Clothing design method and system

Abstract

The invention relates to the technical field of costume design, and discloses a costume design method and system, and the method comprises the steps: building a database of standard costume parts; obtaining keywords and images of user preferences, and learning the keywords by using a machine learning algorithm to obtain user preference features; according to the image, extracting clothes part features of clothes worn by the person in the image; calculating the user preference characteristics and the clothes part characteristics by using a matching algorithm to obtain a matching degree value, and selecting clothes parts from a clothes database according to the matching degree value to obtain an optimal matching scheme; and applying the optimal matching scheme to the model to obtain a dressing model and returning the dressing model to the user. Compared with the prior art, the method has the advantages that the user preferences are learned to obtain the user preference characteristics, and the appropriate clothes are selected from the database according to the preference characteristics to serve as the optimal matching scheme, so that dependence on designers is reduced, and design efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of clothing design, and in particular to a clothing design method and system. Background Art

[0002] The traditional clothing design process relies on the personal experience and professional knowledge of the clothing designer. In today's society with rapid development of information technology, traditional clothing design can no longer meet people's needs. The development trend of the clothing design industry is to realize the automation and intelligence of clothing design and break the traditional clothing design method. A clothing product usually includes different clothing parts, such as: collar, sleeves, main body, etc. Clothing parts make up the whole clothing, express clothing style, and determine clothing category.

[0003] Computer-aided design is now used in clothing design. In the existing design systems or methods, in the clothing design process, the plane and three-dimensional drawings of clothing are generally made according to the clothing component samples, and clothing components of different specifications are made on the basis of the clothing component samples. The clothing components of different specifications are spliced ​​to form clothing patterns, and paper patterns are made according to the specifications of the clothing patterns; or the size data of various parts of the human body are measured using an anthropometer, and technical specifications are calculated; then the calculation results are derived into three-dimensional cutting pieces; and paper patterns are made using cutting piece samples. Whether it is deriving specifications after the sample is made or deriving the sample after the human body parameters are measured, it involves cumbersome and complex technical specifications and process operations.

[0004] At present, each clothing design process needs to go through a complex process of drawing, cutting and sewing. This process is highly dependent on manual operation and the experience and judgment of clothing designers, resulting in low production efficiency and difficulty in responding to market changes flexibly. This production process is not only complicated and time-consuming, but also overly dependent on the experience and knowledge of clothing designers. In addition, different clothing styles and designs require re-making samples, which is difficult to adapt to the changing clothing market, greatly limiting the design efficiency and production efficiency. Summary of the invention

[0005] The present invention provides a clothing design method and system to solve the problem of over-reliance on the designer's experience and knowledge and low design efficiency in the clothing design process.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides a clothing design method, comprising: Establish a clothing database containing standard clothing parts; Obtaining keywords and images preferred by users, and using machine learning algorithms to learn the keywords to obtain user preference features; Extracting features of clothing parts of clothing worn by a person in the image according to the image; Using a matching algorithm to calculate the user preference features and the clothing component features to obtain a matching degree value, and selecting standard clothing components from the clothing database according to the matching degree value to obtain an optimal matching solution; The best matching solution is applied to a preset model model to obtain a clothing model model, and the clothing model model is returned to the user.

[0007] In an optional embodiment, the step of establishing a clothing database including standard clothing components comprises: Obtain clothing pictures and clothing feature information; Classify the clothing according to the clothing picture and the clothing feature information to obtain clothing category information; Encode and record each of the clothing category information to obtain the number of clothing categories and a clothing database containing standard clothing components.

[0008] In an optional implementation, the using of a machine learning algorithm to learn the keywords to obtain user preference features includes: The keyword is used as a feature vector, and the feature vector is learned using a machine learning algorithm to obtain a learning result; wherein the keyword includes a preferred version element and a preferred texture element; A preference histogram is established for each keyword according to the learning result, and the user preference feature is obtained according to the preference histogram.

