collecting the user's personal information. For example, a pop-up window or other forms can be used to prompt the user that the function of generating structured graphic data from the album needs to read the relevant information of the image, and the user is asked whether he agrees. The information is collected with the user's authorization, otherwise the information is not collected.

[0022] In some technical solutions, the photo album in the user's mobile phone can be converted into graphic information based on pre-configured rules. The rules may include: if the shooting time and shooting location of the photo hit the pre-set time and location, then use the text that matches the time and location. For example, the shooting date of multiple photos in a photo album is January 1 and the shooting location is the Summer Palace. The text matched by the time and location is "New Year's Day trip to the Summer Palace". Therefore, a title "New Year's Day trip to the Summer Palace" can be generated, and then some or all images from the photo album are selected as content and placed under the title for users to browse.

[0023] However, the above technical solution lacks the ability to structure the atlas and cannot realize personalized sorting and selection based on the image content. In addition, the copywriting of the above technical solution is pre-configured, which makes it difficult to generate personalized copywriting with rich emotional atmosphere.

[0024] The disclosed embodiment aims to propose an image processing method, which selects a target structure template according to the content of the image, then groups multiple images using a corresponding grouping pattern, and then generates a subset copy for each target subset after grouping based on the image and the image's description text, and then generates graphic and text data based on the target structure template, the image in the target image set and the subset copy.

[0025] The technical solution provided by the disclosed embodiment first selects a target structure template based on the content of the image, and then generates graphic data based on the target structure template, so that the target image set can be personalized and organized based on the image content. In addition, grouping is performed by using the grouping mode for the target structure template, and subset copywriting is generated for each target subset, rather than using pre-configured fixed copywriting, so personalized generation of copywriting can be achieved. The disclosed embodiment can organize the target image set and generate personalized copywriting to improve the user experience.

[0026] The technical solution provided by the present disclosure will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] Figure 1 4 is a system architecture diagram of an image processing method and device according to an embodiment of the present disclosure.

[0028] It should be noted that Figure 1 What is shown is merely an example of a system architecture to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0029] like Figure 1 As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0030] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptops, desktop computers, etc.

[0031] The server 105 may be a server that provides various services, such as a background management server (only an example) that provides support for websites browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process the received user request and other data, and feed back the processing results (such as graphic data obtained or generated according to the user request, etc.) to the terminal device.

[0032] It should be noted that the image processing method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the image processing device provided in the embodiment of the present disclosure can generally be set in the server 105. The image processing method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the image processing device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.

[0033] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0034] Figure 2 is a schematic flowchart of an image processing method according to an embodiment of the present disclosure.

[0035] like Figure 2 As shown, the image processing method 200 may include operations S210 to S240.

[0036] In operation S210, a target structure template is determined from a plurality of predetermined structure templates according to the number of images in a target image set that meet a predetermined condition.

[0037] For example, a user can select an album from a gallery of an electronic device through a front-end operation page, and the album is called an initial image set. The initial image set can be used as a target image set, or the initial image set can be filtered and the filtered initial image set can be used as the target image set.

[0038] For example, the predetermined condition may be that the image quality is higher than a predetermined quality, or the image corresponds to a shooting location, or the shooting time of the image is within a predetermined time period, etc. This embodiment does not limit the predetermined condition.

[0039] For example, the predetermined structure template may include a location structure template, a time structure template, a spare structure template, etc., and the predetermined structure template may be pre-configured according to actual needs.

[0040] For example, a first number of images corresponding to a predetermined shooting location and a second number of images corresponding to a predetermined shooting time may be counted, and when both the first number and the second number reach a predetermined value and the first number is greater than or equal to the second number, a location structure template may be selected as the target structure template. When both the first number and the second number reach a predetermined value and the first number is less than the second number, a time structure template may be selected as the target structure template. In other cases, a spare structure template may be selected as the target structure template.

[0041] In operation S220, a plurality of images in the target image set are grouped according to the grouping mode for the target structure template to obtain at least one target subset.

[0042] For example, the grouping modes of each predetermined structure template may be different from each other, and the grouping mode may include grouping according to at least one predetermined dimension, and the predetermined dimension includes shooting location grouping, shooting time, etc. In addition, for the same predetermined dimension, different grouping precisions may be included. For example, for grouping according to shooting location, different granularities may be subdivided according to province, city, district, etc. For grouping according to shooting time, different granularities may be subdivided according to year, month, day, etc.

[0043] In operation S230 , for each target subset in the at least one target subset, a subset copy is generated according to the image in the target subset and the description text of the image.

[0044] In one example, the description text of the image can be determined based on at least one of the shooting time, shooting location, and category of the image, based on a predetermined text organization rule, wherein the category may include the category of the object contained in the image. For example, the description text of an image is "shooting time: January 1, 2024; category: dog, grass, person; shooting location: Summer Palace".

[0045] In another example, the image description text may describe the content of the image, for example, the description text may be determined by a picture-to-text model. For example, the image is input into the picture-to-text model, and the picture-to-text model generates the description text of the image.

