WO2023207681A1

WO2023207681A1 - Method and apparatus for intelligent clothing matching, and electronic device and storage medium

Info

Publication number: WO2023207681A1
Application number: PCT/CN2023/089038
Authority: WO
Inventors: 黄伟强; 邹星星; 庞楷成
Original assignee: 人工智能设计研究所有限公司; 香港理工大学; 皇家艺术学院
Priority date: 2022-04-28
Filing date: 2023-04-18
Publication date: 2023-11-02
Also published as: WO2023207681A9; CN117056548A

Abstract

Provided in the present application are a method and apparatus for intelligent clothing matching, and an electronic device and a storage medium. The method comprises: acquiring a first clothing image and a second clothing image; performing feature extraction on the first clothing image and the second clothing image, so as to obtain a feature value of a first clothing element of the first clothing image, a feature value of a second clothing element of the first clothing image, a feature value of a first clothing element of the second clothing image and a feature value of a second clothing element of the second clothing image; determining a first matching compatibility and a second matching compatibility between the first clothing image and the second clothing image; generating a clothing matching compatibility and a matching level between the first clothing image and the second clothing image according to the first matching compatibility and the second matching compatibility; and generating a clothing matching combination. By means of the method, a clothing matching combination having a better matching effect can be automatically generated according to a clothing matching compatibility and a matching level of clothing images, thereby improving the efficiency of generating the clothing matching combination, and improving the user experience.

Description

Methods, devices, electronic equipment and storage media for intelligent matching of clothing

Cross-references to related applications

This disclosure requests the priority of the Chinese patent application with application number 202210470116.6 and titled "Method, device, electronic device and storage medium for intelligent matching of clothing" submitted on April 28, 2022. All the Chinese patent applications The contents are incorporated herein by reference in their entirety.

Technical field

The present disclosure relates to the field of image processing technology, and in particular, to a method, device, electronic device and storage medium for intelligent matching of clothing.

Background technique

With the development of computer technology and artificial intelligence technology, more and more data-driven technologies are being applied to various practical products and daily life. In the clothing industry, with the advent of the digital era, the digitization of clothing has attracted more and more attention.

In the clothing matching process of related technologies, the clothing image input by the user is usually obtained, and similar clothing images are recommended to the user based on the clothing image. However, the recommended clothing image and the clothing image input by the user cannot be matched in a complete set, or The matching effect is poor.

It should be noted that the information disclosed in the above background section is only used to enhance understanding of the background of the present disclosure, and therefore may include information that does not constitute prior art known to those of ordinary skill in the art.

Contents of the invention

The purpose of this disclosure is to provide a method, device, electronic device and storage medium for intelligent clothing matching. This method can automatically generate clothing matching combinations with better matching effects based on the clothing matching compatibility and matching levels of clothing images, and improve It improves the efficiency of generating clothing matching combinations, thereby improving user experience.

Additional features and advantages of the disclosure will be apparent from the following detailed description, or, in part, may be learned by practice of the disclosure.

Embodiments of the present disclosure provide a method for intelligent matching of clothing, including: obtaining a first clothing image of a first clothing category and a second clothing image of a second clothing category; and separately comparing the first clothing image and the second clothing image. Perform feature extraction on two clothing images to obtain the characteristic value of the first clothing element of the first clothing image, the characteristic value of the second clothing element of the first clothing image, and the first clothing element of the second clothing image. and the characteristic value of the second clothing element of the second clothing image; determining the first characteristic value between the characteristic value of the first clothing element of the first clothing image and the characteristic value of the first clothing element of the second clothing image. Matching compatibility, and determining the second matching compatibility between the characteristic value of the second clothing element of the first clothing image and the characteristic value of the second clothing element of the second clothing image; according to the first matching compatibility and the second matching Compatibility, generating clothing matching compatibility and matching level of the first clothing image and the second clothing image; according to the clothing matching compatibility of the first clothing image and the second clothing image Compatibility and matching level, generate clothing matching combinations.

In some exemplary embodiments of the present disclosure, the method further includes: according to the first matching compatibility, the second matching compatibility and the clothing matching compatibility, from the first clothing element and the The target clothing element is determined among the second clothing elements; and the reason for generating the matching level is obtained according to the target clothing element and the matching level.

In some exemplary embodiments of the present disclosure, generating a clothing matching combination based on the clothing matching compatibility and matching level of the first clothing image and the second clothing image includes: if the first clothing image and the second clothing image If the matching level of the second clothing image belongs to the first matching level, a first clothing matching combination is generated based on the first clothing image and the second clothing image.

In some exemplary embodiments of the present disclosure, generating a clothing matching combination based on the clothing matching compatibility and matching level of the first clothing image and the second clothing image also includes: if the first clothing image and the second clothing image If the matching level of the second clothing image belongs to the second matching level, then a matching suggestion for replacing the second clothing image is generated according to the target clothing element, wherein the first matching level is better than the second matching level. ;Acquire a third clothing image of the second clothing category according to the matching suggestion, wherein the target clothing element of the third clothing image is different from the target clothing element of the second clothing image; According to the first clothing The image and the third clothing image generate a second clothing matching combination.

In some exemplary embodiments of the present disclosure, the method further includes: determining suitable body feature tags for the clothing matching combination; training to obtain a clothing matching model based on the clothing matching combination and its suitable body feature tags; and obtaining the target object The physical characteristic data of the target object is input into the clothing matching model to obtain a target clothing matching combination suitable for the target object.

In some exemplary embodiments of the present disclosure, the target clothing matching combination includes a first target clothing image and a second target clothing image; wherein the method further includes: generating the target according to the body characteristic data of the target object. a virtual image of an object; determining the dressing order of the first target clothing image and the second target clothing image; and combining the first target clothing image and the second target clothing image based on the dressing order and the The above virtual images are fused to generate the target outfit image.

In some exemplary embodiments of the present disclosure, fusing the first target clothing image and the second target clothing image with the virtual image based on the dressing order to generate a target outfit image includes: respectively The clothing structures of the first target clothing image and the second target clothing image are analyzed to obtain the first clothing front image of the first target clothing image and the second clothing front image of the second target clothing image. According to the dressing order, the virtual image is sequentially fused with the first clothing front image and the second clothing front image to generate the target outfit image.

