Image acquisition method and device

Abstract

The present application discloses an image acquisition method, comprising: obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain the attribute information of the target image to be obtained; obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image contains the initial layout information and visual information of the target image to be obtained; obtaining the target image according to the first completion image, wherein the target image includes at least one layer, and the content in the at least one layer corresponds to the target object or the target constraint information. The method can obtain a target image that meets the user's expectations in a targeted and convenient manner by fitting the design method of the user's offline design image.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an image acquisition method, device, electronic device and storage device. Background Art

[0002] With the continuous development of Internet technology, especially e-commerce technology, electronic transactions through e-commerce platforms have gradually become the mainstream trading method for people.

[0003] At present, when users engage in electronic transactions on e-commerce platforms, they usually need to conduct e-commerce operations, that is, users operate their stores on the e-commerce platform to increase the transaction volume of the store or the store's goods. Usually, one of the main means for users to conduct e-commerce operations is through designing and placing promotional images (Banner). Currently, promotional images are generally obtained in the following ways: 1. Through relevant image processing software design; 2. By selecting the corresponding template on the image design platform or material platform and redesigning the template.

[0004] The first method requires manual design, which is time-consuming and labor-intensive, and has a relatively high communication cost; while the second method is more convenient and labor-saving than the manual method, but it is highly dependent on templates and also cannot obtain the promotional images that users expect to obtain in a targeted and convenient manner. Therefore, the prior art has the problem of not being able to obtain the promotional images that users expect to obtain in a targeted and convenient manner.

[0005] The above uses the example of a user obtaining promotional images required for e-commerce operations to illustrate the problems existing in the prior art; it should be noted that in real life, when it is necessary to obtain other images besides promotional images, the above problems also exist. Summary of the invention

[0006] The embodiment of the present application provides an image acquisition method to solve the problem in the prior art that an image desired by a user cannot be acquired in a targeted and convenient manner.

[0007] The present application provides an image acquisition method, comprising: obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained; obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image contains initial layout information and visual information of the target image to be obtained; obtaining the target image according to the first completion image, wherein the target image includes at least one layer, and the content in the at least one layer corresponds to the target object or the target constraint information.

[0008] Optionally, obtaining a first completion image based on the original image and the target constraint information includes: normalizing the original image and the target constraint information to obtain normalized preprocessing information; inputting the preprocessing information into a planning model to obtain the first completion image, wherein the planning model is used to obtain at least one layer information based on the preprocessing information, and perform layout planning and visual planning on the layer corresponding to the at least one layer information to obtain the sequence relationship and position information between the layers corresponding to the at least one layer information, and the layer information includes layout information and visual information of the layer corresponding to the layer information.

[0009] Optionally, the original image and the target constraint information are normalized to obtain normalized preprocessing information, including: obtaining feature identification information for identifying the target object based on the original image; obtaining target constraint identification information for identifying the target constraint information; merging information in the feature identification information that is similar to the target constraint identification information, and encoding the merged feature identification information and the target constraint identification information to obtain the preprocessing information.

[0010] Optionally, obtaining feature identification information for identifying the target object based on the original image includes: acquiring the feature information of the target object contained in the original image; obtaining the cluster center point of the feature information by performing cluster analysis on the feature information, and using the identification information of the cluster center point as the feature identification information.

[0011] Optionally, inputting the normalized preprocessing information into the planning model to obtain the first completion image includes: acquiring at least one layer information corresponding to the preprocessing information according to the planning model; and performing layout planning and visual planning on the at least one layer information by using the planning model to obtain the first completion image.

[0012] Optionally, obtaining at least one layer information corresponding to the preprocessing information according to the planning model includes: obtaining the target identification information from the preprocessing information; performing predictive analysis on the preprocessing information using the planning model to obtain the layer information to be determined; and determining the at least one layer information according to the target identification information and the layer information to be determined.

[0013] The use of the planning model to perform predictive analysis on the preprocessing information to obtain the layer information to be determined includes: using the planning model to obtain the branch information to be predicted corresponding to the preprocessing information; optimizing the branch information to be predicted by using a pruning algorithm to obtain the optimized branch information to be predicted; and obtaining the layer information to be determined based on the optimized branch information to be predicted.

[0014] Optionally, the obtaining of the first completion image by performing layout planning and visual planning on the at least one layer information using the planning model includes: performing visual transformation processing on the at least one layer information to obtain a visualization image corresponding to the at least one layer information; obtaining sequence relationships and position information between layers corresponding to the at least one layer information by performing layout planning and visual planning on the visualization image; and obtaining the first completion image based on the sequence relationships and position information between layers corresponding to the at least one layer information.

[0015] Optionally, obtaining the target image based on the first degree-of-completion image includes: inputting the first degree-of-completion image into a quantitative upgrading model to obtain a second degree-of-completion image, wherein the number of layers contained in the second degree-of-completion image is not less than the number of layers contained in the first degree-of-completion image, and the quantitative upgrading model is used to convert the first degree-of-completion image into a quantitative image that meets preset conditions; inputting the second degree-of-completion image into a target construction model to obtain an initial target image, wherein the layers contained in the initial target image are filled with original material images; inputting the initial target image and the target constraint information into a post-processing optimization model to obtain an optimized target image, wherein the post-processing optimization model is used to optimize a part of the initial target image according to the target constraint information; and inputting the optimized target image into an image rendering model to obtain the target image.

[0016] Optionally, inputting the second degree-of-completion image into the target construction model to obtain the initial target image includes: obtaining feature information of the second degree-of-completion image according to the target construction model, and obtaining feature category information to which the feature information belongs; obtaining an original material image corresponding to the second degree-of-completion image from material image resource data according to the feature information of the second degree-of-completion image and the feature category information to which the feature information belongs; and filling the original material image into the second degree-of-completion image to obtain the initial target image.

[0017] Optionally, the method includes: inputting the target image into an evaluation model to obtain evaluation data for characterizing a display effect of the target image, wherein the evaluation model is used to obtain evaluation data for characterizing a display effect of an image.

[0018] Optionally, the method is applied to a server, and the method further includes: providing the target image to a client.

[0019] Optionally, the method further includes: providing the evaluation data to the client.

[0020] Optionally, the method is applied to a client, and the method further includes: obtaining the target image provided by a server; and displaying the target image.

[0021] Optionally, it includes: obtaining the evaluation data provided by the server; and displaying the evaluation data.

[0022] Optionally, the method is applied to a server, and the method further includes: providing the first completion image to a client.

[0023] Optionally, the method further includes: receiving supplementary constraint information for the first degree-of-completion image sent by the client; and updating the first degree-of-completion image according to the supplementary constraint information and the original image.