[0046] For example, some or all images in the target subset and the description texts of some or all images can be input into the multimodal macro model, and the multimodal macro model can generate subset copy. The subset copy can include at least one of a subtitle and a subcopy. For example, if the image in the target subset is a seaside image, the subtitle can be "Leisure time on a seaside chair", and the subcopy can be "In Weihai, my friends and I sat on a unique white chair and enjoyed the leisure time at the seaside. The blue sky, the blue sea, the clean beach, everything is so beautiful. This is the simplest happiness!"

[0047] In operation S240, graphic and text data are generated according to the target structure template, the images in the target image set, and the subset text.

[0048] For example, based on the target structure template, the graphic content of the target subset can be displayed in sequence. In addition, in the process of displaying the graphic content of the target subset, the subset text can be displayed first, and then the image in the target subset can be displayed, so as to obtain graphic data.

[0049] After obtaining the graphic data, the graphic data can be output so that the front-end page can display the graphic data.

[0050] According to the image processing method provided by the embodiment of the present disclosure, the method first selects a target structure template based on the content of the image, and then generates graphic data based on the target structure template, so that the target image set can be personalized and organized based on the image content. In addition, grouping is performed by using the grouping mode for the target structure template, and subset copywriting is generated for each target subset, rather than using pre-configured fixed copywriting, so personalized generation of copywriting can be achieved. The embodiment of the present disclosure can organize the target image set and generate personalized copywriting to improve the user experience.

[0051] According to another embodiment of the present disclosure, in the process of determining the target image set, an initial image set can be first obtained, and then the images in the initial image set are screened based on at least one of the quality evaluation value of the images in the initial image set, the similarity between any two images, and the category of the images, and then the target image set is determined based on the screened initial image set.

[0052] This embodiment screens the initial image set by at least one of the quality evaluation value, the similarity between any two images, and the category of the image, so that images with low quality, high similarity or predetermined category in the initial image set can be deleted, thereby improving the quality of the target image set, and further improving the quality of the image and text data after generating the image and text data based on the target image set.

[0053] In one example, the above-mentioned screening process includes quality screening, and the process of quality screening includes: in response to detecting that the quality assessment value of an image in the initial image set is less than or equal to a quality threshold, deleting the image. For example, the quality assessment value may include clarity, aesthetics, etc., and the quality assessment value may be determined by a pre-trained image quality assessment model, and the image is input into the image quality assessment model, and the image quality assessment model outputs quality assessment values ​​such as clarity and aesthetics. If the clarity and aesthetics are lower than the set threshold, it can be determined that the image quality is poor and filtered. In this example, low-quality images can be deleted, thereby improving the image quality of the initial image set after screening.

[0054] In another example, the above-mentioned screening process includes category screening, and the process of category screening includes: in response to detecting that the category of the image in the initial image set includes a predetermined category, deleting the image, wherein the predetermined category includes, for example, screenshots or other pre-set categories. For example, an image intelligent classification operator can be used to classify and identify the image, and a list of predetermined categories can be pre-set. If the confidence of the image in a certain category is greater than a confidence threshold, the image is considered to correspond to the category. After classification and identification, an image can correspond to one or more categories. If the category of the image includes a predetermined category, the image can be deleted. In this example, images of predetermined categories can be deleted, so that images of predetermined categories that are not related to the required category can be deleted, thereby improving the image quality of the initial image set after screening.

[0055] In another example, the above-mentioned screening process includes similarity screening, and the similarity screening process includes: for any two images in the initial image set, the two images are screened according to the similarity between the two images and the quality evaluation values ​​of the two images. In this example, images with high similarity can be deleted, so that the initial image set can be screened similarly to deduplication, thereby improving the image quality of the screened initial image set.

[0056] For example, the images in the initial image set are combined in pairs and then the similarity is calculated. The similarity can be the cosine distance. If the similarity between the two images is higher than the set threshold, the two images are considered similar. Then the quality evaluation values ​​of the two images are compared. If the quality evaluation value of the first image is higher than the quality evaluation value of the second image, the first image is retained and the second image is deleted, thereby retaining an image with higher quality. This example can remove duplicates and retain high-quality images, reduce redundant storage, and improve the overall quality and storage efficiency of the target image set.

[0057] For another example, for each image in the initial image set, a weighted calculation can be performed based on the maximum similarity between the image and other images in the initial image set and the quality assessment value of the image, and images with lower weighted calculation results are deleted, while images with higher weighted calculation results are retained.

[0058] It should be noted that the screening process may include at least one of the above-mentioned quality screening, similarity screening, and category screening, and the above-mentioned screening operations may be performed in a predetermined order, for example, quality screening, similarity screening, and category screening may be performed in sequence, and the initial set of images after screening may be gradually reduced during the processing.

[0059] According to another embodiment of the present disclosure, the predetermined condition includes a first sub-condition, and the first sub-condition includes that the image corresponds to a geographical location. The predetermined condition may also include a second sub-condition, and the second sub-condition includes that the shooting time of the image is within a target time range, and the target time range may be a predetermined time period, such as the whole day of January 1, during the National Day, etc. The target time range may also be a time period determined according to the shooting time of the image, for example, the shooting start and end time range of the images in the target image set is within the target time range, and the duration corresponding to the target time range may be 24 hours.