The present disclosure provides a device for intelligent matching of clothing, including: an image acquisition module for acquiring a first clothing image of a first clothing category and a second clothing image of a second clothing category; a feature extraction module for respectively The first clothing image and the second clothing image perform feature extraction to obtain the characteristic value of the first clothing element of the first clothing image, the characteristic value of the second clothing element of the first clothing image, and the the characteristic value of the first clothing element of the second clothing image and the characteristic value of the second clothing element of the second clothing image; a determination module for determining the characteristic value of the first clothing element of the first clothing image and the second Characteristic value of the first clothing element of the clothing image the first matching compatibility between them, and determining the second matching compatibility between the characteristic value of the second clothing element of the first clothing image and the characteristic value of the second clothing element of the second clothing image; the first generation module, Used to generate the clothing matching compatibility and matching level of the first clothing image and the second clothing image according to the first matching compatibility and the second matching compatibility; a second generation module used to generate the clothing matching compatibility and matching level according to the first clothing image. The clothing matching compatibility and matching level of the clothing image and the second clothing image are used to generate a clothing matching combination.

An embodiment of the present disclosure provides an electronic device, including: at least one processor; a storage device configured to store at least one program, and when the at least one program is executed by the at least one processor, the at least one processor Implement any of the above methods for intelligent matching of clothing.

Embodiments of the present disclosure provide a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program implements any of the above methods for intelligent matching of clothing.

The method for intelligent matching of clothing provided by the embodiment of the present disclosure obtains the feature values of each clothing element of the first clothing image and the second clothing image by performing feature extraction on the first clothing image and the second clothing image of different categories, The clothing matching compatibility and matching level of the first clothing image and the second clothing image are automatically generated based on the matching compatibility between various clothing elements, so that the reason for generating the clothing matching compatibility and matching level can be localized to specific clothing elements, thereby This makes the generated clothing matching compatibility and matching level more accurate; in addition, the clothing matching effect in the clothing matching combination automatically generated based on the clothing matching compatibility and matching level of the first clothing image and the second clothing image is better, improving the generation The efficiency of clothing matching and combination, thereby improving user experience.

It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and do not limit the present disclosure.

Description of the drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure. Obviously, the drawings in the following description are only some embodiments of the present disclosure. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without exerting creative efforts.

FIG. 1 shows a schematic diagram of an exemplary system architecture to which the method for intelligent matching of clothing according to the embodiment of the present disclosure can be applied.

FIG. 2 is a flow chart of a method for intelligent matching of clothing according to an exemplary embodiment.

FIG. 3 is a schematic structural diagram of a fashion compatibility model according to an exemplary embodiment.

FIG. 4 is a schematic diagram showing the application of the fashion compatibility model according to an exemplary embodiment.

FIG. 5 is a schematic diagram of a clothing database with discrimination labels according to an example.

FIG. 6 is a flow chart of a method for intelligent matching of clothing according to an exemplary embodiment.

FIG. 7 is a schematic diagram illustrating determining physical characteristic tags suitable for clothing matching combinations according to an example.

Figure 8 is a schematic diagram of multi-level virtual dressing according to an example.

Figure 9 is a schematic diagram of an intelligent clothing matching system according to an exemplary embodiment.

Figure 10 is a block diagram of a device for intelligent matching of clothing according to an exemplary embodiment.

Figure 11 is a schematic structural diagram of an electronic device according to an exemplary embodiment.

Detailed ways

Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in various forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concepts of the example embodiments. To those skilled in the art. The described features, structures or characteristics may be combined in any suitable manner in one or more embodiments.

Furthermore, the drawings are merely schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and/or processor devices and/or microcontroller devices.

Furthermore, in the description of the present disclosure, "plurality" means at least two, such as two, three, etc., unless otherwise expressly and specifically limited. The terms “first” and “second” are used for descriptive purposes only and shall not be understood as indicating or implying relative importance or implicitly indicating the quantity of indicated technical features. Therefore, features defined as "first" and "second" may explicitly or implicitly include one or more of these features.

As shown in Figure 1, the system architecture may include a server 101, a network 102 and a terminal device 103. The network 102 is a medium used to provide a communication link between the terminal device 103 and the server 101 . Network 102 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

In this embodiment of the present disclosure, the server 101 may be a server that provides various services, such as a background management server that provides support for devices operated by users using the terminal device 103 . The background management server can analyze and process the received request and other data, and feed back the processing results to the terminal device 103 .

In the embodiment of the present disclosure, the terminal device 103 may be a mobile phone, a tablet computer, an e-book reader, smart glasses, a smart home device, an AR (Augmented Reality, augmented reality) device, a VR (Virtual Reality, virtual reality) device, etc. The terminal, or the terminal 120, may also be a personal computer (Personal Computer, PC), such as a laptop computer, a desktop computer, etc.

In this embodiment of the present disclosure, the server 101 may, for example: obtain a first clothing image of a first clothing category and a second clothing image of a second clothing category; perform feature extraction on the first clothing image and the second clothing image respectively to obtain the first clothing image. The characteristic value of the first clothing element of the clothing image and the characteristic value of the second clothing element of the first clothing image, and the characteristic value of the first clothing element of the second clothing image and the characteristic value of the second clothing element of the second clothing image ; Determine a first matching compatibility between the characteristic value of the first clothing element of the first clothing image and the characteristic value of the first clothing element of the second clothing image, and determine the characteristic value of the second clothing element of the first clothing image and the second matching compatibility between the feature values of the second clothing element of the second clothing image; generate the first clothing image based on the first matching compatibility and the second matching compatibility The clothing matching compatibility and matching level of the first clothing image and the second clothing image; generating a clothing matching combination based on the clothing matching compatibility and matching level of the first clothing image and the second clothing image.

In this embodiment of the present disclosure, the server 101 may, for example, send the generated clothing matching combination to the terminal device 103 for display.

In the embodiment of the present disclosure, the target object can, for example, input his or her own physical characteristic data through the terminal device 103; the server 101 can, for example, obtain the physical characteristic data of the target object; input the physical characteristic data of the target object into a clothing matching model to obtain a suitable The target clothing matching combination of the target object; for example, the server 101 can send the target clothing matching combination to the terminal device 103 for display, for the target object to view and select.

It should be understood that the number of terminal equipment 103, network 102 and server 101 in Figure 1 is only illustrative. The server 101 can be a physical server, a server cluster composed of multiple servers, or a cloud server. You can have any number of end devices, networks, and servers based on actual needs.

Below, each step of the method for intelligent matching of clothing in the exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings and embodiments.

As shown in Figure 2, the method for intelligent matching of clothing provided by the embodiment of the present disclosure may include the following steps.