[0024] Optionally, the method further includes: generating constraint information of the original material image corresponding to the first degree-of-completion image; and providing the constraint information of the original material image corresponding to the first degree-of-completion image to the client.

[0025] Optionally, the method is applied to a client, and the method further includes: receiving the first degree of completion image provided by a server; and displaying the first degree of completion image.

[0026] Optionally, the method further includes: obtaining supplementary constraint information for the first degree-of-completion image; and sending the supplementary constraint information to the server.

[0027] Optionally, the method further includes: receiving constraint information of an original material image corresponding to the first degree-of-completion image provided by the server; obtaining the original material image according to the constraint information; and updating material image resource data corresponding to the original material image.

[0028] Optionally, the planning model is obtained in the following manner: obtaining an original sample image containing an original object and original sample constraint information, wherein the original sample constraint information is used to constrain attribute information of a target sample image to be obtained; using the original sample image and the original sample constraint information as sample data to train and obtain the planning model.

[0029] Optionally, the original sample image and the original sample constraint information are used as sample data to train the layout planning model, including: normalizing the original sample image and the original sample constraint information to obtain normalized preprocessed sample information; using the preprocessed sample information to train the planning model so that the prediction error of the planning model reaches a preset convergence condition.

[0030] Optionally, it also includes: obtaining target text information corresponding to the target object; obtaining a first completion image based on the original image and the target constraint information, including: obtaining the first completion image based on the original image, the target text information and the target constraint information.

[0031] Optionally, the content in at least one layer included in the target image corresponds to the target object, the target text information or the target constraint information.

[0032] Optionally, the feature information of the target object includes visual feature information and spatial feature information; the feature identification information includes visual feature identification information for identifying the visual feature information and spatial feature identification information for identifying the spatial feature information.

[0033] Optionally, the visual feature information includes at least one of the following feature information: color feature information, texture feature information, and style feature information.

[0034] Optionally, the target object includes a product object, and the target image is used to be displayed in a product details page corresponding to the product object.

[0035] The present application also provides an image acquisition device, including:

[0036] An information obtaining unit, used to obtain an original image containing a target object, and obtain target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained;

[0037] A first completion image obtaining unit, configured to obtain a first completion image according to the original image and the target constraint information, wherein the first completion image includes initial layout information and visual information of the target image to be obtained;

[0038] A target image obtaining unit is used to obtain the target image according to the first completion image, wherein the target image includes at least one layer, and the content in the at least one layer corresponds to the target object or the target constraint information.

[0039] The present application also provides an electronic device, including:

[0040] processor;

[0041] A memory is used to store a program of the image acquisition method. After the device is powered on and the program of the image acquisition method is run by the processor, the following steps are performed:

[0042] Obtain an original image containing a target object, and obtain target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained; obtain a first completion image based on the original image and the target constraint information, wherein the first completion image contains initial layout information and visual information of the target image to be obtained; obtain the target image based on the first completion image, wherein the target image includes at least one layer, and content in the at least one layer corresponds to the target object or the target constraint information.

[0043] The embodiment of the present application further provides a storage device storing a program of an image acquisition method, wherein the program is executed by a processor to perform the following steps:

[0044] Obtain an original image containing a target object, and obtain target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained; obtain a first completion image based on the original image and the target constraint information, wherein the first completion image contains initial layout information and visual information of the target image to be obtained; obtain the target image based on the first completion image, wherein the target image includes at least one layer, and content in the at least one layer corresponds to the target object or the target constraint information.

[0045] Compared with the prior art, this application has the following advantages:

[0046] The embodiment of the present application provides an image acquisition method, comprising: obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain the attribute information of the target image to be obtained; obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image contains the initial layout information and visual information of the target image to be obtained; obtaining the target image according to the first completion image, wherein the target image includes at least one layer, and the content in the at least one layer corresponds to the target object or the target constraint information. Compared with the method of directly performing refined design to obtain the target image, the method first obtains the first completion image containing the initial layout information and visual information of the target image to be obtained according to the original image and the target constraint information, thereby saving a lot of pre-judgment calculations and reducing the amount of calculations. After obtaining the first completion image, according to the first completion image, the target image corresponding to the content in the layer contained and the target object or the target constraint information can be obtained, and the target image that meets the user's expectations can be easily obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic diagram of an application scenario of an image acquisition method provided in the first embodiment of the present application.

[0048] Figure 2 It is a flowchart of an image acquisition method provided in the first embodiment of the present application.

[0049] Figure 3 It is a schematic diagram of the visualization process of an image acquisition method provided in the first embodiment of the present application.

[0050] Figure 4 It is a schematic diagram of an image acquisition device provided in the second embodiment of the present application.

[0051] Figure 5 It is a schematic diagram of an electronic device provided in the third embodiment of the present application. DETAILED DESCRIPTION

[0052] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0053] In order to enable those skilled in the art to better understand the present application, the following describes in detail the specific application scenarios of the embodiments based on the image acquisition method provided by the present application. The image acquisition method provided in the first embodiment of the present application can be applied to the scenario of interaction between the client and the server, such as Figure 1 As shown, it is a schematic diagram of an application scenario of an image acquisition method provided in the first embodiment of the present application.

[0054] In specific implementation, it is usually based on the user's desire to obtain a target image that meets certain requirements, such as meeting the requirements for product object promotion. After the client obtains the original image uploaded by the user containing the target object, and obtains the target constraint information input by the user for constraining the attribute information of the target image to be obtained, the client establishes a connection with the server. After the connection, the client sends the original image and the target constraint information to the server; after the server obtains the original image and the target constraint information, it obtains a first completion image based on the original image and the target constraint information, wherein the first completion image contains the initial layout information and visual information of the target image to be obtained; and based on the first completion image, obtains the target image, wherein the target image includes at least one layer, and the content in the at least one layer corresponds to the target object or the target constraint information; after the server obtains the target image, it provides the target image to the client; thereafter, the client displays the target image for the user to view or perform secondary editing.

[0055] It should be noted that the client can be a mobile terminal device, such as a mobile phone, a tablet computer, etc., or a common computer device. The server generally refers to a server, which can be a physical server or a cloud server, and is not specifically limited here.

[0056] In addition, in a specific implementation, the image acquisition method can also be applied separately to a client or server computing device. For example, after the client obtains the original image containing the target object and the target constraint information, it obtains the first degree-of-completion image based on the original image and the target constraint information, and obtains the target image based on the first degree-of-completion image. For another example, after the server obtains the original image containing the target object and the target constraint information from other servers, it obtains the first degree-of-completion image based on the original image and the target constraint information, and obtains the target image based on the first degree-of-completion image. After that, the obtained target image is stored in its own storage or remote storage without being provided to the client.