[0060] The process of determining the target structure template may include the following operations: determining whether the number of images in the target image set that meet the first sub-condition is greater than or equal to a second quantity threshold, the second quantity threshold is, for example, 3, 5 or other numbers, and if so, determining the position structure template as the target structure template. If not, it may be determined whether the number of images that meet the second sub-condition is greater than or equal to a third quantity threshold, and if so, the time structure template may be determined as the target structure template, otherwise the spare structure template may be determined as the target structure template. It should be noted that in other embodiments, the order of the above judgment process may be adjusted, and this embodiment does not limit the execution order of the above judgment process.

[0061] This embodiment divides the content structure into a location structure template, a time structure template and a spare template, and selects a suitable target structure template according to the location and time corresponding to the image in the target image set, and the target structure template matches the information corresponding to the image in the target image set. Therefore, when the images in the target image set are subsequently grouped, a target subset that is more reasonably grouped and conforms to the image rules can be obtained.

[0062] According to another embodiment of the present disclosure, the predetermined structure template includes a multi-level classification template and a single-level classification template, the multi-level classification template may include a position structure template and a time structure template, and the single-level classification template is the backup structure template mentioned above. Next, the process of grouping multiple images in the target image set according to the grouping mode used for the target structure template to obtain at least one target subset is described.

[0063] A plurality of images in a target image set may be first divided into an initial subset. After the initial subset is obtained, the target subset may be determined according to the initial subset.

[0064] In one example, if the target structure template is a multi-level classification template, the images in the target image set can be divided into multiple initial subsets based on different classification dimensions, where the categories or precisions of the different classification dimensions are different. In this example, for the case of using a multi-level classification template, the images are grouped in multiple levels based on different classification dimensions or different classification precisions, so that the grouped target subsets have obvious common characteristics, so that the generated graphic data has regularity, avoids the chaos of graphic data, and improves the user experience.

[0065] In one example, if the target structure template is a spare structure template, the images in the target image set can be randomly divided into a predetermined number of initial subsets. The predetermined number can be 1, 2 or other values. In this embodiment, in the case where the target image set cannot use a multi-level classification template as the target structure template, a random grouping method is adopted, which can be used as a fallback strategy to achieve grouping processing of the target images.

[0066] Next, when the target structure template is a multi-level classification template, the process of dividing the images in the target image set into multiple initial subsets based on different classification dimensions is described.

[0067] When the target structure template is a location structure template in a multi-level classification template, the images in the target image set can be grouped at the first level according to the shooting time to obtain at least one subset to be processed, for example, multiple subsets to be processed for different dates such as January 1 and January 2 are generated. Then, for each subset to be processed, the images in the subset to be processed are grouped at the second level according to the shooting location to obtain at least one initial subset, for example, multiple initial subsets such as the Summer Palace in Beijing and the Beihai Park in Beijing are generated for the subset to be processed. In this embodiment, the grouping mode for the location structure template first performs a first level grouping based on the shooting time, and then performs a second level grouping based on the shooting location, so that the initial subsets after grouping conform to the grouping logic of time and place, and the initial subsets after grouping have a certain regularity.

[0068] When the target structure template is a time structure template in a multi-level classification template, the images in the target image set can be grouped at the first level according to the shooting time to obtain at least one subset to be processed, for example, multiple subsets to be processed for different dates such as January 1 and January 2 are generated. Then, for each subset to be processed in at least one subset to be processed, according to the description text of the image in the subset to be processed, for example, the description text of a certain image is "shooting time: January 1, 2024; category: dog, grass, person; shooting location: Summer Palace". Then, based on the large language model, the images in the subset to be processed are grouped at the second level based on time to obtain at least one initial subset. For example, the description text and the first prompt information (prompt) are input into the large language model, and the large language model performs secondary grouping according to time and content to obtain multiple initial subsets such as 8 o'clock-10 o'clock, 17 o'clock-19 o'clock, etc. In this embodiment, the grouping mode for the time structure template first performs primary grouping based on the shooting time, and then performs more fine-grained secondary grouping based on time based on the large language model, so that the initial subsets after grouping conform to the grouping logic of time, and the initial subsets after grouping have a certain regularity, and the use of the large language model for grouping can realize intelligent grouping and improve the accuracy of grouping.

[0069] For example, the first prompt information may be a text in natural language form, for example, the first prompt information is "I took a lot of photos, and now I want to group these photos. I will give you the shooting time of these photos. Please help me group the photos in a reasonable way: ##Photo shooting time##Requirement: Output according to date + time period name (time period), for example: 2024.8.12 Afternoon Wonderful Time (14:30-15:00), Rest and Restart (15:30-17:00), Evening Cozy Time (16:00-19:00), etc. Output in the following format: {"Date": [reference text + time period]}, such as {"August 1, 2023": ["Afternoon Wonderful Time (14:30-15:00)"]}", in some examples, the reference text in the first prompt information may be omitted. If the large language model generates a reference text, the reference text may be added to the graphic data for display.

