Photographing recommendation method, apparatus, device, and storage medium
By generating photo recommendations using a deep learning-based neural network model, the system provides outlines of scenery and people, as well as movement direction hints. This solves the problems of interactive complexity and intuitiveness in existing mobile phone photo recommendation functions, enabling convenient shooting of high-quality photos.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN CHINO E COMM CO LTD
- Filing Date
- 2024-11-28
- Publication Date
- 2026-06-09
AI Technical Summary
Existing commercial mobile phone photography recommendation functions require frequent user interaction, the recommendation prompt interface is not intuitive, and it is difficult for non-professional users to take high-quality photos.
It uses a deep learning-based neural network model to generate outlines of objects and people in the preview image and provides movement direction prompts to guide users in adjusting the shooting position and angle.
It improves the convenience and accuracy of users taking high-quality photos, reduces user interaction, and enhances the intuitiveness of photo recommendations.
Smart Images
Figure CN119922412B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of image processing technology, and in particular to a method, apparatus, device, and storage medium for recommending photos. Background Technology
[0002] With the development of smartphone photography technology and the popularity of social media, people often use their phones to take photos to record life moments, share on social networks, and capture travel scenery.
[0003] However, despite modern mobile phones being equipped with high-resolution cameras and advanced image processing technology, existing commercial photo recommendation functions have many problems, such as requiring frequent user interaction during the recommendation process and having unintuitive recommendation prompts. Non-professional users still face challenges when taking high-quality photos. Summary of the Invention
[0004] Based on this, the purpose of the present invention is to provide a photo recommendation method, apparatus, device, and storage medium, which obtains the corresponding scene outline map and person outline map of the preview image based on a pre-trained photo recommendation model, and generates movement direction prompt information, which is displayed on a preset photo interface to prompt the user to make photo adjustments and meet the user's need for high-quality photo taking.
[0005] Firstly, embodiments of this application provide a photo recommendation method, including the following steps:
[0006] Obtain a preview image and its image annotation data; encode the image annotation data of the preview image to obtain the image annotation encoding representation of the preview image;
[0007] The image annotation encoding of the preview image is input into a preset image recommendation model to generate an image and obtain a recommended image corresponding to the preview image. The image recommendation model is a neural network model based on deep learning.
[0008] The recommended image corresponding to the preview image is segmented into scenery and people to obtain the scenery outline map and the people outline map.
[0009] A real-time image is obtained. Based on the comparison between the real-time image, the outline of the scene, and the outline of the person, movement direction prompt information is obtained. The outline of the scene, the outline of the person, and the movement direction prompt information are displayed on a preset shooting interface. The movement direction prompt information is used to indicate the shooting position and shooting angle.
[0010] Secondly, embodiments of this application provide a photo recommendation device, including:
[0011] The image acquisition module is used to acquire a preview image and the image annotation data of the preview image, and to encode the image annotation data of the preview image to obtain the image annotation encoding representation of the preview image;
[0012] The image generation module is used to input the photo annotation encoding representation of the preview image into a preset photo recommendation model to generate an image and obtain a recommended image corresponding to the preview image. The photo recommendation model is a neural network model based on deep learning.
[0013] The image segmentation module is used to segment the scene and people in the recommended image corresponding to the preview image to obtain the scene outline map and the person outline map.
[0014] The prompt information display module is used to obtain real-time images, compare the real-time images with the outlines of objects and people, obtain movement direction prompt information, and display the outlines of objects and people and the movement direction prompt information on a preset shooting interface. The movement direction prompt information is used to indicate the shooting position and shooting angle.
[0015] Thirdly, embodiments of this application provide a computer device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor; when the computer program is executed by the processor, it implements the steps of the photo recommendation method as described in the first aspect.
[0016] Fourthly, embodiments of this application provide a storage medium storing a computer program that, when executed by a processor, implements the steps of the photo recommendation method as described in the first aspect.
[0017] In this application embodiment, a photo recommendation method, apparatus, device, and storage medium are provided. Based on a pre-trained photo recommendation model, the method obtains the outline of objects and the outline of people corresponding to the preview image, and generates movement direction prompts, which are displayed on a preset photo interface to prompt the user to make adjustments to the photo and meet the user's need for high-quality photos.