[0057] It should be noted that the above application scenarios are merely specific embodiments of the image acquisition method provided in the first embodiment of the present application. The purpose of providing the above application scenarios is to facilitate understanding of the image acquisition method, but not to limit the image acquisition method.

[0058] The first embodiment of the present application provides an image acquisition method, such as Figure 2 As shown, it is a flow chart of an image acquisition method provided by the first embodiment of the present application, such as Figure 3 As shown, it is a schematic diagram of the visualization process of an image acquisition method provided by the first embodiment of the present application. Figure 2 , Figure 3 The image acquisition method provided in the first embodiment of the present application is introduced.

[0059] Step S201, obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained.

[0060] The target object generally refers to an entity or virtual object that exists in the real world or the virtual world. For example, the target object may be a commodity object, or may also be a virtual cartoon object or other object. It should be noted that in the first embodiment of the present application, unless otherwise specified, the target object is a commodity object, and the target image to be obtained is an image for display in the commodity details page corresponding to the commodity object, and the image acquisition method described in this embodiment is described in detail. For example, the target object may be a shoe object of a certain brand, and the target image is a promotional poster corresponding to the shoe object.

[0061] The target constraint information is the attribute information used to constrain the target image to be obtained. For example, the target constraint information may be style information used to constrain the style of the target image, or composition information used to constrain the composition format of the target image, or color matching information used to constrain the main color of the target image, or size information used to constrain the size of the target image. Of course, the target constraint information may also be a combination of one or more information used to constrain the attribute information of the target image. For example, the target constraint information may be "style: motion, composition: left and right, main color: dark gray, size: 1125*352".

[0062] It should be noted that the target constraint information can be freely input by the user or selected from the preset constraint information. Of course, in order to better save computing resources, in the specific implementation, the client can display the target constraint information configuration interface of constraint information of different levels or corresponding to different contents, and the user can select the target constraint information in the configuration interface, and charge according to the user's selection result. After the user pays successfully, the subsequent processing of obtaining the target image is carried out, and the obtained target image is provided to the user; or, according to different user levels, the constraint information matching the user level can be provided to the user in the target constraint information configuration interface for the user to choose.

[0063] For example, in the target constraint information configuration interface, the user can simultaneously select several styles, several composition methods and other constraint information to obtain target images of different styles or different compositions. Afterwards, a corresponding payment interface is generated based on the target constraint information selected by the user. After the user pays successfully, the target image corresponding to the target constraint information is provided to the user.

[0064] When the image acquisition method is applied to a client, the obtaining of the original image containing the target object and the obtaining of the target constraint information may specifically include receiving the original image containing the target object and the target constraint information sent by a user or other client or server. When the image acquisition method is applied to a server, the obtaining of the original image containing the target object and the obtaining of the target constraint information may specifically include obtaining a data request message containing the original image and the target constraint information sent by a client or other server, and then the server obtains the original image containing the target object and the target constraint information from the data request message.

[0065] It should be noted that when the target image to be obtained is used in an e-commerce operation scenario or other promotional scenario, then in order to make the obtained target image more targeted, the method further includes: obtaining target copy information corresponding to the target object.

[0066] The target text information is information that describes the characteristics of the target object in text information. Figure 3 As shown, the copy information for a certain brand of shoe object may be "XXXX YYZZZ", where "XXXX" may be a word that modifies it, such as "feeling comfortable", and "YY ZZZ" may be a series of the shoe object, such as "Plus Max". In specific implementation, the target copy information may be set as needed, which will not be described in detail here.

[0067] Step S202: obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image includes initial layout information and visual information of the target image to be obtained.

[0068] In the first embodiment of the present application, in order to make the target image to be obtained more targeted and more in line with user expectations, after obtaining the original image and target constraint information containing the target object, by fitting the designer's thinking, that is, using a computing device to simulate the designer's offline design of the image to first design a sketch, and then further complete the sketch, and gradually enrich the process of obtaining the target image, the first embodiment of the present application is to provide a method that first obtains the first completion image containing the initial layout information and visual information of the target image to be obtained based on the obtained original image and target constraint information; then, based on the first completion image, obtain the target image. The following is a detailed description of how to obtain the first completion image.

[0069] The initial layout information refers to the initial spatial information of the content contained in the target image to be obtained, or the initial position information in the target image. For example, the target image is composed of two different types of elements, and the two elements are in different layers. According to different constraints such as design techniques, image styles or element types, the layers containing the two elements may have different spatial information; for example, the two layers are layer 1 and layer 2, and the initial layout information may be that layer 1 and layer 2 are arranged left and right. It should be noted that this is just an example. In specific implementation, the initial layout information can be set to other content as needed. For example, while numbering different layers, the left-right relationship, the up-and-down relationship, etc. are replaced by identification characters; after that, a string of encoding information for identifying the initial layout information of the target image can be obtained.

[0070] The initial visual information refers to the visual information represented by the content contained in the target image to be obtained. For example, the initial visual information may be at least one of the visual information such as type information, color information, layer information or decoration information of the content contained in the target image.

[0071] In the first embodiment of the present application, the initial layout information and initial visual information contained in the first completion image are only used to represent the initial information of the target image to be obtained, rather than the final information, that is, it is similar to the sketch of the target image obtained when the designer designs the image offline. It should be noted that the layout information and visual information contained in the target image finally obtained are more intuitive than the above-mentioned initial information in the first completion image. The target image is an image that is a specific quantification of the information contained in the first completion image and is filled with actual image materials.

[0072] The obtaining of the first degree-of-completion image according to the original image and the target constraint information comprises: normalizing the original image and the target constraint information to obtain normalized preprocessing information; inputting the preprocessing information into a planning model to obtain the first degree-of-completion image, wherein the planning model is used to obtain at least one layer information according to the preprocessing information, and perform layout planning and visual planning on the layer corresponding to the at least one layer information to obtain the sequence relationship and position information between the layers corresponding to the at least one layer information, wherein the layer information comprises the layout information and visual information of the layer corresponding to the layer information.

[0073] The planning model specifically obtains at least one layer of information obtained according to the preprocessing information, obtains a visualization image corresponding to the at least one layer of information, and obtains the sequence relationship and relative position information between different layer information by performing layout planning on the visualization image, and then outputs sketch images with different visual effects. In the first embodiment of the present application, the planning model is specifically a long short-term memory model (Long Short-Term Memory, LSTM), and the first completion image is specifically a 16*16 visualization image.