In step S202, a first clothing image of a first clothing category and a second clothing image of a second clothing category are obtained.

In the embodiment of the present disclosure, clothing can be clothing, accessories, shoes, etc.; clothing can be tops, pants, skirts, dresses, etc.; accessories can be glasses, earrings, necklaces, bracelets, hats, bags (handbags, shoulder bags, etc.) , backpack, cross-body bag), etc., the shoes can be leather shoes, sports shoes, canvas shoes, boots, etc., but the disclosure is not limited thereto.

In the embodiment of the present disclosure, the clothing category is used to represent a specific category of clothing, such as tops, bottoms, shoes, bags, accessories, etc.

In the embodiment of the present disclosure, multiple clothing images can be obtained, in which each clothing image corresponds to a different clothing category. Multiple clothing corresponding to these multiple clothing images can form a clothing matching combination.

For example, you can obtain a top image, a bottom image, a bag image, a shoe image, and an accessory image, and the clothing corresponding to these images can form a clothing matching combination.

In the following examples, the description is based on the example of acquiring two clothing images (ie, the first clothing image and the second clothing image) and processing the two clothing images, but the disclosure is not limited thereto.

In the embodiment of the present disclosure, the clothing category of the first clothing image is different from the clothing category of the second clothing image. The first clothing corresponding to the first clothing image and the second clothing corresponding to the second image can form a clothing matching combination.

For example, the first clothing image of the first clothing category is a top image, and the second clothing image of the second clothing category is a bottom clothing image.

Both the first clothing image and the second clothing image in the embodiment of the present disclosure may be images of single items of clothing, or may be images of clothing combinations, which the disclosure is not limited to.

In step S204, feature extraction is performed on the first clothing image and the second clothing image respectively to obtain the feature value of the first clothing element of the first clothing image, the feature value of the second clothing element of the first clothing image, and the second clothing element. Clothing pictures The characteristic value of the first clothing element of the image and the characteristic value of the second clothing element of the second clothing image.

Among them, clothing elements may include but are not limited to color, print, material, silhouette and design.

In the embodiment of the present disclosure, the feature values of multiple clothing elements of the first clothing image and the second clothing image can be extracted respectively. The following description takes the extraction of two clothing elements (ie, the first clothing element and the second clothing element) as an example. , but the present disclosure is not limited thereto.

In the embodiment of the present disclosure, the fashion compatibility model can be used to perform feature extraction on the first clothing image to obtain the feature value of the first clothing element of the first clothing image and the feature value of the second clothing element of the first clothing image; you can use The fashion compatibility model performs feature extraction on the second clothing image, and obtains the feature value of the first clothing element of the second clothing image and the feature value of the second clothing element of the second clothing image.

For example, if the first clothing image of the first clothing category is a top image, the second clothing image of the second clothing category is a bottom image, the first clothing element is color, and the second clothing element is print, then feature extraction is performed on the top image , it can be obtained that the color of the top image is red and the print is plant printing; by feature extraction of the bottom image, it can be obtained that the color of the bottom image is orange and the print is floral print.

For example, referring to Figure 3, a fashion compatibility model (also known as an interpretable deep neural network model) may include a feature extraction network and an internal factor compatibility network, where the feature extraction network is used to extract features from clothing images to obtain The eigenvalues of clothing elements of clothing images, and the internal factor compatibility network is used to analyze the compatibility between clothing elements of clothing images.

Continuing to refer to Figure 3, the top image 301 and the bottom image can be input into the feature extraction network of the fashion compatibility model. The feature extraction network characterizes the color elements, printing elements, material elements, outline elements and design elements of the top image 301. Extract and obtain the color information 3011, printing information 3012, material information 3013, outline information 3014 and design information 3015 of the top image 301; the feature extraction network extracts the color elements, printing elements, material elements, outline elements and design elements of the bottom image 302 Feature extraction is performed to obtain the color information 3021, print information 3022, material information 3023, outline information 3024 and design information 3025 of the underwear image 302.

In step S206, a first matching compatibility between the characteristic value of the first clothing element of the first clothing image and the characteristic value of the first clothing element of the second clothing image is determined, and the second clothing of the first clothing image is determined. Second matching compatibility between the feature value of the element and the feature value of the second clothing element of the second clothing image.

In the embodiment of the present disclosure, the first matching compatibility and the second matching compatibility are used to evaluate the degree of suitable matching between clothing elements, and the clothing matching compatibility is used to evaluate the degree of suitable matching between clothing elements.

Among them, the first matching compatibility, the second matching compatibility and the clothing matching compatibility can all be expressed by numerical values. The larger the value, the more suitable the clothing elements or clothing are to be matched together.

For example, the collocation compatibility between the color of the upper garment image and the color of the lower garment image may be determined as the first collocation compatibility, and the collocation compatibility between the print of the upper garment image and the print of the lower garment image may be determined as the second collocation compatibility. sex.

For example, if red and orange are suitable for matching together, it can be determined that the first matching compatibility between the color of the top image and the color of the bottom image is larger; if plant printing and floral printing are not suitable for matching together, then Can be sure The numerical value corresponding to the second matching compatibility between the printing of the upper garment image and the printing of the lower garment image is determined to be smaller.

Continuing to refer to Figure 3, the color information 3011 of the top image 301 and the color information 3021 of the bottom image 302 are input to the internal factor compatibility network, and the first color information 3011 of the top image 301 and the color information 3021 of the bottom image 302 are obtained. Compatibility 303; input the printing information 3012 of the top image 301 and the printing information 3022 of the bottom image 302 into the internal factor compatibility network to obtain the second compatibility of the printing information 3012 of the top image 301 and the printing information 3022 of the bottom image 302 304; Input the material information 3013 of the top image 301 and the material information 3023 of the bottom image 302 into the internal factor compatibility network to obtain the third compatibility of the material information 3013 of the top image 301 and the material information 3023 of the bottom image 302 305; Input the outline information 3014 of the top image 301 and the outline information 3024 of the bottom image 302 into the internal factor compatibility network to obtain the fourth compatibility 306 of the outline information 3014 of the top image 301 and the outline information 3024 of the bottom image 302 ; Input the design information 3015 of the top image 301 and the design information 3025 of the bottom image 302 into the internal factor compatibility network to obtain the fifth compatibility 307 of the design information 3015 of the top image 301 and the design information 3025 of the bottom image 302.

In step S208, the clothing matching compatibility and matching level of the first clothing image and the second clothing image are generated based on the first matching compatibility and the second matching compatibility.