[0018] To better understand and implement this invention, the following detailed description is provided in conjunction with the accompanying drawings. Attached Figure Description
[0019] Figure 1 A schematic flowchart illustrating the photo recommendation method provided in the first embodiment of this application;
[0020] Figure 2 A flowchart illustrating the photo recommendation method provided in the second embodiment of this application;
[0021] Figure 3 This is a flowchart illustrating step S6 of the photo recommendation method provided in the third embodiment of this application.
[0022] Figure 4 This is a flowchart illustrating step S6 of the photo recommendation method provided in the fourth embodiment of this application;
[0023] Figure 5 A schematic flowchart illustrating the photo recommendation method provided in the fifth embodiment of this application;
[0024] Figure 6 This is a schematic diagram of the photograph recommendation device provided in the sixth embodiment of this application;
[0025] Figure 7 This is a schematic diagram of the structure of a computer device provided in the seventh embodiment of this application. Detailed Implementation
[0026] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.
[0027] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.
[0028] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."
[0029] Please see Figure 1 , Figure 1 The flowchart of the photo recommendation method provided in the first embodiment of this application is shown. The method includes the following steps:
[0030] S1: Obtain a preview image and the image annotation data of the preview image, encode the image annotation data of the preview image, and obtain the image annotation encoding representation of the preview image.
[0031] The execution entity of the photo recommendation method is the recommendation device (hereinafter referred to as the recommendation device). In an optional embodiment, the recommendation device may be a computer device, a server, or a server cluster composed of multiple computer devices.
[0032] In this embodiment, the recommending device can obtain a preview image input by the user, or it can obtain a preview image sent by the camera terminal through a data transmission connection. The camera terminal can be a device with camera functionality, such as a mobile phone, tablet, or interactive whiteboard. The preview image is generated by the user taking a photo through the preview interface of the camera terminal.
[0033] The recommended device obtains the image annotation data of the preview image. The image annotation data is constructed by the camera terminal synchronously recording relevant information of the preview image through a sensor device or a public API interface when generating the preview image. The image annotation data includes information on people, scenery, and environment. The information on people includes the number of people, skin color, clothing, and height of each person. The information on scenery includes background building information and building angle information. The information on environment includes weather, sunlight, light, humidity, temperature, time of day, and geographical location information.
[0034] The recommended device can use one-hot encoding or numerical encoding to encode the image annotation data of the preview image to obtain the image annotation encoding representation of the preview image.
[0035] S2: Input the photo annotation code of the preview image into a preset photo recommendation model to generate an image and obtain a recommended image corresponding to the preview image.
[0036] The photo recommendation model is a neural network model based on deep learning.
[0037] In this embodiment, the recommendation device inputs the photo annotation data of the preview image into a preset photo recommendation model for image generation. The photo recommendation model generates the best shooting position and angle image based on the geographical location information and specific scenic spots in the preview image and the photo annotation code representation, as well as the person information, scenery information and environmental information in the photo annotation code representation. The recommended image corresponding to the preview image is obtained, wherein the recommended image contains the best order and position for taking photos of each person.
[0038] In an optional embodiment, step S5 is further included: training the neural network model to be trained. (See [link to previous document]). Figure 2 , Figure 2 The flowchart of step S5 in the photo recommendation method provided in the second embodiment of this application is shown below, including steps S51 to S52:
[0039] S51: Obtain several test images and the image annotation codes of the test images.
[0040] In this embodiment, the recommended device obtains several test images and several photographic annotation codes representing the test images.
[0041] S52: Input several test images and their image annotation encodings into the image recommendation model to be trained, and use an attention mechanism to extract feature vectors from the test images to obtain image feature representations of the test images.
[0042] In this embodiment, the recommendation device inputs several test images and their image annotation codes into the image recommendation model to be trained. It then uses an attention mechanism to extract feature vectors from the test images to obtain image feature representations of the test images, thereby capturing information such as color, texture, and shape of the images and improving the accuracy of image generation.
[0043] S53: The image feature representation of the test image is concatenated with the corresponding photo annotation encoding representation to obtain a concatenated representation of several test images; an unsupervised training method is used to train the photo recommendation model to be trained based on the concatenated representation of several test images to obtain the photo recommendation model.