[0009] In an optional implementation, establishing a preference histogram for each keyword according to the learning result, and obtaining the user preference feature according to the preference histogram includes: Analyze the learning results to obtain the probability of occurrence of each keyword and establish a preference histogram; According to the preference histogram, the relevance between the user and each of the keywords is recorded to obtain the user preference characteristics; Wherein, the preference histogram is established according to the following formula: in, represents the preference histogram, represents the number of the keywords, Respectively represent The probability of a keyword appearing.

[0010] In an optional implementation, extracting, based on the image, features of clothing parts of clothing worn by a person in the image, includes: Using a semantic segmentation algorithm, extracting key points of clothes and human body from the image, and calculating vectors between the key points to obtain semantic vectors; The semantic vector is transformed to generate a feature vector in a high-dimensional space, and the feature vector is integrated to obtain clothing component features.

[0011] In an optional implementation, the using of a semantic segmentation algorithm to extract key points of clothing and human body from the image and calculating vectors between the key points to obtain a semantic vector includes: Using a semantic segmentation algorithm, extracting key points of clothes and human body in the image, wherein the key points include corner point coordinates and contour line point coordinates; Using the Euclidean distance measurement method to calculate the vector between the key points, to obtain the semantic vector; The semantic vector is calculated according to the following formula: in, represents the semantic vector, represents the horizontal coordinate of the first key point, represents the horizontal coordinate of the second key point, represents the ordinate of the first key point, Indicates the ordinate of the second key point.

[0012] In an optional implementation, transforming the semantic vector to generate a feature vector in a high-dimensional space includes: Combining all the semantic vectors to form a feature matrix; Calculating the covariance of the feature matrix to obtain a covariance matrix; Performing eigenvalue decomposition on the covariance matrix to obtain the eigenvector; Wherein, the covariance matrix is ​​calculated according to the following formula: in, represents the covariance matrix, represents the number of semantic vectors, represents the feature matrix, Represents the mean matrix of the feature matrix.

[0013] In an optional implementation, the use of a matching algorithm to calculate the user preference features and the clothing component features to obtain a matching degree value, and selecting standard clothing components from a clothing database according to the matching degree value to obtain an optimal matching solution includes: Using the Euclidean distance metric to calculate the distance between the user preference feature and the clothing component feature to obtain a matching degree value; The matching degree value is multiplied and matched with the preference histogram, and standard clothing parts are selected from the clothing database to obtain the best matching solution; The matching degree value is calculated according to the following formula: in, represents the matching degree value, represents the user preference characteristics, represents the characteristics of the clothing component, Indicates the number of clothing categories, Indicates the current Kind of clothing.

[0014] In an optional embodiment, the clothing component features include clothing morphological features and human body morphological features; wherein the clothing morphological features include the style, texture and color of the clothing; and the human body morphological features include the body shape, skin color, face shape and limb proportions of the human body.

[0015] In a second aspect, the present invention provides a clothing design system, comprising: A database building module, used for building a clothing database containing standard clothing parts; A user preference feature learning module, which is used to obtain keywords and images that users prefer, and use a machine learning algorithm to learn the keywords to obtain user preference features; A clothing component feature extraction module, used to extract clothing component features of clothing worn by a person in the image based on the image; The best matching solution calculation module is used to calculate the user preference characteristics and the clothing component characteristics using a matching algorithm to obtain a matching degree value, and select standard clothing components from the clothing database according to the matching degree value to obtain the best matching solution; The data display module is used to apply the optimal matching solution to a preset model model to obtain a clothing model model, and return the clothing model model to the user.

[0016] In a third aspect, the present invention further provides an electronic device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements any one of the above-mentioned clothing design methods when executing the computer program.

[0017] In a fourth aspect, the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the clothing design methods described above.