In the embodiment of the present disclosure, the matching level is a level used to evaluate suitable matching between clothing items. For example, the matching level may include three levels: good, medium, and poor.

In the embodiment of the present disclosure, the clothing matching compatibility between the first clothing image and the second clothing image, and the clothing matching compatibility between the first clothing image and the second clothing image can be generated based on the matching compatibility between the clothing elements of the first clothing image and the second clothing image. 2. Matching level of clothing images.

Specifically, for example, the average of the matching compatibility between the clothing elements of the first clothing image and the second clothing image can be used as the clothing matching compatibility of the first clothing image and the second clothing image; for example, each of the clothing matching compatibility can also be Set weights for the clothing elements, and calculate the clothing matching compatibility of the first clothing image and the second clothing image based on the matching compatibility and weight of each clothing element of the first clothing image and the second clothing image; for example, fashion compatibility can also be used The sex model generates clothing matching compatibility of the first clothing image and the second clothing image.

Specifically, the clothing matching compatibility can be represented by a numerical value. For example, the clothing matching compatibility can be a numerical value between 0 and 1; the first clothing image and the second clothing image are determined according to the clothing matching compatibility of the first clothing image and the second clothing image. The matching level of the second clothing image, for example, when the clothing matching compatibility falls within the first numerical range, the matching level is the first matching level, and when the clothing matching compatibility falls within the second numerical range, the matching level is the second matching level.

Continuing to refer to FIG. 3 , for example, the first compatibility 303 of the color information 3011 of the upper garment image 301 and the color information 3021 of the lower garment image 302 may be based on the second compatibility 302 of the printing information 3012 of the upper garment image 301 and the lower garment image 302 . Compatibility 304, the third compatibility 305 of the material information 3013 of the upper garment image 301 and the material information 3023 of the lower garment image 302, the fourth compatibility 306 of the outline information 3014 of the upper garment image 301 and the outline information 3024 of the lower garment image 302, And the fifth compatibility 307 of the design information 3015 of the upper garment image 301 and the design information 3025 of the lower garment image 302 determines the clothing compatibility and matching level of the upper garment image 301 and the lower garment image 302, wherein the upper garment image 301 and the lower garment image 302 The matching level of 302 can be "poor".

In an exemplary embodiment, the method may further include: based on the first collocation compatibility, the second collocation compatibility and the service According to the compatibility of clothing matching, the target clothing element is determined from the first clothing element and the second clothing element; the reason for generating the matching level is obtained based on the target clothing element and the matching level.

In the embodiment of the present disclosure, the target clothing element may be the clothing element that has the greatest impact on clothing matching compatibility among various clothing elements.

Specifically, among the matching compatibility corresponding to each clothing element of the first clothing image and the second clothing image, the clothing element corresponding to the matching compatibility with the largest difference in clothing matching compatibility can be used as the target clothing element.

For example, continuing to refer to Figure 3, if the difference between the second compatibility 304 of the printing information 3012 of the top image 301 and the printing information 3022 of the bottom image 302 is the largest and the clothing matching compatibility, then the printing element is used as the target clothing element.

In the embodiment of the present disclosure, the specific reason for generating the matching level can be obtained based on the determined target clothing element and matching level.

For example, continuing to refer to Figure 3, the matching level of the top image 301 and the bottom image 302 is "poor", and the reason 308 for generating the matching level "poor" can be traced to: "The printing of the top image and the printing of the bottom image do not match. ".

In this disclosed embodiment, each clothing element corresponds to an internal factor compatibility network. Taking clothing element A as an example, its corresponding internal factor compatibility network can output clothing element A of the first clothing image and the second clothing image. The probability of belonging to each collocation level.

When the clothing matching level (that is, the clothing matching level of the overall image) is "bad", the difference in the positive contribution of each clothing element can be calculated through gradient reasoning: contribu ⁺ _bad -contribu ⁺ _normal , where contribu ⁺ _bad represents the probability that the internal factor compatibility network output is "bad", contribu ⁺ _normal represents the probability that the internal factor compatibility network output is "medium", that is, calculates the probability that the internal factor compatibility network output corresponding to each clothing element is poor. With the probability that the output is medium, the clothing element corresponding to the largest difference in the positive contribution is selected as the target clothing element. The reason why the overall clothing matching level of the image is "poor" is that the target clothing element does not match.

In the embodiment of the present disclosure, on the basis of automatically generating the clothing matching compatibility and matching level of the first clothing image and the second clothing image, the reason for generating the matching level can also be automatically obtained, that is, the fashion compatibility in the embodiment of the present disclosure. The gender model is an interpretable neural network model that allows users to more intuitively know the reasons for generating matching levels, making it easier to adjust clothing matching later.

In step S210, a clothing matching combination is generated based on the clothing matching compatibility and matching level of the first clothing image and the second clothing image.

In the embodiment of the present disclosure, the first clothing image and the second clothing image can be directly formed into a clothing matching combination according to the clothing matching compatibility and matching level of the first clothing image and the second clothing image, or the second clothing image can be replaced In order to make the third clothing image more suitable for the first clothing image, the first clothing image and the third clothing image are formed into a clothing matching combination.

In an exemplary embodiment, if the matching level of the first clothing image and the second clothing image belongs to the first matching level, the first clothing matching combination is generated based on the first clothing image and the second clothing image.

In an exemplary embodiment, if the matching level of the first clothing image and the second clothing image belongs to the second matching level, a matching suggestion for replacing the second clothing image is generated according to the target clothing element, wherein the first matching level is better than The second matching level; according to the matching suggestions, the third clothing image of the second clothing category is obtained, wherein the target of the third clothing image The clothing element and the target clothing element of the second clothing image are different; the second clothing matching combination is generated based on the first clothing image and the third clothing image.

The first matching level may be "good", for example, and the second matching level may be "poor", for example.

Specifically, if the matching level of the first clothing image and the second clothing image is "good", then the first clothing corresponding to the first clothing image and the second clothing corresponding to the second clothing image can be directly formed into a clothing matching combination.

Specifically, if the matching level of the first clothing image and the second clothing image is "poor", a matching suggestion for replacing the second clothing image can be generated based on the target clothing element (for example, if the target clothing element is "color", then The matching suggestion can be "change the color of the second clothing"). According to the matching suggestion, the second clothing image is replaced with a third clothing image, and the first clothing corresponding to the first clothing image and the third clothing corresponding to the third clothing image are composed. An outfit combination.