[0070] Next, the process of determining the target subset based on the initial subset is described.

[0071] In one example, the initial subset may be determined as the target subset.

[0072] In another example, for each initial subset in at least one initial subset, when the number of images in the initial subset is greater than a first number threshold, the similarity between any two images in the initial subset is determined, and then some images in the initial subset are deleted according to the similarity between the any two images and the quality evaluation values ​​of the any two images. Then, the target subset is determined according to the initial subset after deleting some images.

[0073] For example, any two images in the initial subset are combined into image pairs, and the similarity between the two images in the image pair is calculated, and then sorted in descending order according to the similarity. For the image pair with the highest similarity, the quality evaluation values ​​of the two images in the image pair are compared, and the image with the low quality evaluation value is deleted and the image with the high quality evaluation value is retained. For the remaining image pairs, they are still sorted in descending order according to the similarity and the image pair with the highest similarity is reselected, and then the image with the low quality evaluation value in the image pair is deleted and the image with the high quality evaluation value is retained. The above operation is repeated until the number of images is equal to the first number threshold.

[0074] This embodiment can select images in the initial subset and remove images with high similarity and poor quality. Compared with the solution of directly determining the initial subset as the target subset without screening, this example can improve the image quality of the target subset by selecting images. In addition, this embodiment can also limit the number of images in each initial subset to within the first number threshold range, avoiding an excessive number of images in the target subset, thereby avoiding redundant generated graphic data.

[0075] According to another embodiment of the present disclosure, graphic data may be generated based on a target text.

[0076] In one example, the target copy may include at least one of the general title and the general copy of the target image set. For example, the category (such as sky, night view, person, etc.), shooting time (such as January 1, 2024, etc.), and shooting location (such as the Summer Palace in Beijing, etc.) of the target image set can be parsed, and this information can be converted into a description text of the image. Then, the general title and the general copy for the target image set can be generated based on the large language model according to the description text of the image in the target image set. For example, the description text and the second prompt information of each image in the target image set are input into the large language model, and the second prompt information can require the large language model to generate the general title and the general copy according to the target image set. This example uses a large model to generate the general copy and the general title, which can efficiently produce creative content, improve the quality of the general copy and the general title, save manpower time, and has a variety of styles to meet multiple needs.

[0077] In another example, the target copy may include a subset copy, and the subset copy may include at least one of a subtitle and a subcopy. For example, the main image may be determined from the target subset according to at least one of the category and the quality assessment value of the image. For example, an image with the most corresponding categories in the target subset may be selected as the main image, or an image with the highest quality assessment value may be selected as the main image, or an image with the highest quality assessment value may be selected from images with a corresponding category number greater than a predetermined number as the main image. Then, subtitles and subcopy for the target subset may be generated based on the multimodal large model according to the main image and the description text of other images in the target subset except the main image. For example, the main image is input into the multimodal large model with image information, and the remaining images are input into the multimodal large model with text information describing the text, and the third prompt information is input into the multimodal large model at the same time, requiring the multimodal large model to generate subtitles and subcopy. This example uses the multimodal large model to fuse images and texts, and intelligently generates subcopy and subtitles, which can ensure that the generated content has a certain relevance to the image and text, while saving manpower time, and has diverse styles to meet multiple needs.

[0078] Figure 3 It is a schematic diagram of an application scenario of the image processing method according to an embodiment of the present disclosure.

[0079] In this embodiment, the user can select an album from the album list through the front-end operation page. For example, the user's album list includes the following albums: Shanghai Travel Album_1, Beijing Travel Album_2, Summer Weihai Album_3. The user selects the "Summer Weihai" album Album_3 and initiates a request to generate graphic content.

[0080] In addition, the server or client is deployed with a structured content generation service 300, which is used to respond to the received request and execute the image processing method according to the album selected by the user, thereby generating graphic data Data, and then displaying the graphic data Data generated by the structured content generation service to the user through the front-end page.

[0081] When executing the image processing method, the structured content generation service 300 can use generative large models such as the large language model Model_1 and the multimodal large model Model_2, and can also use the image database DB to store information such as image category, shooting time, shooting location, description text, etc.

[0082] The structured content generation service 300 in this embodiment can execute an image processing method to perform information recognition, content organization, and text generation on a large number of disordered photos stored by users to form structured graphic data, thereby realizing the generation function from photo albums to structured graphic data and improving user experience.

[0083] Figure 4 It is a schematic diagram of the principle of the image processing method according to an embodiment of the present disclosure.

[0084] Next, the working principle of the structured content generation service is explained.

[0085] Query processing 401 may be performed first, for example, querying the database for meta information of each image in the album and the initial image set, the meta information including shooting time, shooting location, category and other information.

[0086] Next, the initial image set may be screened to obtain a target image set. The screening process may include quality screening 402, similarity screening 403, and category screening 404. The screening process may refer to the above, and will not be described in detail in this embodiment.