[0018] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a clothing design method, comprising establishing a clothing database including standard clothing parts; obtaining keywords and images preferred by a user, and using a machine learning algorithm to learn the keywords to obtain user preference features; extracting clothing part features of clothing worn by a person in the image based on the image; using a matching algorithm to calculate the user preference features and the clothing part features to obtain a matching degree value, and selecting standard clothing parts from the clothing database based on the matching degree value to obtain an optimal matching scheme; applying the optimal matching scheme to a preset model model to obtain a dressed model model, and returning the dressed model model to the user.

[0019] The method learns the user's preferences to obtain the user's preference characteristics, and selects suitable clothing from a clothing database containing standard clothing parts as the user's best matching solution based on the preference characteristics, thereby reducing dependence on the clothing designer's experience and knowledge and improving design efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of a clothing design method provided by a first embodiment of the present invention; Figure 2 It is a schematic diagram of the structure of a clothing design system provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] Reference Figure 1 The first embodiment of the present invention provides a clothing design method, comprising the following steps: S1, establish a clothing database containing standard clothing parts; S2, obtaining keywords and images preferred by the user, and using a machine learning algorithm to learn the keywords to obtain user preference features; S3, extracting features of clothing parts of clothing worn by the person in the image according to the image; S4, using a matching algorithm to calculate the user preference features and the clothing component features to obtain a matching degree value, and selecting standard clothing components from the clothing database according to the matching degree value to obtain an optimal matching solution; S5, applying the best matching solution to a preset model model to obtain a clothing model model, and returning the clothing model model to the user.

[0023] It should be noted that traditional clothing design solutions usually rely on the experience and knowledge of designers, which requires long-term practice and training. At the same time, designers sometimes need to communicate frequently with multiple users. This one-to-many, high-frequency communication process inevitably reduces the efficiency of the design process and limits the improvement of efficiency. What is more complicated is that some users who conduct customized design only have conceptual imaginations of ideal clothing combinations and may not be able to accurately express the clothing combinations they want, which is often one of the reasons for increasing clothing design time.

[0024] In the present invention, by using the clothing design system in the computer program, the efficiency in the clothing design process can be greatly accelerated. With the efficient machine learning algorithm in the clothing design system, the user only needs to input enough preference features, and the system can accurately analyze and understand the user's personalized preference features, including but not limited to color preferences, style tendencies, material selection, etc., and then automatically generate a series of best matching solutions that meet the user's expectations and display them to the user. At the same time, during the design process, the system has a high degree of flexibility and interactivity, and the user can modify his or her preference features at any time according to the changes in his or her own needs. The system can also respond instantly and make corresponding adjustments to the best matching solution. This customized and instant feedback mechanism not only greatly accelerates the efficiency of clothing design, but also effectively reduces excessive dependence on designers.

[0025] In step S1, the manufacturer needs to pre-establish a clothing database containing standard clothing parts. It should be noted that the clothing database not only needs to ensure the storage of text data, but also should be able to allow the storage of image data.

[0026] In one embodiment, the establishment of a clothing database containing standard clothing components includes: obtaining clothing pictures and clothing feature information; classifying the clothing according to the clothing pictures and the clothing feature information to obtain clothing category information; encoding and recording each of the clothing category information to obtain the number of clothing categories and the clothing database containing standard clothing components.

[0027] Specifically, when classifying clothing based on the clothing picture and the clothing feature information, a strategy method of picture extraction and template segmentation can be adopted. First, the clothing in the picture is extracted using a picture extraction method, and is split into multiple template components using a template segmentation method and recorded separately. Then, it is combined with the clothing feature information to form the clothing category information. Among them, the template components of the clothing include jackets, shirts, trousers, dresses, pants, tops, trousers, accessories, shoes, and hats; the clothing feature information includes the brand, fabric, price, model, and manufacturer of the current clothing. Finally, after obtaining the clothing category information, each of the clothing category information is numbered according to a unified standard, and then stored in the clothing database in an orderly manner according to the number.