[0044] In this embodiment, the recommendation device concatenates the image feature representation of the test image with the corresponding photo annotation encoding representation to obtain a concatenated representation of several test images. An unsupervised training method is used to train the photo recommendation model based on the concatenated representations of the several test images to obtain the photo recommendation model. This enables the photo recommendation model to pay more attention to important information, to pinpoint specific scenic spots based on the geographical location information in the test images and photo annotation encoding representations, and to comprehensively analyze and generate the optimal shooting position and angle image based on the person, scenery, and environmental information in the photo annotation encoding representations.
[0045] In an optional embodiment, step S6 is further included: adjusting the model parameters of the photo recommendation model. (See [link to relevant documentation]). Figure 3 , Figure 3The flowchart of step S6 in the photo recommendation method provided in the third embodiment of this application is shown below, including steps S61 to S62:
[0046] S61: Obtain the photo annotation encoding representation of several verification images, input the photo annotation encoding representation of several verification images into the photo recommendation model to generate images, and obtain the recommended images corresponding to several verification images and the image feature representation of the recommended images.
[0047] In this embodiment, the recommendation device obtains the photo annotation code representations of several verification images, inputs the photo annotation code representations corresponding to the several verification images into the photo recommendation model to generate images, and obtains the recommended images corresponding to the several verification images and the image feature representations of the recommended images.
[0048] S62: Using a consistency calculation method, consistency calculation is performed based on the photo annotation code representations of several verification images and the photo annotation code representations of the corresponding recommended images to obtain the consistency between the several verification images and the corresponding recommended images. Based on the consistency between the several verification images and the corresponding recommended images, the model parameters of the photo recommendation model are adjusted.
[0049] In this embodiment, the recommendation device employs a consistency calculation method. It performs consistency calculations based on the image annotation codes of several verification images and the corresponding recommendation images to obtain the consistency between the verification images and the corresponding recommendation images. Specifically, the consistency calculation method can be the Kappa test, the ICC intragroup correlation coefficient calculation method, or the Kendall concordance coefficient calculation method.
[0050] Based on the consistency between several verification images and the corresponding recommended images, the model parameters of the photo recommendation model are adjusted to adjust the degree of emphasis the model places on important information, thereby improving the accuracy of the model in generating recommended images.
[0051] Please see Figure 4 , Figure 4 The flowchart of step S6 in the photo recommendation method provided in the fourth embodiment of this application is shown below. It also includes steps S63 to S64, as follows:
[0052] S63: Obtain image feature representations of several verification images, input the photo annotation encoding representations corresponding to several verification images into the photo recommendation model for image generation, and obtain recommended images corresponding to several verification images and image feature representations of the recommended images.
[0053] In this embodiment, the recommendation device obtains image feature representations of several verification images, inputs the photo annotation encoding representations corresponding to several verification images into the photo recommendation model for image generation, and obtains recommended images corresponding to several verification images and image feature representations of the recommended images.
[0054] S64: Using a similarity calculation method, similarity is calculated based on the image feature representations of several verification images and the image feature representations of the corresponding recommended images to obtain the similarity between the several verification images and the corresponding recommended images. Based on the similarity between the several verification images and the corresponding recommended images, the model parameters of the photo recommendation model are adjusted.
[0055] In this embodiment, the recommendation device employs a similarity calculation method. Based on the image feature representations of several verification images and the corresponding image feature representations of the recommended images, a similarity score is calculated to obtain the similarity between the verification images and the corresponding recommended images. Specifically, the similarity calculation method may be a structural similarity index (SSIM) calculation method and a peak signal-to-noise ratio (PSNR) calculation method.
[0056] The recommendation device adjusts the model parameters of the photo recommendation model based on the similarity between several verification images and corresponding recommended images, so as to further adjust the model's emphasis on important information and thus improve the accuracy of the model in generating recommended images.
[0057] S3: Perform scene and person segmentation on the recommended image corresponding to the preview image to obtain the scene outline map and the person outline map.
[0058] In this embodiment, the recommendation device uses an image segmentation model to segment the scene and people in the recommended image corresponding to the preview image according to the category segmentation method, and obtains the scene outline map and the person outline map. The selection can be made by category or by selecting one by one to determine the scene outline map and the person outline map in the recommended image.