Referring to Figure 4, a set of clothing outfits (which may include a first clothing image, a second clothing image, a third clothing image, and a fourth clothing image) can be input into the interpretable fashion compatibility model, and the obtained output result can be Including compatibility score (i.e., the compatibility of the above-mentioned clothing matching), overall evaluation (i.e., the above-mentioned matching grade), reasons (i.e., the reasons for generating the above-mentioned matching grade) and suggestions (i.e., the above-mentioned matching suggestions).

For example, if a set of clothing 401 is input into the fashion compatibility model 404, the output result 405 obtained is: "Compatibility score: 0.24; Overall evaluation: bad; Reason: The color of the bag is caused by the excessive colors of the clothes. The entire outfit has lost its color balance; suggestion: change the bag to black."; For another example, if a set of clothing outfit 402 is input into the fashion compatibility model 404, the output result 406 obtained is; "Compatibility score: 0.59; Overall evaluation: normal; reason: there are no matching errors and no highlights."; For another example, if a set of clothing 403 is input into the fashion compatibility model 404, the output result 407 is: "Compatibility score: 0.87; Overall rating: Good; Reason: The pants and shoes are of the same color and are harmonious, while the bright colors bring a bright atmosphere to the entire outfit."

The explainable fashion compatibility model provided by the embodiments of the present disclosure can automatically generate the clothing matching compatibility and matching level of the first clothing image and the second clothing image, and can also automatically obtain the generation reason and matching of the matching level. Suggestions, that is, on the basis of giving the compatibility score of clothing, a clear judgment can also be given, the reasons for making the conclusion and suggestions for improvement.

In the embodiment of the present disclosure, a deep neural network model can be obtained as an interpretable fashion compatibility model based on data-driven algorithm and corresponding database training.

The training process of the fashion compatibility model will be described below, but the disclosure is not limited thereto.

First, a large amount of clothing combination data can be collected and multi-dimensional attribute annotation can be performed on the clothing combination data. The multi-dimensional attributes can include but are not limited to identification labels, fabrics, colors, prints, clothing categories, etc. Among them, the discriminant label is used to evaluate the degree of suitability of matching between the clothing in the clothing combination. The discriminant label can include but is not limited to "good", "normal", and "poor". The discriminant label can be judged by people with a fashion background. The annotations are obtained, and a database for training the compatibility of fashion items can be constructed based on the clothing combination data and its annotations. The database with discriminant labels is shown in Figure 5, for example.

Based on the constructed database, an interpretable deep neural network framework is designed (refer to Figure 3). The annotated labels and corresponding clothing pictures are input into the deep neural network. After training iterations, the corresponding fashion item compatibility model can be obtained. type.

In some embodiments, on the basis of automatically generating the clothing matching compatibility and matching level of the first clothing image and the second clothing image, the reason for generating the matching level can also be automatically obtained, that is, the fashion compatibility in the embodiments of the present disclosure. The gender model is an interpretable neural network model that allows users to more intuitively know the reasons for generating matching levels, making it easier to adjust clothing matching later.

Based on the embodiment of Figure 2, the method for intelligent matching of clothing provided by the embodiment of Figure 6 may also include the following steps.

In step S602, physical feature tags suitable for the clothing matching combination are determined.

Among them, physical characteristic tags may include but are not limited to skin color, body shape, height, hairstyle, personality, etc.

In the embodiment of the present disclosure, a large number of high-quality clothing matching combinations can be automatically generated according to the method provided in the embodiment of Figure 2. After generating the clothing matching combinations according to the embodiment of Figure 2, it can be determined whether each clothing matching combination is suitable or not. physical characteristics label.

Referring to Figure 7 , the clothing matching combination 701 and the body feature tags can be displayed on the designer's client. For example, the body feature tags can include but are not limited to: Factor 1: Body shape (straight-cylindrical, funnel-shaped with wide top and narrow bottom, hourglass shape, Large frame (sporty, inverted triangle), pear-shaped (triangular, spoon-shaped, hourglass-shaped with narrow top and wide bottom), apple-shaped, (round, diamond-shaped)), factor 2: skin color (yellowish, dark) , fair skin, bronze), Factor 3: Hair style (long curly hair, long straight hair, medium-long curly hair, medium-long straight hair, short curly hair, short straight hair), Factor 4: Hair color (ginger, black, dark brown, light Brown, grey/silver, gold), Factor 5: Height (tall, medium, short), Factor 6: Breasts (large, medium, small), Factor 7: Contrast of hair color and skin tone (high contrast (dark) Hair color/dark eyes & light skin tone; light hair color/light eyes & dark skin tone), low contrast (dark hair color/dark eyes & dark skin tone; light hair color/light eyes & light skin tone)).

The designer can click on the physical feature tag that is suitable for the clothing matching combination 701 on the client, or click the physical feature tag that is not suitable for the clothing matching combination 701, as the physical feature tag of the clothing matching combination 701, for training the clothing matching model .

In step S604, a clothing matching model is obtained through training based on clothing matching combinations and their suitable body feature labels.

In the embodiment of the present disclosure, a deep neural network model can be trained based on each clothing matching combination and its suitable or unsuitable body feature labels to obtain a clothing matching model.

In step S606, the body characteristic data of the target object is obtained.

In the embodiment of the present disclosure, the target object may be, for example, a user of a shopping platform, and the physical characteristic data of the target object may be, for example, the user's skin color, body shape, height, hairstyle, personality, etc.

In step S608, the target object's physical characteristic data is input into the clothing matching model to obtain a target clothing matching combination suitable for the target object.

In this disclosed embodiment, by inputting the user's physical characteristic data into the clothing matching model, a personalized target clothing matching combination suitable for the user can be obtained.

Specifically, although each user's physical characteristic data is different, that is, each user's skin color, body shape, height, hairstyle, etc. are different, the clothing matching model provided by the embodiment of the present disclosure can be a model with various physical characteristics. Users of the data recommend suitable target clothing combinations.

The target clothing matching combination may include multiple target clothing images. The following description takes the target clothing matching combination including two target clothing images (ie, the first target clothing image and the second target clothing image) as an example. However, this disclosure does not Limited to this.

In an exemplary embodiment, the method may further include: generating a virtual image of the target object according to the physical characteristic data of the target object; determining the dressing order of the first target clothing image and the second target clothing image; and converting the first target clothing image into The second target clothing image is fused with the virtual image based on the dressing order to generate the target clothing image.