[0028] It should be noted that the number is the unique identification of clothing, which plays an important role in clothing design, production, sales, and inventory management. Therefore, before the uniform standard numbering of each clothing category information is performed, it is necessary to ensure that the number used is unique and easy to identify. At the same time, the standard clothing parts included in the clothing database should be able to be produced by current manufacturers. In addition, the clothing database must also have good scalability to meet the growing personalized needs.

[0029] In step S2, the keywords and images preferred by the user are obtained, and the keywords are learned using a machine learning algorithm to obtain user preference features, including the following steps: S21, the system obtains keywords and images preferred by the user.

[0030] Specifically, the system obtains keywords and image data input by the user. The input data may include the user's body information (such as height, weight), preferred clothing information (such as clothing style, material), and other additional information (such as accepted price, home address). The image data is an example photo of the user's preferred clothing.

[0031] It should be noted that when the system obtains the above information, it needs to make a simple judgment and remove the keywords and images that are not related to clothing design to avoid affecting the subsequent results. In addition, the system should also save the above user personal information and establish a corresponding security mechanism.

[0032] S22, taking the keyword as a feature vector, using a machine learning algorithm to learn the feature vector, and obtaining a learning result; wherein the keyword includes a preferred version element and a preferred texture element.

[0033] Specifically, the system first uses a text quantization algorithm to establish the keywords as feature vectors, and then uses a machine learning algorithm to learn the feature vectors to obtain learning results. Among them, the machine learning algorithm includes a decision tree algorithm and a support vector machine algorithm. The decision tree algorithm recursively divides the feature space, and with the information gain and Gini impurity criteria as a guide, it classifies similar feature vectors into the same category. The support vector machine algorithm is used to find an optimal separation hyperplane in the feature space formed by the feature vectors to maximize the intervals between different categories. The two machine learning algorithms can classify the feature vectors into different categories according to the correlation between the feature vectors, and count the number and probability of their occurrence as the learning results.

[0034] It should be noted that the decision tree algorithm and the support vector machine algorithm have certain requirements for the number of keywords. In particular, when the number of keywords is too small (such as the number of keywords is less than 24), the learning results may be overfitted, that is, too consistent with the special case of the sample, and the generalization ability when facing new data is significantly reduced. Therefore, in order to ensure the generalization of the algorithm during machine learning, it is necessary to ensure that the number of input keywords is as large as possible. At the same time, since the decision tree algorithm and the support vector machine algorithm are relatively traditional algorithms, there may be a slight decrease in computational efficiency and accuracy when processing large-scale data sets or complex pattern recognition tasks. Therefore, when hardware resources are sufficient, the machine learning algorithm here can be replaced with a deep learning algorithm with higher efficiency and accuracy, such as a convolutional neural network.

[0035] S23, establishing a preference histogram for each of the keywords according to the learning result, and obtaining the user preference feature according to the preference histogram.

[0036] In one embodiment, the step of establishing a preference histogram for each keyword according to the learning results, and obtaining the user preference characteristics according to the preference histogram, includes: parsing the learning results to obtain the probability of occurrence of each keyword, and establishing a preference histogram; according to the preference histogram, recording the degree of correlation between the user and each keyword to obtain the user preference characteristics; wherein the preference histogram is established according to the following formula, in, represents the preference histogram, represents the number of the keywords, Respectively represent The probability of a keyword appearing.

[0037] Specifically, the system uses a data processing algorithm to analyze and count the learning results, and obtains the probability of occurrence of each keyword based on the learning results, and then establishes a preference histogram for each keyword that records the correlation between the keyword and the probability of keyword occurrence. Finally, the user preference characteristics are summarized based on the preference histogram and stored in the corresponding file. It should be noted that the preference histogram records the number and probability of occurrence of each keyword, but is stored in text form rather than image form in the program algorithm, and will only be output in image form when the user needs to view it. The user preference characteristics record the clothing preference characteristics of each user, and after the user modifies the input keyword, the system will also make corresponding adjustments to the user preference characteristics.