[0059] S4: Obtain a real-time image, compare the real-time image, the outline of the scene, and the outline of the person to obtain movement direction prompts, and display the outline of the scene, the outline of the person, and the movement direction prompts on a preset shooting interface.
[0060] In this embodiment, the recommended device obtains a real-time image sent by the camera terminal, wherein the real-time image is an image acquired by the camera terminal in real time on the camera interface.
[0061] The recommended device compares the preview image, the outline of the scenery, and the outline of the person to obtain movement direction prompts. These prompts indicate the shooting location and angle. Specifically, the recommended device compares the scenery area in the real-time image with the outline of the scenery to obtain movement direction prompts for the scenery, and compares the outline of the person area in the real-time image with the outline of the person to obtain movement direction prompts for the person.
[0062] The recommendation device sends location information for both objects and people to the camera terminal. The camera terminal then displays outlines of the objects and people, along with the location information, on its preset camera interface. Specifically, the location information can be a simple arrow or a location description. Based on a pre-trained photo recommendation model, the device obtains the corresponding outlines of objects and people from the preview image and generates location information, which is then displayed on the preset camera interface to prompt the user to make adjustments and meet their needs for high-quality photos.
[0063] In one optional embodiment, please refer to Figure 5 , Figure 5 The flowchart of the photo recommendation method provided in the fifth embodiment of this application is shown, including step S7, which is after step S4, and is specifically as follows:
[0064] S7: Determine whether the scenery and people in the real-time image match the outline of the scenery and the outline of the people, and control the camera terminal to take a picture.
[0065] The judgment result includes a correct judgment result and a failed judgment result.
[0066] In this embodiment, the recommended device determines whether the scenery and people in the real-time image match the scenery outline map and the people outline map. If the scenery area in the real-time image matches the scenery outline map and the people area in the real-time image matches the people outline map, a correct judgment result is obtained; otherwise, a failure judgment result is obtained.
[0067] If the judgment result is correct, it is recommended that the device send a photo-taking command to the photo-taking terminal. The photo-taking terminal responds to the photo-taking command and performs the photo-taking operation to control the photo-taking terminal to achieve convenient and accurate acquisition of the best photos and meet the user's high-quality photo-taking needs.
[0068] Please refer to Figure 6 , Figure 6This is a schematic diagram of the structure of the photo recommendation device provided in the sixth embodiment of this application. The device can be implemented entirely or partially through software, hardware, or a combination of both. The device 6 includes:
[0069] The image acquisition module 61 is used to acquire a preview image and the image annotation data of the preview image, and to encode the image annotation data of the preview image to obtain the image annotation encoding representation of the preview image;
[0070] Image generation module 62 is used to input the photo annotation encoding representation of the preview image into a preset photo recommendation model to generate an image and obtain a recommended image corresponding to the preview image, wherein the photo recommendation model is a neural network model based on deep learning;
[0071] Image segmentation module 63 is used to segment the scene and people in the recommended image corresponding to the preview image to obtain the scene outline map and the person outline map.
[0072] The prompt information display module 64 is used to obtain a real-time image, compare the real-time image with the outline of the scene and the outline of the person to obtain movement direction prompt information, and display the outline of the scene, the outline of the person and the movement direction prompt information on a preset shooting interface. The movement direction prompt information is used to indicate the shooting position and shooting angle.
[0073] In this embodiment, an image acquisition module obtains a preview image and its image annotation data, encodes the image annotation data to obtain an image annotation encoding representation of the preview image, and then uses an image generation module to input the image annotation encoding representation of the preview image into a preset image recommendation model to generate a recommended image corresponding to the preview image. The image recommendation model is a deep learning-based neural network model. An image segmentation module segments the recommended image corresponding to the preview image into scenery and people, obtaining scenery outlines and people outlines. A prompt information display module obtains a real-time image, compares the real-time image, scenery outlines, and people outlines to obtain movement direction prompts, and displays the scenery outlines, people outlines, and movement direction prompts on a preset camera interface. The movement direction prompts indicate the camera's movement position and angle. Based on a pre-trained photo recommendation model, the system obtains the outlines of objects and people corresponding to the preview image, and generates movement direction prompts, which are displayed on the preset photo interface to prompt users to make adjustments and meet their needs for high-quality photos.