[0074] That is, the image acquisition method includes a preprocessing module, which is used to perform preprocessing and analysis on the original image and target constraint information, and then normalize the information in the original image and the target constraint information to obtain preprocessing information containing common information of the target object and the target constraint information in the original image; after the normalized preprocessing information is obtained through the preprocessing module, the preprocessing information is input into the planner of the image acquisition method, specifically: the obtained preprocessing information is input into a planning model used for layout planning and visual planning to obtain a first completion image.

[0075] The normalization process of the original image and the target constraint information to obtain normalized preprocessing information includes: obtaining feature identification information for identifying the target object according to the original image; obtaining target constraint identification information for identifying the target constraint information; merging information similar to the target constraint identification information in the feature identification information, and encoding the merged feature identification information and the target constraint identification information to obtain the preprocessing information. Since the original image usually includes different feature information, such as color feature information, texture feature information, and style feature information, in addition, the convolution feature information of the image can also be extracted by a convolutional neural network model for extracting image feature information. Therefore, in order to normalize the original image and the target constraint information, it is necessary to quantize the feature information of the original image and the target constraint information. Specifically, different types of feature information and constraint information can be encoded in advance, and then the feature information and target constraint information of the original image are quantized using the corresponding encoding identifiers.

[0076] The step of obtaining feature identification information for identifying the target object based on the original image includes: obtaining feature information of the target object contained in the original image; obtaining cluster center points of the feature information by performing cluster analysis on the feature information, and using identification information of the cluster center points as the feature identification information. It should be noted that the feature information of the target object includes visual feature information and spatial feature information; the feature identification information includes visual feature identification information for identifying the visual feature information and spatial feature identification information for identifying the spatial feature information. The visual feature information includes at least one of the following feature information: color feature information, texture feature information, and style feature information.

[0077] In the first embodiment of the present application, in order to conveniently and purposefully obtain a target image that meets the user's expectations, feature information of different categories of images, design methods corresponding to these feature information, element types corresponding to these design methods, and spatial quantization information and visual quantization information are classified and encoded in advance. For example, St types of theme styles, P types of design methods, T types of element types, Sp types of spatial quantization information, Vi types of visual quantization information and other information are obtained in advance, where St, P, T, Sp, and Vi are all not less than 0. It can be seen that the image acquisition method can have at least St*P*T*Sp*Vi different branch image design methods according to the differences between the original image obtained and the target constraint information. Therefore, feature identification information is obtained by extracting the feature information of the original image obtained, and target constraint information for identifying target constraint information is obtained, and information with similar or similar content in the obtained feature identification information and target constraint identification information is merged.

[0078] For example, the style information obtained from the feature information of the original image is merged with the style information in the target constraint information, and then the feature identification information and the target constraint identification information after the merge are encoded to obtain the preprocessing information. In the first embodiment of the present application, the normalized preprocessing information is specifically a 64-bit sequence code. For example, the preprocessing information can be a sequence code composed of "style identification information, category identification information, style feature information original code, category feature information original code", etc. Of course, in the specific implementation, the processing information can also be set to other bits as needed; in addition, how to use the convolutional neural network model to extract the feature information of the original image, and how to perform cluster analysis on the obtained feature information of different categories to obtain the detailed processing of the cluster center points are described in detail in the prior art and will not be repeated here.

[0079] It should be noted that theme style refers to different theme styles divided according to different user constraints or user expectations; design method refers to different design methods obtained after data mining of existing images, such as imaging method, composition method, etc.; element type refers to the basic unit that constitutes the image; spatial quantitative information refers to the position information of different element types corresponding to different design methods in two-dimensional space, specifically, after dividing the image into 16*16 image blocks, the position information of the image blocks where different element types are located is used as the spatial quantitative information of the element type; visual quantitative information refers to the quantitative information of the visual features of different element types.

[0080] The above describes in detail how to normalize the original image containing the target object and the target constraint information in the preprocessing module to obtain the normalized preprocessing information. After the preprocessing information is obtained by the preprocessing module, the preprocessing information is input into the planner of the image acquisition method to obtain the first completion image, which is described in detail below.

[0081] The step of inputting the normalized preprocessing information into the planning model to obtain the first completion image includes: acquiring at least one layer information corresponding to the preprocessing information according to the planning model; and performing layout planning and visual planning on the at least one layer information by using the planning model to obtain the first completion image.

[0082] The step of obtaining at least one layer information corresponding to the preprocessing information according to the planning model includes: obtaining the target identification information from the preprocessing information; performing predictive analysis on the preprocessing information using the planning model to obtain the layer information to be determined; and determining the at least one layer information according to the target identification information and the layer information to be determined. The step of performing predictive analysis on the preprocessing information using the planning model to obtain the layer information to be determined includes: obtaining the branch information to be predicted corresponding to the preprocessing information using the planning model; optimizing the branch information to be predicted using a pruning algorithm to obtain the optimized branch information to be predicted; and obtaining the layer information to be determined according to the optimized branch information to be predicted.

[0083] The process of the planning model obtaining at least one layer information according to the received preprocessing information is specifically: performing predictive analysis on the received preprocessing information, such as performing branch prediction on the element types, design methods, etc. of the sketch image to be obtained, so as to obtain at least one layer information in the sketch image; it should be noted that the planning model can usually obtain a number of branch information to be predicted according to the preprocessing information, wherein the branch information to be predicted is a prediction branch for obtaining different contents constituting the first degree of completion image. For example, the planning model predicts various branches of the composition according to the preprocessing information, such as image background branches, decoration branches, object identification branches, etc. The different branches can specifically constitute a multi-branch selection tree, and the layer information finally output is at least one branch path of the multi-branch selection tree, that is, the output is at least one layer information to be determined, and then, after matching the at least one layer information to be determined with the target constraint information, the layer information to be determined that matches the target constraint information is used as the layer information to be finally obtained. It should be noted that the at least one layer information may specifically be a 64-bit sequence code; in addition, in the layer information prediction model, a pruning algorithm may be used in conjunction with acceleration to increase the computing speed, which will not be elaborated here.

[0084] After obtaining at least one layer information corresponding to the preprocessing information according to the preprocessing information, the first completion image is obtained by performing layout planning and visual planning on the at least one layer information using the planning model, including: performing visual transformation processing on the at least one layer information to obtain a visual image corresponding to the at least one layer information; obtaining sequence relationship and position information between layers corresponding to the at least one layer information by performing layout planning and visual planning on the visual image; and obtaining the first completion image according to the sequence relationship and position information between layers corresponding to the at least one layer information.

[0085] In addition, it should be noted that in the process of using the planning model to obtain the first degree-of-completion image, a graphics processing unit (GPU) optimization strategy or a GPU optimization algorithm may be used for acceleration to increase the speed of obtaining the first degree-of-completion image.