[0087] Next, a target structure template may be selected from the predetermined structure templates. For example, a location structure template determination 411 may be performed first, i.e., determining whether to use the location structure template as the target structure template. If so, a grouping process 421 is performed based on the location structure. If not, a time structure template determination 412 is performed, i.e., determining whether to use the time structure template as the target structure template. If so, a grouping process 422 is performed based on the time structure. If not, a spare structure template is used as the target structure template 413, and then a grouping process 423 is performed based on the used structure template.

[0088] In the grouping process, for example, in the case where the target structure template is a location structure template, the grouping process 421 may include first generating multiple subsets to be processed for different dates such as January 1, January 2, and then generating multiple initial subsets such as Beijing Summer Palace, Beijing Beihai Park, etc. for the subsets to be processed. For example, in the case where the target structure template is a time structure template, the grouping process 422 may include first generating multiple subsets to be processed for different dates such as January 1, January 2, and then for each subset to be processed in at least one subset to be processed, according to the description text of the image in the subset to be processed, based on the large language model, performing secondary grouping based on time on the images in the subset to be processed, and obtaining initial subsets, for example, obtaining multiple initial subsets such as 8 o'clock-10 o'clock, 17 o'clock-19 o'clock, etc. For example, in the case where the target structure template is a spare structure template, the grouping process 423 may include dividing the images in the target image set into an initial subset.

[0089] After obtaining the above initial subsets, the images in the initial subsets may be selected to obtain a target subset. For example, for each initial subset in at least one initial subset, when the number of images in the initial subset is greater than a first number threshold, similarity calculation 424 may be performed between any two images in the initial subset, and then similarity is sorted in descending order 425, and then the image with a low quality evaluation value is deleted from the two images with the highest similarity, and the image with a high quality evaluation value is retained 426, thereby obtaining a target subset.

[0090] Next, a target copy 430 may be generated, and the target copy may include a general title, a general copy, a subtitle, and a subcopy. For example, a general title and a general copy for the target image set may be generated based on a large language model according to the description text of the image in the target image set. For another example, a main image may be determined from the target subset according to at least one of the category and the quality assessment value of the image. Then, a subtitle and a subcopy for the target subset may be generated based on the multimodal large model according to the description text of the main image and other images in the target subset except the main image.

[0091] After the target copy is obtained, data encapsulation may be performed to generate graphic data 440. For example, based on the structure represented by the target structure template, the main title, the main copy, the subtitle, the subcopy and the target subset are combined to obtain graphic data.

[0092] Figure 5 Detailed description of the invention is given in FIG.

[0093] like Figure 5 As shown, in this embodiment, the graphic data Data includes a general title 501, a general text 502, a first subtitle 5031, a first subtext 5041, a first target subset 5051, a second subtitle 5032, a second subtext 5042 and a second target subset 5052, wherein the first target subset 5051 and the second target subset 5052 respectively include a number of grouped images. It can be seen that the target structure template in this example is a time structure template. In the grouping process, a first-level grouping is first performed based on the shooting time, and then a more fine-grained second-level grouping is performed based on the time.

[0094] Figure 6 It is a schematic diagram of the front-end display effect according to an embodiment of the present disclosure.

[0095] like Figure 6As shown, after obtaining the graphic data, the graphic data can be output to the client or other terminal device for display. During the display process, the general title 601 and the general text 602 can be displayed, and then the first subtitle 6031, the first subtext 6041 and the corresponding target subset 6051 are displayed below, and then the second subtitle 6032, the second subtext 6042 and the corresponding target subset 6052 are displayed. It can be seen that in this embodiment, the target structure template is a location structure template, and each target subset is first grouped based on the shooting time, and then second grouped based on the shooting location.

[0096] Figure 7 is a schematic structural block diagram of an image processing device according to an embodiment of the present disclosure.

[0097] like Figure 7 As shown, the image processing device 700 may include: a target structure template determination module 710 , a grouping module 720 , a subset text generation module 730 , and a graphic and text generation module 740 .

[0098] The target structure template determination module 710 is used to determine a target structure template from a plurality of predetermined structure templates according to the number of images in the target image set that meet a predetermined condition.

[0099] The grouping module 720 is used to perform grouping processing on multiple images in the target image set according to the grouping mode used for the target structure template to obtain at least one target subset.

[0100] The subset copy generating module 730 is used to generate a subset copy for each target subset in at least one target subset according to the images in the target subset and the description texts of the images.

[0101] The image and text generation module 740 is used to generate image and text data according to the target structure template, the images in the target image set and the subset text.

[0102] According to another embodiment of the present disclosure, the subset copy generation module includes: a main image determination submodule and a subset copy generation submodule. The main image determination submodule is used to determine the main image from the target subset according to at least one of the category and quality evaluation value of the image. The subset copy generation submodule is used to generate a subtitle and subcopy for the target subset based on the multimodal large model according to the main image and the description text of other images in the target subset except the main image, wherein the subset copy includes the subtitle and the subcopy.