[0038] Furthermore, after using the formula to calculate the preference histogram, the system will record the degree of correlation between the user and each of the keywords according to the preference histogram, that is, learn and record the user's preferred version elements (such as clothing style and model) and preferred texture elements (such as clothing material and pattern), so as to accurately extract the user's preference characteristics.

[0039] In step S3, the program algorithm extracts the clothing component features of the clothing worn by the person in the image. The clothing component features include clothing morphological features and human body morphological features; the clothing morphological features include the pattern, texture and color of the clothing; and the human body morphological features include the body shape, skin color, face shape and body part proportions of the human body.

[0040] In one embodiment, extracting the features of clothing parts of clothing worn by a person in the image according to the image comprises the following steps: S31, using a semantic segmentation algorithm, extracting key points of clothes and human body from the image, and calculating vectors between the key points to obtain semantic vectors; S32, transforming the semantic vector to generate a feature vector in a high-dimensional space, and integrating the feature vector to obtain clothing component features.

[0041] In step S31, the use of a semantic segmentation algorithm to extract key points of clothes and human body from the image, and calculating the vectors between the key points to obtain a semantic vector includes: using a semantic segmentation algorithm to extract key points of clothes and human body in the image; wherein the key points include corner point coordinates and contour line point coordinates; using a Euclidean distance measurement method to calculate the vectors between the key points to obtain the semantic vector; wherein the semantic vector is calculated according to the following formula, in, represents the semantic vector, represents the horizontal coordinate of the first key point, represents the horizontal coordinate of the second key point, represents the ordinate of the first key point, Indicates the ordinate of the second key point.

[0042] Specifically, the system uses the semantic segmentation algorithm to perform semantic segmentation on the image, thereby extracting the key points of the clothes and the human body in the image, and then uses them to calculate the vectors between the key points as the semantic vectors to participate in subsequent calculations. Among them, the semantic segmentation algorithm is a pixel-level target detection algorithm, which can well distinguish the human body, clothes and environmental background from the image, thereby ensuring the accurate extraction of the key points of the clothes and the human body. After the key points are extracted, the vectors between the key points are calculated respectively to obtain the semantic vectors and temporarily record them. Among them, the semantic vectors calculated by the key points not only contain the shape information of the clothes and the human body, but also imply the relative position and movement trend between objects, which can provide a calculation basis for subsequent recognition and calculation tasks, such as posture estimation and behavior recognition.

[0043] Furthermore, there is a certain degree of correlation between the key points, and the degree of correlation records the degree of influence between the clothes and the human body. For the key points with a low degree of correlation, the semantic vectors formed by these key points have little effect on the subsequent calculation results. Therefore, the system can choose to calculate the degree of correlation between all the key points first when calculating the vectors between the key points, and skip the key points with a degree of correlation less than a certain threshold to establish the semantic vector, thereby reducing unnecessary calculation burden and resource consumption.

[0044] In step S32, the semantic vector is transformed to generate a feature vector in a high-dimensional space, including: combining all the semantic vectors to form a feature matrix; calculating the covariance of the feature matrix to obtain a covariance matrix; performing eigenvalue decomposition on the covariance matrix to obtain the feature vector; wherein the covariance matrix is ​​calculated according to the following formula, in, represents the covariance matrix, represents the number of semantic vectors, represents the feature matrix, Represents the mean matrix of the feature matrix.

[0045] Specifically, the system first combines all the semantic vectors by substituting the coordinate values ​​of the semantic vectors into the feature matrix. The non-zero elements in the feature matrix represent the semantic vectors of the current coordinates, and the zero elements represent that there are no semantic vectors for the current coordinates. Then, the system calculates the mean matrix of the feature matrix, and substitutes the feature matrix and the mean matrix into the formula to calculate the covariance to obtain the covariance matrix. Finally, the covariance matrix is ​​subjected to eigenvalue decomposition to obtain the eigenvector. The system then integrates the eigenvector to obtain the clothing component features and records them in the corresponding files.