[0086] In addition, the planning model can be obtained in the following manner: obtaining an original sample image containing an original object and original sample constraint information, wherein the original sample constraint information is used to constrain the attribute information of the target sample image to be obtained; using the original sample image and the original sample constraint information as sample data to train and obtain the planning model.

[0087] The method of using the original sample image and the original sample constraint information as sample data to train and obtain the planning model includes: normalizing the original sample image and the original sample constraint information to obtain normalized preprocessed sample information; and using the preprocessed sample information to train the planning model so that the prediction error of the planning model reaches a preset convergence condition.

[0088] In a specific implementation, if the target text information corresponding to the target object is also obtained in step S201, then the step of obtaining the first completion image according to the original image and the target constraint information includes: obtaining the first completion image according to the original image, the target text information and the target constraint information. The step of normalizing the original image and the target constraint information to obtain normalized preprocessing information includes: normalizing the original image, the target text information and the target constraint information to obtain normalized preprocessing information. It should be noted that when the target text information is normalized, it is specifically based on the image style constraint information in the target constraint information. After the target text information is converted into a corresponding visual image, it is encoded according to the image style constraint information and the feature identification information corresponding to the visual image to obtain the identification information corresponding to the target text information.

[0089] According to the above description, the image acquisition method provided in the first embodiment of the present application simulates the designer's thinking process of designing sketches through different network models, and fits the original image containing the target object and the target constraint information to obtain a serialized first completion image that meets the user's expectations and is composed of different layer information. Through the serialized layer information, the initial layout information and visual information of the target image to be obtained can be determined.

[0090] In addition, in order to further reduce the amount of calculation, after obtaining the first degree of completion image, the first degree of completion image can be provided to the user for viewing, so that the user can confirm whether the preliminary design style meets expectations, and the user can provide further supplementary constraint information based on the first degree of completion graphic to improve the accuracy of the target image finally obtained. That is to say, when the method is applied to the server, the method also includes: providing the first degree of completion image to the client. In addition, the method also includes: receiving supplementary constraint information for the first degree of completion image sent by the client; updating the first degree of completion image according to the supplementary constraint information and the original image. Correspondingly, when the method is applied to the client, the method also includes: receiving the first degree of completion image provided by the server; displaying the first degree of completion image; in addition, it also includes: obtaining supplementary constraint information for the first degree of completion image; and sending the supplementary constraint information to the server.

[0091] It should also be noted that after the server obtains the first degree of completion image, it can also generate constraint information of the original material image to be filled in different layers, for example, the size information of the original material image of the product LOGO in the LOGO layer of the corresponding product object, and provide the constraint information to the client for the user or the material provider for providing the material image to view the constraint information, and generate the corresponding original material image according to the constraint information, and update the corresponding material image resource data, so that the server can obtain the original material image in the subsequent process of generating the target image to provide the accuracy of the target image, that is, when the method is applied to the server, it also includes: generating constraint information of the original material image corresponding to the first degree of completion image; providing the constraint information of the original material image corresponding to the first degree of completion image to the client. Correspondingly, when the method is applied to the client, it also includes: receiving constraint information of the original material image corresponding to the first degree of completion image provided by the server; obtaining the original material image according to the constraint information; and updating the material image resource data corresponding to the original material image.

[0092] Step S203: obtaining the target image according to the first completion image, wherein the target image includes at least one layer, and the content in the at least one layer corresponds to the target object or the target constraint information.

[0093] After the above step S202, the target image can be obtained by further quantitatively analyzing the obtained first degree-of-completion image and filling different materials into the layers of the first degree-of-completion image accordingly.

[0094] The method of obtaining the target image according to the first degree-of-completion image includes: inputting the first degree-of-completion image into a quantitative upgrading model to obtain a second degree-of-completion image, wherein the number of layers contained in the second degree-of-completion image is not less than the number of layers contained in the first degree-of-completion image, and the quantitative upgrading model is used to convert the first degree-of-completion image into a quantitative image that meets preset conditions; inputting the second degree-of-completion image into a target construction model to obtain an initial target image, wherein the layers contained in the initial target image are filled with original material images; inputting the initial target image and the target constraint information into a post-processing optimization model to obtain an optimized target image, wherein the post-processing optimization model is used to optimize a part of the initial target image according to the target constraint information; and inputting the optimized target image into an image rendering model to obtain the target image.

[0095] In this step, it specifically includes an actor, a builder, a post-processing module and a renderer. The actor is specifically used to upgrade the image quality of the first completion image so that the resolution of the first completion image meets the preset conditions. Correspondingly, the layer information in the first completion image can also be further split to obtain more fine-grained layer information; the builder is used to retrieve the corresponding original material image in the pre-acquired material image resource data according to the feature information in the layer information in the image output by the actor and the feature category information corresponding to the feature information, and fill the retrieved original material image into the corresponding layer to obtain the initial target image; the post-processing module is specifically used to optimize the local part of the image, such as optimizing the outline, border or font of the image element; the rendering model is used to combine the above layer information to obtain the target image.

[0096] The quantitative upgrade model is specifically a generative adversarial network model (GAN), which is used to perform image quality upgrade processing from coarse to fine and layer splitting processing on the first completion image obtained in the builder layer by layer to make it meet the preset conditions, so as to improve the fineness of the target image to be obtained. In the first embodiment of the present application, the preset condition is specifically to upgrade the 16*16 first completion image to a 64*64 second completion image. Since how to upgrade the image quality is a prior art, it will not be repeated here.

[0097] The step of inputting the second degree-of-completion image into the target construction model to obtain the initial target image includes: obtaining feature information of the second degree-of-completion image according to the target construction model, and obtaining feature category information to which the feature information belongs; obtaining an original material image corresponding to the second degree-of-completion image from material image resource data according to the feature information of the second degree-of-completion image and the feature category information to which the feature information belongs; and filling the original material image into the second degree-of-completion image to obtain the initial target image.

[0098] It should be noted that, in specific implementation, the quantitative upgrade model, the target construction model, the post-processing optimization model and the image rendering model can be obtained by separate training using the corresponding sample data, or these models can be jointly trained according to the correlation between these models, which will not be repeated here. In addition, in specific implementation, of course, all network models used in the image acquisition method, such as the above-mentioned planning model, quantitative upgrade model, target construction model, post-processing optimization model and image rendering model, can also be jointly trained, and in the process of joint training, the prediction errors of the above models are adjusted through the corresponding loss functions of the above models and the hyperparameters between the models as a whole, so that the above models reach their respective preset convergence conditions to improve the accuracy of the output results of the above models, which will not be repeated here.