[0103] According to another embodiment of the present disclosure, it also includes: a title copy generation module, which is used to generate a general title and a general copy for the target image set based on the large language model according to the description text of the image in the target image set.

[0104] According to another embodiment of the present disclosure, the graphic and text generation module includes: a graphic and text generation submodule, which is used to combine the main title, main copy, sub-title, sub-copy and target subset based on the structure represented by the target structure template to obtain graphic and text data.

[0105] According to another embodiment of the present disclosure, the grouping module includes: a first grouping submodule, a second grouping submodule and a target subset determination submodule. The first grouping submodule is used to divide the images in the target image set into multiple initial subsets based on different classification dimensions in response to detecting that the target structure template is a multi-level classification template. Among them, the categories of different classification dimensions are different or the precision is different. The second grouping submodule is used to randomly divide the images in the target image set into a predetermined number of initial subsets in response to detecting that the target structure template is a spare structure template. The target subset determination submodule is used to determine at least one target subset based on the initial subset.

[0106] According to another embodiment of the present disclosure, the first grouping submodule includes: a first grouping unit and a second grouping unit. The first grouping unit is used for, in response to detecting that the target structure template is a position structure template in a multi-level classification template, performing a primary grouping of images in the target image set according to the shooting time to obtain at least one subset to be processed. For each subset to be processed, performing a secondary grouping of the images in the subset to be processed according to the shooting position to obtain at least one initial subset. The second grouping unit is used for, in response to detecting that the target structure template is a time structure template in a multi-level classification template, performing a primary grouping of images in the target image set according to the shooting time to obtain at least one subset to be processed; for each subset to be processed in at least one subset to be processed, performing a secondary grouping of the images in the subset to be processed based on time based on the description text of the images in the subset to be processed based on the large language model to obtain at least one initial subset.

[0107] According to another embodiment of the present disclosure, the target subset determination submodule includes: a similarity determination unit, a deletion unit and a target subset determination unit. The similarity determination unit is used to determine, for each initial subset in at least one initial subset, the similarity between any two images in the initial subset when the number of images in the initial subset is greater than a first quantity threshold. The deletion unit is used to delete some images in the initial subset according to the similarity between any two images and the quality evaluation values ​​of each of the any two images. The target subset determination unit is used to determine the target subset according to the initial subset after deleting some images.

[0108] According to another embodiment of the present disclosure, the target structure template determination module includes: a first determination submodule, a second determination submodule and a third determination submodule. The first determination submodule is used to determine the location structure template as the target structure template in response to detecting that the number of images in the target image set that meet the first subcondition in the predetermined condition is greater than or equal to the second quantity threshold; wherein the first subcondition includes that the image corresponds to a geographical location. The second determination submodule is used to determine the time structure template as the target structure template in response to detecting that the number of images in the target image set that meet the first subcondition is less than the second quantity threshold, and the number of images that meet the second subcondition in the predetermined condition is greater than or equal to the third quantity threshold; the second subcondition includes that the shooting time of the image is within the target time period. The third determination submodule is used to determine the backup structure template as the target structure template in response to detecting that the number of images in the target image set that meet the first subcondition is less than the second quantity threshold, and the number of images that meet the second subcondition is less than the third quantity threshold.

[0109] According to another embodiment of the present disclosure, it further includes: a screening module and a target image set determination module. The screening module is used to screen the images in the initial image set according to at least one of the quality evaluation value of the images in the initial image set, the similarity between any two images, and the category of the images. The target image set determination module is used to determine the target image set based on the screened initial image set.

[0110] According to another embodiment of the present disclosure, the screening module includes at least one of the following: a first deletion submodule, a second deletion submodule and a screening submodule. The first deletion submodule is used to delete the image in response to detecting that the quality assessment value of the image in the initial image set is less than or equal to the quality threshold. The second deletion submodule is used to delete the image in response to detecting that the category of the image in the initial image set includes a predetermined category. The screening submodule is used to screen any two images in the initial image set according to the similarity between the two images and the quality assessment values ​​of the two images.

[0111] According to another embodiment of the present disclosure, the screening submodule includes: a deleting unit, configured to delete the second image in response to detecting that the similarity between the two images is greater than or equal to a similarity threshold and the quality assessment value of the first image of the two images is higher than the quality assessment value of the second image.

[0112] According to another embodiment of the present disclosure, the description text is determined according to at least one of the shooting time, shooting location, and category of the image.

[0113] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, including at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned image processing method.

[0114] According to an embodiment of the present disclosure, the present disclosure further provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute the above-mentioned image processing method.

[0115] According to an embodiment of the present disclosure, the present disclosure further provides a computer program product, including a computer program, and the computer program implements the above-mentioned image processing method when executed by a processor.

[0116] Figure 8 1 is a block diagram of an electronic device for implementing the image processing method of an embodiment of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0117] like Figure 8 As shown, the device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 to a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0118] A number of components in the device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0119] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as image processing methods. For example, in some embodiments, the image processing method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the image processing method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform the image processing method in any other appropriate manner (e.g., by means of firmware).