In the embodiment of the present disclosure, a virtual image of the user can be generated based on the user's physical characteristic data. For example, a virtual image template can be obtained first, and the virtual image template can be deformed according to the user's physical characteristic data (such as height, weight, skin color, etc.) to obtain the user's virtual image. Since the user's virtual image is based on the user's Body feature data is generated, so each user's virtual image is specific and personalized.

In the embodiment of the present disclosure, the dressing order can be set for each target clothing image. For example, the dressing order of the first target clothing image can be set to the first dressing order, and the dressing order of the second target clothing image can be set to For the second dressing order, when fusing the first target clothing image and the second target clothing image with the virtual image based on the dressing order, first fuse the first target clothing image and the virtual image to generate an intermediate fusion image, and then fuse the first target clothing image and the virtual image, and then fuse the first target clothing image and the second target clothing image with the virtual image. The two target clothing images are fused with the intermediate fusion image to generate the target outfit image.

When the target clothing matching combination includes multiple target clothing images, a dressing order is set for each target clothing image, and each target clothing image and the virtual image are fused according to the dressing order to generate a target clothing image. That is, the method provided by the embodiment of the present disclosure can realize multi-level virtual dressing.

In the embodiment of the present disclosure, the above dressing sequence can be preset according to dressing logic, or can be adjusted according to user instructions. The dressing logic may include, for example: wearing a skirt inside a shirt, wearing a jacket outside a vest, wearing a sweater outside a shirt, etc.

For example, the dressing logic can be preset to "wear a shirt first and then a skirt", then first fuse the shirt with the user's virtual image, and then fuse the skirt with the user's virtual image to generate the target outfit image; After generating the target outfit image, if the user wants to adjust the order of outfits, they can select the skirt and drag it, and then the skirt can be superimposed on the outside of the shirt to regenerate the target outfit image.

In the embodiment of the present disclosure, the fusion process of the above target clothing image and the user's virtual image can be performed through deep neural Network model is processed. For example, a deep neural network model can be used to detect key points on the target clothing image and the user's virtual image respectively. Based on the key points obtained through the detection, the target clothing image is superimposed on the user's virtual image to generate a target outfit image.

For example, referring to Figure 8, the target clothing matching combination includes a first target clothing image 801, a second target clothing image 802, a third target clothing image 803, and a fourth target clothing image 804. For example, the order of dressing is the first target clothing image. 801, the third target clothing image 803, the second target clothing image 802 and the fourth target clothing image 804, then the first target clothing image 801, the third target clothing image 803, the second target clothing image 802 and the fourth target clothing image are sequentially The four target clothing images of clothing image 804 are fused with the user's virtual image to generate target outfit image 805.

In an exemplary embodiment, fusing the first target clothing image and the second target clothing image with the virtual image based on the dressing order to generate the target outfit image may include: separately merging the first target clothing image and the second target clothing image. Analyze the clothing structure of the image to obtain the first clothing front image of the first target clothing image and the second clothing front image of the second target clothing image; according to the dressing order, the virtual image is sequentially matched with the first clothing front image The image is fused with the second clothing front image to generate a target outfit image.

Specifically, when the target clothing image is displayed alone, both the front and part of the back part of the clothing may be displayed. However, when the target clothing image and the virtual image are fused, part of the back part may be blocked by the virtual image. That is, in the generated target The above-mentioned back part does not need to be shown in the outfit image. For example, when the round-neck T-shirt image is displayed alone, the back piece around the collar will be displayed. However, when the round-neck T-shirt is worn on the virtual image of the target object, the back piece around the collar will be hidden by the target object. The virtual image is blocked, so before fusing the target clothing image with the virtual image, the segmentation model can be used to segment the target clothing image, that is, to distinguish the front and back pieces of the target clothing image.

In some embodiments, the clothing structure of the first target clothing image can be analyzed to obtain the first clothing front piece image and the first clothing back piece image of the first target clothing image; and the clothing structure of the second target clothing image can be analyzed. Analyze and obtain the second clothing front image and the second clothing back image of the second target clothing image; according to the dressing order, fuse the virtual image with the first clothing front image and the second clothing front image in turn to generate Target outfit images.

In some embodiments, the clothing back piece image of the target clothing image also needs to be superimposed on the virtual image (such as a tuxedo). Then, the first target clothing image is analyzed to obtain the first clothing front piece image and the first clothing back piece image. , and after analyzing the second target clothing image to obtain the second clothing front image and the second clothing back image, the first clothing front image, the first clothing back image, the second clothing front image and the second clothing front image are determined. The sub-dressing order of the two clothing back piece images is based on the sub-dressing order of the first clothing front piece image, the first clothing back piece image, the second clothing front piece image and the second clothing back piece image. The sub-dressing image, the first clothing back piece image, the second clothing front piece image and the second clothing back piece image are sequentially superimposed on the virtual image of the target object to generate a target outfit image.

The method for intelligent matching of clothing provided by the embodiments of the present disclosure is to obtain a clothing matching model based on clothing matching combinations and their suitable body feature labels through training, and input the physical feature data of the target object into the clothing matching model to obtain a suitable target By using the object's target clothing matching combination, you can obtain a personalized target clothing matching combination, obtain a more intuitive matching effect, and provide better styling services for the target object.

In some embodiments, a body image that can intuitively express the target object can be generated based on the physical characteristic data of the target object. The virtual image of the body features makes the generated target outfit image closer to the real effect of the target object wearing the target clothing.

In some embodiments, the target clothing image and the virtual image of the target object are fused according to the dressing order of the target clothing image, which can reflect a multi-level clothing matching effect, thereby enhancing the fashion sense of clothing matching.

Referring to Figure 9, the clothing intelligent matching system provided by the embodiment of the present disclosure may include an interpretable compatibility learning and understanding module of fashion items (ie, the above-mentioned fashion compatibility model), a personalized fashion matching module with aesthetic cognition, and multiple Hierarchical virtual dressing module.

Among them, the interpretable compatibility learning and understanding module of fashion items can perform clothing matching recommendation, clothing matching evaluation and clothing matching generation. For example, the purchased goods can be input into the fashion compatibility model, and the fashion compatibility model can determine the purchased items. The clothing matching compatibility and matching level between the product and each clothing image in the database will recommend clothing with higher clothing matching compatibility with the purchased product to the user who purchased the product. For example, the fashion compatibility model can determine The clothing matching compatibility and matching level of each clothing image in a set of clothing, as well as the reasons for generating the matching level, find unreasonable places and give matching suggestions; for another example, a large number of high-quality can be automatically generated through the fashion compatibility model clothing combinations.