[0046] It should be noted that the feature matrix should be a square matrix, that is, the number of rows and columns of the feature matrix should be equal, so that the covariance can be calculated. Therefore, if the feature matrix obtained after combining the semantic vectors is not a square matrix, an appropriate number of zero elements can be added to the feature matrix to make it a square matrix.

[0047] In another embodiment, in order to eliminate the standard deviation in the covariance matrix and prevent the generation of a pseudo solution of the minimum eigenvector, the deviation of the covariance matrix can be corrected by introducing a regularization factor. The covariance matrix is ​​corrected according to the following formula: in, represents the modified covariance matrix, After the correction operation, the covariance matrix can be directly replaced with the corrected covariance matrix, and subsequent calculations can be continued.

[0048] In step S4, the user preference feature and the clothing component feature are calculated using a matching algorithm to obtain a matching degree value, and a standard clothing component is selected from the clothing database according to the matching degree value to obtain an optimal matching solution, including: using a Euclidean distance metric to calculate the distance between the user preference feature and the clothing component feature to obtain a matching degree value; multiplying and matching the matching degree value with the preference histogram, and selecting a standard clothing component from the clothing database to obtain an optimal matching solution; wherein the matching degree value is calculated according to the following formula, in, represents the matching degree value, represents the user preference characteristics, represents the characteristics of the clothing component, Indicates the number of clothing categories, Indicates the current Kind of clothing.

[0049] Specifically, the system first reads the user preference features and clothing component features originally stored in the form of coordinates from the corresponding file, and then directly uses the Euclidean metric to calculate the distance between the two. The distance obtained is regarded as the matching degree value and can be recorded in the file for frequent use. Then the matching degree value is multiplied and matched with the preference histogram to obtain the matching degree value and the number and probability of occurrence of the keywords stored in the preference histogram. Finally, the system uses the Hungarian algorithm to screen from the clothing database to obtain standard clothing components that meet the user's preference characteristics, and record them as the best matching solution and store them in the file. Among them, the best matching solution records the user preference characteristics, the standard clothing components and the matching degree value in the form of a triple. At the same time, the Hungarian algorithm is a combinatorial optimization algorithm that solves the task allocation problem in polynomial time, and can also be replaced by other more suitable task allocation algorithms.

[0050] In another embodiment, the matching algorithm can also be replaced by the following matching algorithm: first, the Gaussian kernel function corresponding to the semantic vector is calculated, and the inner product of the Gaussian kernel function and the preference histogram composed of the user preference features is calculated to obtain a first matching set; then, the Euclidean distances from the semantic vectors of different categories to all the user preference features are calculated respectively, and the first matching set is sorted according to the distance to obtain a second matching set; finally, the first matching set and the second matching set are combined to obtain the best matching solution. This dual screening mechanism not only enhances the accuracy of matching, but also gives the algorithm the ability to flexibly adapt to the preferences of diverse users, so as to construct a more comprehensive and personalized best matching solution.

[0051] In step S5, firstly, a standard three-dimensional model needs to be pre-established for the mannequin model.

[0052] Specifically, the system will obtain the user's body shape features from the clothing component features, and establish a standard three-dimensional model with the same body shape as the current user based on the body shape features, wherein the body shape features include body shape, skin color, face shape and body part proportion information.

[0053] Next, the color RGB values ​​of the corresponding component features are introduced into the standard three-dimensional model.

[0054] Specifically, the system first uses the vertex position of the model as the center point of the Gaussian kernel function, then calculates the mean color value near the vertex position, and finally sets the RGB value of the vertex position as the mean of the Gaussian curve.