[0120] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0121] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0122] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0123] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0124] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0125] A computer system may include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship to each other.

[0126] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0127] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. An image processing method, comprising: Determining a target structure template from a plurality of predetermined structure templates according to the number of images in the target image set that meet a predetermined condition; performing grouping processing on a plurality of images in the target image set according to a grouping mode for the target structure template to obtain at least one target subset; For each target subset in the at least one target subset, generating a subset copy according to the images in the target subset and the description texts of the images; as well as Graphic and text data are generated according to the target structure template, the images in the target image set and the subset text.

2. The method according to claim 1, wherein: The generating of subset text according to the images and the description texts of the images in the target subset comprises: determining a main image from the target subset according to at least one of a category and a quality assessment value of the image; and According to the main image and the description texts of other images in the target subset except the main image, a subtitle and subcopy for the target subset are generated based on a multimodal large model; wherein the subset copy includes the subtitle and the subcopy.

3. The method according to claim 2, further comprising: According to the description texts of the images in the target image set, a general title and a general copy for the target image set are generated based on a large language model.

4. The method according to claim 3, wherein: The generating of graphic and text data according to the target structure template, the image in the target image set and the subset text comprises: Based on the structure represented by the target structure template, the general title, the general text, the subtitle, the subtext and the target subset are combined to obtain the graphic data.

5. The method according to claim 1, wherein: The step of grouping the plurality of images in the target image set according to the grouping mode for the target structure template to obtain at least one target subset includes: In response to detecting that the target structure template is a multi-level classification template, dividing the images in the target image set into a plurality of initial subsets based on different classification dimensions; wherein the different classification dimensions have different categories or different precisions; and In response to detecting that the target structure template is a spare structure template, randomly dividing the images in the target image set into a predetermined number of initial subsets; and The at least one target subset is determined based on the initial subset.

6. The method according to claim 5, wherein: In response to detecting that the target structure template is a multi-level classification template, dividing the images in the target image set into a plurality of initial subsets based on different classification dimensions comprises: In response to detecting that the target structure template is a position structure template in the multi-level classification template, the images in the target image set are grouped in a primary manner according to the shooting time to obtain at least one subset to be processed; for each subset to be processed, the images in the subset to be processed are grouped in a secondary manner according to the shooting position to obtain at least one initial subset; In response to detecting that the target structure template is a time structure template in the multi-level classification template, the images in the target image set are grouped at the first level according to the shooting time to obtain at least one subset to be processed; for each of the at least one subset to be processed, the images in the subset to be processed are grouped at the second level based on time based on a large language model according to the description text of the images in the subset to be processed to obtain the at least one initial subset.

7. The method according to claim 5, wherein: Determining the at least one target subset according to the initial subset includes: For each initial subset in the at least one initial subset, when the number of images in the initial subset is greater than a first number threshold, determining the similarity between any two images in the initial subset; Deleting some images in the initial subset according to the similarity between the arbitrary two images and the quality evaluation values ​​of the arbitrary two images; and The target subset is determined according to the initial subset after deleting part of the images.

8. The method according to claim 1, wherein: Determining the target structure template from a plurality of predetermined structure templates according to the number of images in the target image set that meet the predetermined condition comprises: In response to detecting that the number of images in the target image set that meet the first sub-condition of the predetermined condition is greater than or equal to a second number threshold, determining the location structure template as the target structure template; wherein the first sub-condition includes that the image corresponds to a geographical location; In response to detecting that the number of images in the target image set that meet the first condition is less than the second number threshold, and the number of images that meet the second sub-condition in the predetermined condition is greater than or equal to a third number threshold, determining the time structure template as the target structure template; the second sub-condition includes that the shooting time of the image is within the target time period; and In response to detecting that the number of images in the target image set that meet the first sub-condition is less than the second number threshold, and the number of images that meet the second sub-condition is less than the third number threshold, a backup structure template is determined as the target structure template.

9. The method according to claim 1, further comprising: Performing screening processing on the images in the initial image set according to at least one of a quality evaluation value of the images in the initial image set, a similarity between any two images, and a category of the images; as well as The target image set is determined according to the screened initial image set.

10. The method according to claim 9, wherein: The screening process of the images in the initial image set according to at least one of the quality evaluation value of the images in the initial image set, the similarity between any two images, and the category of the images comprises at least one of the following: In response to detecting that the quality assessment value of an image in the initial image set is less than or equal to a quality threshold, deleting the image; In response to detecting that the category of an image in the initial set of images includes a predetermined category, deleting the image; as well as For any two images in the initial image set, screening processing is performed on the two images according to the similarity between the two images and the quality evaluation values ​​of the two images respectively.

11. The method according to claim 10, wherein: The screening process of the two images according to the similarity between the two images and the quality evaluation values ​​of the two images comprises: In response to detecting that the similarity between the two images is greater than or equal to a similarity threshold, and a quality evaluation value of a first image of the two images is higher than a quality evaluation value of a second image, the second image is deleted.

12. The method according to claim 1, wherein: The description text is determined according to at least one of the shooting time, shooting location, and category of the image.