[0099] At this point, the target image that meets the user's expectations can be obtained in a targeted and convenient manner. It should be noted that if the target text information corresponding to the target object is also obtained in step S201, then the content in at least one layer included in the target image corresponds to the target object, the target text information or the target constraint information.

[0100] After obtaining the target image through the above steps, when the image obtaining method is applied to the client, the method further includes: obtaining the target image provided by the server and displaying the target image. When the image obtaining method is applied to the server, the method further includes: providing the target image to the client.

[0101] In addition, after obtaining the target image, the method further includes: inputting the target image into the evaluation model to obtain evaluation data for characterizing the display effect of the target image, wherein the evaluation model is used to obtain evaluation data for characterizing the display effect of the image. That is, through the evaluation model for evaluating the display effect of the target image, the evaluation data corresponding to the target image is obtained, and the evaluation model can specifically evaluate the target image through at least one dimension such as color dimension and layer layout dimension, and obtain the evaluation data.

[0102] After obtaining the evaluation data, when the image obtaining method is applied to the client, the method further includes: obtaining the evaluation data provided by the server and displaying the evaluation data. When the image obtaining method is applied to the server, the method further includes: providing the evaluation data to the client.

[0103] It should be noted that since the target image ultimately obtained in the present application is a structured image composed of different layers, after the target image is provided to the user, the user can also perform secondary editing based on the target image, for example, using drawing software corresponding to the format of the target image to edit the target image, such as replacing the material image in a certain layer, or editing the text information in a certain layer, such as the font, font size, font color, etc. of the text, so that the user can easily obtain the desired image.

[0104] In summary, the image acquisition method provided by the first embodiment of the present application includes: obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain the attribute information of the target image to be obtained; obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image contains the initial layout information and visual information of the target image to be obtained; obtaining the target image according to the first completion image, wherein the target image includes at least one layer, and the content in the at least one layer corresponds to the target object or the target constraint information. Compared with the method of directly performing refined design to obtain the target image, the method first obtains the first completion image containing the initial layout information and visual information of the target image to be obtained according to the original image and the target constraint information, thereby saving a lot of pre-judgment calculations and reducing the amount of calculations. After obtaining the first completion image, according to the first completion image, the target image corresponding to the content in the layer contained and the target object or the target constraint information can be obtained, and the target image that meets the user's expectations can be easily obtained.

[0105] Corresponding to the image acquisition method provided in the first embodiment of the present application, the second embodiment of the present application further provides an image acquisition device, see Figure 4 , which is a schematic diagram of an image acquisition device provided in the second embodiment of the present application. Since the device embodiment is basically similar to the method embodiment, the description is relatively simple. For relevant parts, please refer to the partial description of the method embodiment. The device embodiment described below is only illustrative. An image acquisition device provided in the second embodiment of the present application includes the following parts:

[0106] The information obtaining unit 401 is used to obtain an original image containing a target object and obtain target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained;

[0107] A first completion image obtaining unit 402 is used to obtain a first completion image according to the original image and the target constraint information, wherein the first completion image includes initial layout information and visual information of the target image to be obtained;

[0108] The target image obtaining unit 403 is used to obtain the target image according to the first completion image, wherein the target image includes at least one layer, and the content in the at least one layer corresponds to the target object or the target constraint information.

[0109] Optionally, the first completion image acquisition unit is specifically used to: normalize the original image and the target constraint information to obtain normalized pre-processing information; input the pre-processing information into the planning model to obtain the first completion image, wherein the layout planning model is used to obtain at least one layer information based on the pre-processing information, and perform layout planning and visual planning on the layer corresponding to the at least one layer information to obtain the sequence relationship and position information between the layers corresponding to the at least one layer information, and the layer information includes the layout information and visual information of the layer corresponding to the layer information.

[0110] Optionally, the original image and the target constraint information are normalized to obtain normalized preprocessing information, including: obtaining feature identification information for identifying the target object based on the original image; obtaining target constraint identification information for identifying the target constraint information; merging information in the feature identification information that is similar to the target constraint identification information, and encoding the merged feature identification information and the target constraint identification information to obtain the preprocessing information.

[0111] Optionally, obtaining feature identification information for identifying the target object based on the original image includes: acquiring the feature information of the target object contained in the original image; obtaining the cluster center point of the feature information by performing cluster analysis on the feature information, and using the identification information of the cluster center point as the feature identification information.

[0112] Optionally, inputting the normalized preprocessing information into the planning model to obtain the first completion image includes: acquiring at least one layer information corresponding to the preprocessing information according to the planning model; and performing layout planning and visual planning on the at least one layer information by using the planning model to obtain the first completion image.

[0113] Optionally, obtaining at least one layer information corresponding to the preprocessing information according to the planning model includes: obtaining the target identification information from the preprocessing information; performing predictive analysis on the preprocessing information using the planning model to obtain the layer information to be determined; and determining the at least one layer information according to the target identification information and the layer information to be determined.

[0114] Optionally, the use of the planning model to perform predictive analysis on the preprocessing information to obtain the layer information to be determined includes: using the planning model to obtain branch information to be predicted corresponding to the preprocessing information; optimizing the branch information to be predicted by using a pruning algorithm to obtain optimized branch information to be predicted; and obtaining the layer information to be determined based on the optimized branch information to be predicted.

[0115] Optionally, performing layout planning and visual planning on the at least one layer information by using the planning model includes: performing visual transformation processing on the at least one layer information to obtain a visual image corresponding to the at least one layer information; obtaining sequence relationships and position information between layers corresponding to the at least one layer information by performing layout planning and visual planning on the visual image; and obtaining the first completion image based on the sequence relationships and position information between layers corresponding to the at least one layer information.

[0116] Optionally, the target image acquisition unit is specifically used to: input the first degree-of-completion image into a quantitative upgrade model to obtain a second degree-of-completion image, wherein the number of layers contained in the second degree-of-completion image is not less than the number of layers contained in the first degree-of-completion image, and the quantitative upgrade model is used to convert the first degree-of-completion image into a quantitative image that meets preset conditions; input the second degree-of-completion image into a target construction model to obtain an initial target image, wherein the layers contained in the initial target image are filled with original material images; input the initial target image and the target constraint information into a post-processing optimization model to obtain an optimized target image, wherein the post-processing optimization model is used to optimize a part of the initial target image according to the target constraint information; input the optimized target image into an image rendering model to obtain the target image.

[0117] Optionally, inputting the second degree-of-completion image into the target construction model to obtain the initial target image includes: obtaining feature information of the second degree-of-completion image according to the target construction model, and obtaining feature category information to which the feature information belongs; obtaining an original material image corresponding to the second degree-of-completion image from material image resource data according to the feature information of the second degree-of-completion image and the feature category information to which the feature information belongs; and filling the original material image into the second degree-of-completion image to obtain the initial target image.