Among them, the body characteristic information input by target objects with different body shapes, different skin colors and different hair colors can be input to a personalized fashion matching module with aesthetic cognition, and a personalized fashion matching suitable for the target object can be output.

Among them, the multi-level virtual dressing module can superimpose clothing matching combinations onto the virtual image of the target object in order. For details, please refer to the embodiment in Figure 6, and the disclosure will not be repeated here.

It should be noted that the above-mentioned drawings are only schematic illustrations of processes included in methods according to exemplary embodiments of the present disclosure, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal sequence of these processes. In addition, it is also easy to understand that these processes may be executed synchronously or asynchronously in multiple modules, for example.

The following are device embodiments of the present disclosure, which can be used to perform method embodiments of the present disclosure. For details not disclosed in the device embodiments of the disclosure, please refer to the method embodiments of the disclosure.

As shown in Figure 10, the device 1000 for intelligent matching of clothing may include an image acquisition module 1002, a feature extraction module 1004, a determination module 1006, a first generation module 1008 and a second generation module 1010.

Among them, the image acquisition module 1002 is used to acquire the first clothing image of the first clothing category and the second clothing image of the second clothing category; the feature extraction module 1004 is used to obtain the first clothing image and the second clothing image respectively. Perform feature extraction to obtain the characteristic value of the first clothing element of the first clothing image, the characteristic value of the second clothing element of the first clothing image, and the characteristic value of the first clothing element of the second clothing image. and the characteristic value of the second clothing element of the second clothing image; the determination module 1006 is used to determine the characteristic value of the first clothing element of the first clothing image and the characteristic value of the first clothing element of the second clothing image. The first matching compatibility, and determining the second matching compatibility between the characteristic value of the second clothing element of the first clothing image and the characteristic value of the second clothing element of the second clothing image; the first generation module 1008 is used to determine the second matching compatibility according to The first matching compatibility and the second matching compatibility are used to generate the clothing matching compatibility and matching level of the first clothing image and the second clothing image; the second generation module 1010 is used to generate the clothing matching compatibility and matching level according to the first clothing image and the second clothing image. The clothing matching compatibility and matching level of the first clothing image and the second clothing image generate a clothing matching combination.

In some exemplary embodiments of the present disclosure, the device 1000 for intelligent matching of clothing may further include: an element determination module configured to determine the first matching compatibility, the second matching compatibility, and the clothing matching compatibility. property, determining a target clothing element from the first clothing element and the second clothing element; and a reason obtaining module, configured to obtain a reason for generating the matching level based on the target clothing element and the matching level.

In some exemplary embodiments of the present disclosure, the second generation module 1010 is configured to, if the matching level of the first clothing image and the second clothing image belongs to the first matching level, generate the first clothing image and the second clothing image according to the first matching level. The two clothing images generate a first clothing matching combination.

In some exemplary embodiments of the present disclosure, the second generation module 1010 is configured to generate, if the matching level of the first clothing image and the second clothing image belongs to the second matching level, according to the target clothing element. Replace the matching suggestions of the second clothing image, wherein the first matching level is better than the second matching level; obtain a third clothing image of the second clothing category according to the matching suggestions, wherein the third clothing image is The target clothing element of the third clothing image is different from the target clothing element of the second clothing image; a second clothing matching combination is generated based on the first clothing image and the third clothing image.

In some exemplary embodiments of the present disclosure, the device 1000 for intelligent matching of clothing may further include: a label determination module for determining suitable body feature labels for the clothing matching combination; and a model training module for determining the appropriate body feature labels for the clothing matching combination. The matching combination and its suitable body feature label training are used to obtain a clothing matching model; the data acquisition module is used to obtain the body feature data of the target object; the combination acquisition module is used to input the body feature data of the target object into the clothing matching model to obtain the target clothing matching combination suitable for the target object.

In some exemplary embodiments of the present disclosure, the target clothing matching combination includes a first target clothing image and a second target clothing image; wherein, the device 1000 for intelligent clothing matching may further include: an image generation module, configured to The physical characteristic data of the target object generates a virtual image of the target object; an order determination module is used to determine the dressing order of the first target clothing image and the second target clothing image; an image fusion module is used to The first target clothing image and the second target clothing image are fused with the virtual image based on the dressing order to generate a target outfit image.

In some exemplary embodiments of the present disclosure, the image fusion module is configured to analyze the clothing structure of the first target clothing image and the second target clothing image respectively, and obtain the first target clothing image of the first target clothing image. The clothing front image and the second clothing front image of the second target clothing image; according to the dressing order, the virtual image is sequentially matched with the first clothing front image and the second clothing front image. The image is fused to generate the target outfit image.

It should be noted that the block diagrams shown in the above figures are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and/or processor devices and/or microcontroller devices.

FIG. 11 is a schematic structural diagram of an electronic device according to an exemplary embodiment. It should be noted that the electronic device shown in FIG. 11 is only an example and should not have any influence on the functions and scope of use of the embodiments of the present disclosure. limit.

As shown in FIG. 11 , the electronic device 1100 includes a central processing unit (CPU) 1101 that can operate according to a program stored in a read-only memory (ROM) 1102 or loaded from a storage portion 1108 into a random access memory (RAM) 1103 And perform various appropriate actions and processing. In the RAM 1103, various programs and data required for the operation of the system 1100 are also stored. CPU 1101, ROM 1102 and RAM 1103 are connected to each other through bus 1104. An input/output (I/O) interface 1105 is also connected to bus 1104.

The following components are connected to the I/O interface 1105: an input section 1106 including a keyboard, a mouse, etc.; an output section 1107 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., speakers, etc.; and a storage section 1108 including a hard disk, etc. ; and a communication section 1109 including a network interface card such as a LAN card, a modem, etc. The communication section 1109 performs communication processing via a network such as the Internet. Driver 1110 is also connected to I/O interface 1105 as needed. Removable media 1111, such as magnetic disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on the drive 1110 as needed, so that a computer program read therefrom is installed into the storage portion 1108 as needed.

In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product including a computer program carried on a computer-readable medium, the computer program containing program code for performing the method illustrated in the flowchart. In such embodiments, the computer program may be downloaded and installed from the network via communication portion 1109 and/or installed from removable media 1111 . When the computer program is executed by the central processing unit (CPU) 1101, the above-described functions defined in the system of the present disclosure are performed.

It should be noted that the computer-readable medium shown in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the above two. The computer-readable storage medium may be, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any combination thereof. More specific examples of computer readable storage media may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard drive, random access memory (RAM), read only memory (ROM), removable Programmed read-only memory (EPROM or flash memory), fiber optics, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above. In this disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code therein. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device . Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the foregoing.