13. An image processing device, comprising: A target structure template determination module, used to determine a target structure template from a plurality of predetermined structure templates according to the number of images in a target image set that meet predetermined conditions; A grouping module, used for grouping a plurality of images in the target image set according to a grouping mode used for the target structure template to obtain at least one target subset; A subset copy generation module, configured to generate a subset copy for each target subset in the at least one target subset according to the images and the description texts of the images in the target subset; as well as The image and text generation module is used to generate image and text data according to the target structure template, the images in the target image set and the subset text.

14. The device according to claim 13, wherein: The subset copywriting generation module includes: a main image determination submodule, configured to determine a main image from the target subset according to at least one of a category and a quality assessment value of the image; and A subset copy generation submodule is used to generate a subtitle and subcopy for the target subset based on the main image and the description text of other images in the target subset except the main image, based on the multimodal large model; wherein the subset copy includes the subtitle and the subcopy.

15. The apparatus according to claim 14, further comprising: A title and copywriting generation module is used to generate a general title and a general copy for the target image set based on the description text of the image in the target image set and based on a large language model.

16. The device according to claim 15, wherein: The image and text generation module includes: The graphic and text generation submodule is used to combine the general title, the general text, the subtitle, the subtext and the target subset based on the structure represented by the target structure template to obtain the graphic and text data.

17. The device according to claim 13, wherein: The grouping module comprises: A first grouping submodule is configured to, in response to detecting that the target structure template is a multi-level classification template, divide the images in the target image set into a plurality of initial subsets based on different classification dimensions; wherein the different classification dimensions have different categories or different precisions; and A second grouping submodule is configured to randomly divide the images in the target image set into a predetermined number of initial subsets in response to detecting that the target structure template is a spare structure template; and The target subset determination submodule is used to determine the at least one target subset according to the initial subset.

18. The device according to claim 17, wherein: The first grouping submodule comprises: A first grouping unit is configured to, in response to detecting that the target structure template is a position structure template in the multi-level classification template, perform a primary grouping of images in the target image set according to shooting time to obtain at least one subset to be processed; and for each subset to be processed, perform a secondary grouping of the images in the subset to be processed according to the shooting position to obtain at least one initial subset; The second grouping unit is used for, in response to detecting that the target structure template is a time structure template in the multi-level classification template, performing a primary grouping of the images in the target image set according to the shooting time to obtain at least one subset to be processed; for each of the at least one subset to be processed, performing a time-based secondary grouping of the images in the subset to be processed based on the large language model according to the description text of the images in the subset to be processed to obtain the at least one initial subset.

19. The device according to claim 17, wherein: The target subset determination submodule includes: A similarity determination unit is used to determine, for each initial subset in the at least one initial subset, a similarity between any two images in the initial subset when the number of images in the initial subset is greater than a first number threshold; a deleting unit, configured to delete some images in the initial subset according to the similarity between the arbitrary two images and the quality evaluation values ​​of the arbitrary two images; and The target subset determining unit is used to determine the target subset according to the initial subset after deleting part of the image.

20. The device according to claim 13, wherein: The target structure template determination module comprises: A first determination submodule is configured to determine the location structure template as the target structure template in response to detecting that the number of images in the target image set that meet the first sub-condition in the predetermined condition is greater than or equal to a second number threshold; wherein the first sub-condition includes that the image corresponds to a geographical location; A second determination submodule is configured to determine the time structure template as the target structure template in response to detecting that the number of images in the target image set that meet the first sub-condition is less than the second number threshold, and the number of images that meet the second sub-condition in the predetermined condition is greater than or equal to a third number threshold; the second sub-condition includes that the shooting time of the image is within a target time period; and The third determination submodule is used to determine the spare structure template as the target structure template in response to detecting that the number of images in the target image set that meet the first sub-condition is less than the second number threshold, and the number of images that meet the second sub-condition is less than the third number threshold.

21. The apparatus according to claim 13, further comprising: A screening module, configured to screen the images in the initial image set according to at least one of a quality evaluation value of the images in the initial image set, a similarity between any two images, and a category of the images; as well as The target image set determination module is used to determine the target image set according to the screened initial image set.

22. The device according to claim 21, wherein The screening module includes at least one of the following: A first deleting submodule, configured to delete the image in response to detecting that the quality assessment value of the image in the initial image set is less than or equal to a quality threshold; a second deleting submodule, configured to delete the image in response to detecting that the category of the image in the initial image set includes a predetermined category; as well as The screening submodule is used to screen any two images in the initial image set according to the similarity between the two images and the quality assessment values ​​of the two images.

23. The device according to claim 22, wherein: The screening submodule comprises: The deleting unit is configured to delete the second image in response to detecting that the similarity between the two images is greater than or equal to a similarity threshold and the quality evaluation value of the first image of the two images is higher than the quality evaluation value of the second image.

24. The device according to claim 13, wherein: The description text is determined according to at least one of the shooting time, shooting location, and category of the image.

25. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 12.

26. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 12.

27. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 12.

Images