[0074] Please refer to Figure 7 , Figure 7 This is a schematic diagram of the structure of a computer device provided in the seventh embodiment of this application. The computer device 7 includes: a processor 71, a memory 72, and a computer program 73 stored in the memory 72 and executable on the processor 71. The computer device can store multiple instructions, which are applicable to the method steps shown in the first to fifth embodiments above being loaded and executed by the processor 71. For the specific execution process, please refer to the specific descriptions shown in the first to fifth embodiments, which will not be repeated here.
[0075] The processor 71 may include one or more processing cores. The processor 71 connects to various parts of the server using various interfaces and lines, and executes various functions and processes data of the photo-recommendation device 6 by running or executing instructions, programs, code sets, or instruction sets stored in the memory 72, and by calling data stored in the memory 72. Optionally, the processor 71 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 71 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required to be displayed on the touch screen; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 71 and may be implemented as a separate chip.
[0076] The memory 72 may include random access memory (RAM) or read-only memory. Optionally, the memory 72 may include a non-transitory computer-readable storage medium. The memory 72 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 72 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch instructions), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data involved in the various method embodiments described above, etc. Optionally, the memory 72 may also be at least one storage device located remotely from the aforementioned processor 71.
[0077] This application also provides a storage medium that can store multiple instructions. These instructions are applicable to being loaded and executed by a processor using the method steps shown in the first to fifth embodiments described above. For details of the execution process, please refer to the specific descriptions shown in the first to fifth embodiments, which will not be repeated here.
[0078] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0079] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.
[0080] Those skilled in the art will recognize that the units and algorithm steps of the various examples 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 implemented in hardware or software depends on the specific application and design constraints of the algorithm. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0081] In the embodiments provided by this invention, it should be understood that the disclosed apparatus / terminal devices and methods can be implemented in other ways. For example, the apparatus / terminal device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.
[0082] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0083] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0084] If the integrated module / unit is implemented as 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, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms.
[0085] This invention is not limited to the above-described embodiments. If any modifications or variations to this invention do not depart from the spirit and scope of this invention, and if such modifications and variations fall within the scope of the claims and equivalent technologies of this invention, then this invention also intends to include such modifications and variations.
Claims
1. A photo recommendation method, characterized in that, Includes the following steps: A preview image and its image annotation data are obtained. The image annotation data of the preview image is encoded to obtain the image annotation encoding representation of the preview image. The image annotation data includes information about people, scenery, and environment. The environment information includes geographical location information. The image's photographic annotation encoding is input into a preset image recommendation model. The model identifies the scenic spot based on the geographic location information in the image and the annotation encoding. It then performs a comprehensive analysis based on the scenic spot and the information about people, scenery, and the environment in the annotation encoding to generate the optimal shooting position and angle image, thus generating a recommended image corresponding to the preview image. The image recommendation model is a deep learning-based neural network model. The recommended image includes scenery, people, and the optimal order and position for taking photos of people. The recommended image corresponding to the preview image is segmented into scenery and people to obtain scenery outline map and person outline map; A real-time image is obtained. Based on the comparison between the real-time image, the outline of the scene, and the outline of the person, movement direction prompt information is obtained. The outline of the scene, the outline of the person, and the movement direction prompt information are displayed on the preset shooting interface. The movement direction prompt information is used to indicate the shooting position and shooting angle. The preset photo recommendation model is obtained as follows: Obtain several test images and their photographic annotation codes; Several test images and their image annotation codes are input into the image recommendation model to be trained. An attention mechanism is used to extract feature vectors from the test images to obtain image feature representations of the test images. The image feature representations include color, texture, and shape information. The image feature representation of the test image is concatenated with the corresponding photo annotation encoding representation to obtain a concatenated representation of several test images. An unsupervised training method is used to train the photo recommendation model to be trained based on the concatenated representation of several test images to obtain the photo recommendation model. The model can identify specific scenic spots based on the geographical location information in the test images and photo annotation encoding representations, and generate the best shooting position and angle image based on the comprehensive analysis of the person information, scenery information and environmental information in the photo annotation encoding representations. A number of verification images are obtained by taking photos and labeling them with encoding. The photos and labeling them with encoding are then input into the photo recommendation model to generate images. The result is a number of recommended images corresponding to the verification images, the image feature representations of the recommended images, and the photos and labeling them with encoding. A consistency calculation method is used to calculate the consistency between the verification images and the corresponding recommended images based on the image feature representations of the verification images and the corresponding recommended images. A similarity calculation method is used to calculate the similarity between the verification images and the corresponding recommended images based on the image feature representations of the verification images and the corresponding recommended images. Based on the consistency between the verification images and the corresponding recommended images, and the similarity between the verification images and the corresponding recommended images, the model parameters of the photo recommendation model are adjusted to obtain the photo recommendation model.