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operations of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagrams may represent a module, segment, or portion of code that contains one or more An executable instruction used to implement specified logical functions. It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown one after another may actually execute substantially in parallel, or they may sometimes execute in the reverse order, depending on the functionality involved. It will also be noted that each block in the block diagram or flowchart illustration, and combinations of blocks in the block diagram or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or operations, or may be implemented by special purpose hardware-based systems that perform the specified functions or operations. Achieved by a combination of specialized hardware and computer instructions.

The units involved in the embodiments of the present disclosure can be implemented in software or hardware. The described unit may also be provided in a processor. For example, it may be described as follows: a processor includes a sending unit, an acquisition unit, a determining unit and a first processing unit. The names of these units do not constitute a limitation on the unit itself under certain circumstances. For example, the sending unit can also be described as "a unit that sends a picture acquisition request to the connected server."

As another aspect, the present disclosure also provides a computer-readable storage medium. The computer-readable storage medium may be included in the electronic device described in the above embodiments; it may also exist independently without being assembled into the electronic device. in electronic equipment. The computer-readable storage medium carries one or more programs. When the one or more programs are executed by an electronic device, the electronic device implements the method described in the above embodiments. For example, the electronic device can implement various steps as shown in Figure 2.

According to one aspect of the present disclosure, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the methods provided in various optional implementations of the above embodiments.

It should be understood that any number of elements in the drawings of the present disclosure is for illustration rather than limitation, and any naming is only for differentiation without any limiting meaning.

Other embodiments of the disclosure will be readily apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptations of the disclosure that follow the general principles of the disclosure and include common common sense or customary technical means in the technical field that are not disclosed in the disclosure. . It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It is to be understood that the present disclosure is not limited to the precise structures described above and illustrated in the accompanying drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the disclosure is limited only by the appended claims.

Claims

A method for intelligent matching of clothing, including:

Obtain the first clothing image of the first clothing category and the second clothing image of the second clothing category;

Perform feature extraction on the first clothing image and the second clothing image respectively to obtain the characteristic value of the first clothing element of the first clothing image and the characteristic value of the second clothing element of the first clothing image, And the characteristic value of the first clothing element of the second clothing image and the characteristic value of the second clothing element of the second clothing image;

Determining a first matching compatibility between the characteristic value of the first clothing element of the first clothing image and the characteristic value of the first clothing element of the second clothing image, and determining the sum of the characteristic values of the second clothing element of the first clothing image the second matching compatibility between the feature values of the second clothing elements of the second clothing image;

According to the first matching compatibility and the second matching compatibility, generate the clothing matching compatibility and matching level of the first clothing image and the second clothing image;

A clothing matching combination is generated based on the clothing matching compatibility and matching level of the first clothing image and the second clothing image.
The method of claim 1, further comprising:

Determine a target clothing element from the first clothing element and the second clothing element according to the first matching compatibility, the second matching compatibility and the clothing matching compatibility;

The reason for generating the matching level is obtained according to the target clothing element and the matching level.
The method according to claim 1 or 2, wherein generating a clothing matching combination based on the clothing matching compatibility and matching level of the first clothing image and the second clothing image includes:

If the matching level of the first clothing image and the second clothing image belongs to the first matching level, a first clothing matching combination is generated based on the first clothing image and the second clothing image.
The method according to claim 3, wherein generating a clothing matching combination according to the clothing matching compatibility and matching level of the first clothing image and the second clothing image further includes:

If the matching level of the first clothing image and the second clothing image belongs to the second matching level, a matching suggestion for replacing the second clothing image is generated according to the target clothing element, wherein the first matching The level is better than the second matching level;

Obtain a third clothing image of the second clothing category according to the matching suggestion, wherein the target clothing element of the third clothing image is different from the target clothing element of the second clothing image;

A second clothing matching combination is generated based on the first clothing image and the third clothing image.
The method of claim 1, further comprising:

Determine the physical characteristic tags suitable for the clothing matching combination;

Obtain a clothing matching model based on training of the clothing matching combinations and their suitable body feature labels;

Obtain the physical characteristic data of the target object;

Input the physical characteristic data of the target object into the clothing matching model to obtain a target clothing matching combination suitable for the target object.
The method according to claim 5, wherein the target clothing matching combination includes a first target clothing image and a second target clothing image;

Wherein, the method also includes:

Generate a virtual image of the target object based on the physical characteristic data of the target object;

Determine the dressing order of the first target clothing image and the second target clothing image;

The first target clothing image and the second target clothing image are fused with the virtual image based on the dressing order to generate a target outfit image.
The method according to claim 6, wherein the first target clothing image and the second target clothing image are fused with the virtual image based on the dressing order to generate a target outfit image, including:

Analyze the clothing structure of the first target clothing image and the second target clothing image respectively to obtain the first clothing front image of the first target clothing image and the second clothing front image of the second target clothing image. Clothing front image;

According to the dressing order, the virtual image is sequentially fused with the first clothing front image and the second clothing front image to generate the target outfit image.
A device for intelligent matching of clothing, including:

An image acquisition module, configured to acquire a first clothing image of a first clothing category and a second clothing image of a second clothing category;

A feature extraction module, configured to perform feature extraction on the first clothing image and the second clothing image respectively, and obtain the characteristic value of the first clothing element of the first clothing image and the second characteristic value of the first clothing image. The characteristic value of the clothing element, the characteristic value of the first clothing element of the second clothing image, and the characteristic value of the second clothing element of the second clothing image;

A determination module for determining a first matching compatibility between the characteristic value of the first clothing element of the first clothing image and the characteristic value of the first clothing element of the second clothing image, and determining the second clothing of the first clothing image. the second matching compatibility between the feature value of the element and the feature value of the second clothing element of the second clothing image;

A first generation module, configured to generate the clothing matching compatibility and matching level of the first clothing image and the second clothing image according to the first matching compatibility and the second matching compatibility;

The second generation module is used to generate a clothing matching combination based on the clothing matching compatibility and matching level of the first clothing image and the second clothing image.
An electronic device, including:

at least one processor;

A storage device configured to store at least one program, so that when the at least one program is executed by the at least one processor, the at least one processor implements the method according to any one of claims 1 to 7.
A computer-readable storage medium having computer-executable instructions stored thereon, wherein the method according to any one of claims 1 to 7 is implemented when the executable instructions are executed by a processor.