[0118] Optionally, the device further includes an evaluation unit, configured to input the target image into an evaluation model to obtain evaluation data for characterizing a display effect of the target image, wherein the evaluation model is configured to obtain evaluation data for characterizing a display effect of an image.

[0119] Optionally, when the device is applied to a server, the device further includes a providing unit, configured to provide the target image to a client.

[0120] Optionally, the providing unit is further configured to provide the target image to the client.

[0121] Optionally, when the device is applied to a client, the device further includes a display unit, which is used to obtain the target image provided by a server; and display the target image.

[0122] Optionally, the display unit is further used to obtain the evaluation data provided by the server; and display the evaluation data.

[0123] Optionally, the information obtaining unit is further used to obtain target text information corresponding to the target object; the first completion image obtaining unit is further used to obtain the first completion image based on the original image, the target text information and the target constraint information.

[0124] Optionally, the content in at least one layer included in the target image in the target image obtaining unit corresponds to the target object, the target text information or the target constraint information.

[0125] Optionally, the feature information of the target object includes visual feature information and spatial feature information; the feature identification information includes visual feature identification information for identifying the visual feature information and spatial feature identification information for identifying the spatial feature information.

[0126] Optionally, the visual feature information includes at least one of the following feature information: color feature information, texture feature information, and style feature information.

[0127] Optionally, the target object includes a commodity object.

[0128] Corresponding to the image acquisition method provided in the first embodiment of the present application, the third embodiment of the present application further provides an electronic device, see Figure 5 , which is a schematic diagram of an electronic device provided in the third embodiment of the present application. Since the electronic device embodiment is basically similar to the method embodiment, the description is relatively simple. For the relevant parts, please refer to the partial description of the method embodiment. The electronic device embodiment described below is only illustrative. The electronic device provided in the third embodiment of the present application includes:

[0129] Processor 501;

[0130] The memory 502 is used to store a program for the image acquisition method. After the device is powered on and the program for the image acquisition method is run by the processor, the following steps are performed:

[0131] Obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained;

[0132] Obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image includes initial layout information and visual information of the target image to be obtained;

[0133] The target image is obtained according to the first completion image, wherein the target image includes at least one layer, and content in the at least one layer corresponds to the target object or the target constraint information.

[0134] Corresponding to the image acquisition method provided in the first embodiment of the present application, the fourth embodiment of the present application further provides a storage device. Since the storage device embodiment is basically similar to the method embodiment, the description is relatively simple. For the relevant parts, please refer to the partial description of the method embodiment. The storage device embodiment described below is only illustrative. The fourth embodiment of the present application provides a storage device, which stores a program of the image acquisition method. The program is executed by a processor to perform the following steps:

[0135] Obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained;

[0136] Obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image includes initial layout information and visual information of the target image to be obtained;

[0137] The target image is obtained according to the first completion image, wherein the target image includes at least one layer, and content in the at least one layer corresponds to the target object or the target constraint information.

[0138] Although the present application is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

[0139] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0140] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0141] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0142] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

Claims

1. A method for obtaining an image, characterized in that: include: Obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained; Obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image includes initial layout information and visual information of the target image to be obtained; According to the first completion image, the target image is obtained, wherein the target image includes at least one layer, and content in the at least one layer corresponds to the target object or the target constraint information; Wherein, obtaining the first completion image based on the original image and the target constraint information includes: obtaining feature identification information for identifying the target object based on the original image; obtaining target constraint identification information for identifying the target constraint information; merging information in the feature identification information that is similar to information in the target constraint identification information, and encoding the merged feature identification information and the target constraint identification information to obtain preprocessing information; and obtaining the first completion image based on the preprocessing information.

2. The image acquisition method according to claim 1, characterized in that: The obtaining the first completion image based on the preprocessing information includes: The preprocessing information is input into a planning model to obtain the first completion image, wherein the planning model is used to obtain at least one layer information based on the preprocessing information, and perform layout planning and visual planning on the layer corresponding to the at least one layer information to obtain the sequence relationship and position information between the layers corresponding to the at least one layer information, and the layer information includes layout information and visual information of the layer corresponding to the layer information.

3. The image acquisition method according to claim 2, characterized in that: The step of obtaining feature identification information for identifying the target object according to the original image includes: Acquire feature information of the target object contained in the original image; The cluster center point of the feature information is obtained by performing cluster analysis on the feature information, and identification information of the cluster center point is used as the feature identification information.

4. The image acquisition method according to claim 2, characterized in that: The step of inputting the preprocessing information into a planning model to obtain the first completion image includes: According to the planning model, obtaining at least one layer information corresponding to the preprocessing information; The first completion image is obtained by performing layout planning and visual planning on the at least one layer information using the planning model.

5. The image acquisition method according to claim 4, characterized in that: The step of acquiring at least one layer of information corresponding to the preprocessing information according to the planning model includes: Obtaining target identification information from the preprocessing information; Use the planning model to perform prediction analysis on the preprocessing information to obtain the layer information to be determined; The at least one layer information is determined according to the target identification information and the layer information to be determined.

6. The image acquisition method according to claim 5, characterized in that: The using the planning model to perform predictive analysis on the preprocessing information to obtain the layer information to be determined includes: Using the planning model, obtaining to-be-predicted branch information corresponding to the preprocessing information; By using a pruning algorithm to optimize the branch information to be predicted, the optimized branch information to be predicted is obtained; The to-be-determined layer information is obtained according to the optimized to-be-predicted branch information.

7. The image acquisition method according to claim 5, characterized in that: The obtaining of the first completion image by performing layout planning and visual planning on the at least one layer information by using the planning model includes: Performing visualization conversion processing on the at least one layer of information to obtain a visualization image corresponding to the at least one layer of information; By performing layout planning and visual planning on the visualization image, a sequence relationship and position information between layers corresponding to the at least one layer information is obtained; The first degree-of-completion image is obtained according to the sequence relationship and position information between the layers corresponding to the at least one layer information.

8. The image acquisition method according to claim 1, characterized in that: The step of obtaining the target image according to the first completion image includes: Inputting the first degree-of-completion image into a quantitative upgrading model to obtain a second degree-of-completion image, wherein the number of layers included in the second degree-of-completion image is not less than the number of layers included in the first degree-of-completion image, and the quantitative upgrading model is used to convert the first degree-of-completion image into a quantitative image that meets a preset condition; Inputting the second completion image into a target construction model to obtain an initial target image, wherein the layer contained in the initial target image is filled with the original material image; Inputting the initial target image and the target constraint information into a post-processing optimization model to obtain an optimized target image, wherein the post-processing optimization model is used to optimize a part of the initial target image according to the target constraint information; The optimized target image is input into an image rendering model to obtain the target image.