2. The photo recommendation method according to claim 1, characterized in that: The photo annotation data includes information on people, scenery, and environment. The information on people includes the number of people, each person's skin color, clothing, and height. The information on scenery includes background building information and building angle information. The information on environment also includes weather, sunlight, light, humidity, temperature, and time of day.
3. The photo recommendation method according to claim 2, characterized in that, It also includes the following steps: The system determines whether the objects and people in the real-time image match the outlines of the objects and people, and then controls the camera terminal to take a picture.
4. A photo recommendation device, characterized in that, include: An image acquisition module is used to acquire a preview image and its image annotation data, and to encode the image annotation data of the preview image to obtain an image annotation encoding representation of the preview image; wherein, the image annotation data includes information about people, scenery, and environment; the environment information includes geographical location information; An image generation module is used to input the photo annotation code representation of the preview image into a preset photo recommendation model. The photo recommendation model locates the scenic spot based on the geographical location information in the preview image and the photo annotation code representation. Based on the scenic spot and the information on people, scenery, and environment in the photo annotation code representation, it performs a comprehensive analysis to generate the best shooting position and angle image, and generates a recommended image corresponding to the preview image. The photo recommendation model is a neural network model based on deep learning. The recommended image includes scenery, people, and the best order and position for taking photos of people. The image segmentation module is used to segment the scene and people in the recommended image corresponding to the preview image to obtain the scene outline map and the person outline map. The prompt information display module is used to obtain real-time images, compare the real-time images, the outline of the scenery, and the outline of the person to obtain movement direction prompt information, and display the outline of the scenery, the outline of the person, and the movement direction prompt information on a preset shooting interface. The movement direction prompt information is used to indicate the shooting position and shooting angle. The preset photo recommendation model is obtained as follows: Obtain several test images and their photographic annotation codes; Several test images and their image annotation codes are input into the image recommendation model to be trained. An attention mechanism is used to extract feature vectors from the test images to obtain image feature representations of the test images. The image feature representations include color, texture, and shape information. The image feature representation of the test image is concatenated with the corresponding photo annotation encoding representation to obtain a concatenated representation of several test images. An unsupervised training method is used to train the photo recommendation model to be trained based on the concatenated representation of several test images to obtain the photo recommendation model. The model can identify specific scenic spots based on the geographical location information in the test images and photo annotation encoding representations, and generate the best shooting position and angle image based on the comprehensive analysis of the person information, scenery information and environmental information in the photo annotation encoding representations. A number of verification images are obtained by taking photos and labeling them with encoding. The photos and labeling them with encoding are then input into the photo recommendation model to generate images. The result is a number of recommended images corresponding to the verification images, the image feature representations of the recommended images, and the photos and labeling them with encoding. A consistency calculation method is used to calculate the consistency between the verification images and the corresponding recommended images based on the image feature representations of the verification images and the corresponding recommended images. A similarity calculation method is used to calculate the similarity between the verification images and the corresponding recommended images based on the image feature representations of the verification images and the corresponding recommended images. Based on the consistency between the verification images and the corresponding recommended images, and the similarity between the verification images and the corresponding recommended images, the model parameters of the photo recommendation model are adjusted to obtain the photo recommendation model.
5. A computer device, characterized in that, include: A processor, a memory, and a computer program stored in the memory and executable on the processor; the computer program, when executed by the processor, implements the steps of the photo recommendation method as described in any one of claims 1 to 3.
6. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, implements the steps of the photo recommendation method as described in any one of claims 1 to 3.