[0055] It should be noted that the Gaussian kernel function will obtain different results according to the different coordinates in the three-dimensional space, so a gradient color will be presented on the standard three-dimensional model, so that different parts of the standard three-dimensional model can be more clearly distinguished. This process not only ensures a smooth transition of colors, but also enables different parts of the model to present a more delicate and layered color effect.

[0056] Finally, the corresponding clothing in the best matching solution is placed on the standard three-dimensional model to obtain a clothing model, and the final effect is returned to the user.

[0057] Specifically, the system first obtains the corresponding clothing in the clothing database according to the best matching plan; then places the clothing on the corresponding part of the standard three-dimensional model to obtain a dressing model model; finally, renders the dressing model model to generate a rendering, and returns the rendering to the user for viewing.

[0058] In another embodiment, the user can set the appearance of the rendering, including adding texture, changing material, color, mapping, wireframe, mesh display, filling details, lighting effects, reflection effects, fog effects, background blur, local refresh, zoom, transform copy, perspective and move, and can select the body shape of the model, hide hair and accessories, and the viewing angle during preview.

[0059] To facilitate understanding of the present invention, the working process of the present invention is described below using a common scenario as an example: In this embodiment, the clothing factory plans to use the clothing design system to design personalized clothing for each user and perform customized production. The system aims to reduce the dependence on the knowledge and experience of clothing designers and improve the efficiency of design. The specific implementation steps are as follows: Step 1: The garment processing factory collects statistics on the garments that can be produced in advance, and inputs the garment information data into the garment database for storage.

[0060] Step 2: The user sends his personal clothing preferences, such as gender, body shape, and preferred clothing style, to the clothing factory in the form of text or pictures, and the clothing factory then inputs the user's preference data into the system.

[0061] Step 3: The system uses machine learning algorithms to learn the user's clothing preferences and summarize the user's preference characteristics.

[0062] Step 4: The system extracts clothing component features of the clothing worn by the person in the picture from the user's picture, including information about the clothing and the body.

[0063] Step 5: The system calculates the matching degree between the user and the clothes according to the characteristics of the clothes parts and the user's preference characteristics, and searches for suitable clothes in the clothing database according to the matching degree value to form the best matching plan.

[0064] Step 6: The system applies the best matching solution to the preset model model, obtains the model model and renders it, and finally displays the generated rendering to the user. The user can modify the rendering at any time and regenerate a new rendering.

[0065] Through the above steps, the system can efficiently design personalized clothing for users without relying too much on the designer's experience and knowledge.

[0066] In summary, the present invention provides a clothing design method, including establishing a clothing database containing standard clothing parts; obtaining user-favorite keywords and images, and using a machine learning algorithm to learn the keywords to obtain user-favorite features; extracting clothing part features of the clothing worn by the person in the image according to the image; using a matching algorithm to calculate the user-favorite features and the clothing part features to obtain a matching degree value, and selecting standard clothing parts from the clothing database according to the matching degree value to obtain an optimal matching solution; applying the optimal matching solution to a preset model model to obtain a dressed model model, and returning the dressed model model to the user. The method obtains the user's favorite features by learning the user's preferences, and selects appropriate clothing from the clothing database containing standard clothing parts as the user's best matching solution according to the favorite features, thereby reducing the reliance on the designer's experience and knowledge and improving design efficiency.

[0067] Reference Figure 2 The second embodiment of the present invention provides a clothing design system, comprising: A database building module, used for building a clothing database containing standard clothing parts; A user preference feature learning module, which is used to obtain keywords and images that users prefer, and use a machine learning algorithm to learn the keywords to obtain user preference features; Among them, the user preference feature learning module includes a user information input module, which is used to obtain the user's preference keywords and images input and judge the input; a preference feature extraction module, which is used to use a machine learning algorithm to learn the keywords and record the learning results; a user preference feature extraction module, which is used to extract the current user's preference features from the learning results, summarize them into user preference features and store them.