9. The image acquisition method according to claim 8, characterized in that: The step of inputting the second completion image into the target construction model to obtain an initial target image includes: constructing a model according to the target, obtaining feature information of the second degree-of-completion image, and obtaining feature category information to which the feature information belongs; According to the feature information of the second degree-of-completion image and the feature category information to which the feature information belongs, obtaining an original material image corresponding to the second degree-of-completion image from the material image resource data; The original material image is filled into the second completion image to obtain the initial target image.

10. The image acquisition method according to claim 1, characterized in that: include: The target image is input into an evaluation model to obtain evaluation data for characterizing a display effect of the target image, wherein the evaluation model is used to obtain evaluation data for characterizing a display effect of an image.

11. The image acquisition method according to claim 10, characterized in that: The method is applied to the server, and the method further includes: The target image is provided to the client.

12. The image acquisition method according to claim 11, characterized in that: Also includes: The evaluation data is provided to the client.

13. The image acquisition method according to claim 10, characterized in that: The method is applied to a client, and the method further comprises: Obtain the target image provided by the server; The target image is displayed.

14. The image acquisition method according to claim 13, characterized in that: include: Obtaining the evaluation data provided by the server; The evaluation data is presented.

15. The image acquisition method according to claim 1, characterized in that: The method is applied to the server, and the method further includes: The first completion image is provided to the client.

16. The image acquisition method according to claim 15, characterized in that: Also includes: receiving supplementary constraint information for the first degree-of-completion image sent by the client; The first degree-of-completion image is updated according to the supplemental constraint information and the original image.

17. The image acquisition method according to claim 15, characterized in that: Also includes: generating constraint information of the original material image corresponding to the first degree-of-completion image; The constraint information of the original material image corresponding to the first degree-of-completion image is provided to the client.

18. The image acquisition method according to claim 1, characterized in that: The method is applied to a client, and the method further comprises: Receiving the first completion image provided by the server; The first completion image is displayed.

19. The image acquisition method according to claim 18, characterized in that: Also includes: Acquire supplementary constraint information for the first degree-of-completion image; The supplementary constraint information is sent to the server.

20. The image acquisition method according to claim 18, characterized in that: Also includes: Receiving constraint information of an original material image corresponding to the first degree-of-completion image provided by the server; According to the constraint information, obtaining the original material image; The material image resource data corresponding to the original material image is updated.

21. The image acquisition method according to claim 5, characterized in that: The planning model is obtained in the following way: Obtaining an original sample image containing an original object and original sample constraint information, wherein the original sample constraint information is used to constrain attribute information of a target sample image to be obtained; The original sample image and the original sample constraint information are used as sample data to train and obtain the planning model.

22. The image acquisition method according to claim 21, characterized in that: The method of using the original sample image and the original sample constraint information as sample data to train and obtain the layout planning model includes: Normalizing the original sample image and the original sample constraint information to obtain normalized pre-processed sample information; The planning model is trained using the preprocessed sample information so that the prediction error of the planning model reaches a preset convergence condition.

23. The image acquisition method according to claim 1, characterized in that: Also includes: Obtaining target copy information corresponding to the target object; The step of obtaining a first completion image according to the original image and the target constraint information includes: The first completion image is obtained according to the original image, the target text information and the target constraint information.

24. The image acquisition method according to claim 23, characterized in that: The content in at least one layer included in the target image corresponds to the target object, the target text information or the target constraint information.

25. The image acquisition method according to claim 3, characterized in that: The characteristic information of the target object includes visual characteristic information and spatial characteristic information; The feature identification information includes visual feature identification information for identifying the visual feature information and spatial feature identification information for identifying the spatial feature information.

26. The image acquisition method according to claim 25, characterized in that: The visual feature information includes at least one of the following feature information: color feature information, texture feature information, and style feature information.

27. The image acquisition method according to claim 1, characterized in that: The target object includes a commodity object, and the target image is used to be displayed in a commodity details page corresponding to the commodity object.

28. An image acquisition device, characterized in that: include: An information obtaining unit, used to obtain an original image containing a target object, and obtain target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained; A first completion image obtaining unit, configured to obtain a first completion image according to the original image and the target constraint information, wherein the first completion image includes initial layout information and visual information of the target image to be obtained; a target image obtaining unit, configured to obtain the target image according to the first completion image, wherein the target image includes at least one layer, and content in the at least one layer corresponds to the target object or the target constraint information; Among them, the first completion image acquisition unit is used to obtain the first completion image according to the original image and the target constraint information through the following steps: according to the original image, obtain feature identification information for identifying the target object; obtain target constraint identification information for identifying the target constraint information; merge the information in the feature identification information that is similar to the target constraint identification information, and encode the merged feature identification information and the target constraint identification information to obtain preprocessing information; based on the preprocessing information, obtain the first completion image.

29. An electronic device, characterized in that: include: processor; A memory is used to store a program of the image acquisition method. After the device is powered on and the program of the image acquisition method is run by the processor, the following steps are performed: Obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained; Obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image includes initial layout information and visual information of the target image to be obtained; According to the first completion image, the target image is obtained, wherein the target image includes at least one layer, and content in the at least one layer corresponds to the target object or the target constraint information; Wherein, obtaining the first completion image based on the original image and the target constraint information includes: obtaining feature identification information for identifying the target object based on the original image; obtaining target constraint identification information for identifying the target constraint information; merging information in the feature identification information that is similar to information in the target constraint identification information, and encoding the merged feature identification information and the target constraint identification information to obtain preprocessing information; and obtaining the first completion image based on the preprocessing information.

30. A storage device, characterized in that: A program for storing an image acquisition method is executed by a processor to perform the following steps: Obtaining an original image containing a target object, and obtaining target constraint information, wherein the target constraint information is used to constrain attribute information of the target image to be obtained; Obtaining a first completion image according to the original image and the target constraint information, wherein the first completion image includes initial layout information and visual information of the target image to be obtained; According to the first completion image, the target image is obtained, wherein the target image includes at least one layer, and content in the at least one layer corresponds to the target object or the target constraint information; Wherein, obtaining the first completion image based on the original image and the target constraint information includes: obtaining feature identification information for identifying the target object based on the original image; obtaining target constraint identification information for identifying the target constraint information; merging information in the feature identification information that is similar to information in the target constraint identification information, and encoding the merged feature identification information and the target constraint identification information to obtain preprocessing information; and obtaining the first completion image based on the preprocessing information.

Images