[0068] A clothing component feature extraction module, used to extract clothing component features of clothing worn by a person in the image based on the image; Among them, the clothing part feature extraction module includes a semantic segmentation module, which is used to extract the key points of the person and the clothes from the image; a semantic vector calculation module, which is used to calculate the semantic vector formed between the key points; a feature vector calculation module, which is used to calculate the feature vector of the matrix obtained by the semantic vector; and a clothing part feature integration module, which is used to summarize the clothing part features according to the feature vector and store them.

[0069] The best matching solution calculation module is used to calculate the user preference characteristics and the clothing component characteristics using a matching algorithm to obtain a matching degree value, and select standard clothing components from the clothing database according to the matching degree value to obtain the best matching solution; Among them, the best matching solution calculation module includes a matching degree calculation module, which is used to use a formula to calculate the user preference characteristics and the clothing component characteristics to obtain a matching degree value; a clothing component selection module, which is used to select standard clothing components from the clothing database according to the matching degree value to obtain the best matching solution and store it.

[0070] The data display module is used to apply the optimal matching solution to a preset model model to obtain a clothing model model, and return the clothing model model to the user.

[0071] Among them, the data display module includes a model generation module, which is used to construct a standard three-dimensional model close to the user's body shape according to the characteristics of the clothing parts; a model model coloring module, which is used to use a Gaussian kernel function to color the standard three-dimensional model; a model model rendering module, which is used to apply the best matching plan to the standard three-dimensional model, obtain a dressed model model and render it, obtain a final rendering, and display the final rendering to the user.

[0072] It should be noted that a clothing design system provided in an embodiment of the present invention is used to execute all process steps of a clothing design method in the above embodiment, and the working principles and beneficial effects of the two correspond one to one, so they will not be repeated here.

[0073] The embodiment of the present invention further provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a clothing design program. When the processor executes the computer program, the steps in the above-mentioned clothing design method embodiments are implemented, such as Figure 1 Alternatively, when the processor executes the computer program, the functions of each module / unit in the above-mentioned system embodiments are realized, such as the user preference feature learning module.

[0074] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0075] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. The electronic device may include more or fewer components than the above components, or may combine certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0076] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electronic device, and uses various interfaces and lines to connect various parts of the entire electronic device.

[0077] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the electronic device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0078] Wherein, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or system capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0079] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0080] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A clothing design method, characterized in that: include: Establish a clothing database containing standard clothing parts; Obtaining keywords and images preferred by users, and using machine learning algorithms to learn the keywords to obtain user preference features; Extracting features of clothing parts of clothing worn by a person in the image according to the image; Using a matching algorithm to calculate the user preference features and the clothing component features to obtain a matching degree value, and selecting standard clothing components from the clothing database according to the matching degree value to obtain an optimal matching solution; The best matching solution is applied to a preset model model to obtain a clothing model model, and the clothing model model is returned to the user.

2. The clothing design method according to claim 1, characterized in that: The step of establishing a clothing database containing standard clothing parts comprises: Obtain clothing pictures and clothing feature information; Classify the clothing according to the clothing picture and the clothing feature information to obtain clothing category information; Encode and record each of the clothing category information to obtain the number of clothing categories and a clothing database containing standard clothing components.

3. The clothing design method according to claim 1, characterized in that: The method of using a machine learning algorithm to learn the keywords to obtain user preference features includes: The keyword is used as a feature vector, and the feature vector is learned using a machine learning algorithm to obtain a learning result; wherein the keyword includes a preferred version element and a preferred texture element; A preference histogram is established for each keyword according to the learning result, and the user preference feature is obtained according to the preference histogram.

4. The clothing design method according to claim 3, characterized in that: The step of establishing a preference histogram for each keyword according to the learning result, and obtaining the user preference feature according to the preference histogram, includes: Analyze the learning results to obtain the probability of occurrence of each keyword and establish a preference histogram; According to the preference histogram, the relevance between the user and each of the keywords is recorded to obtain the user preference characteristics; Wherein, the preference histogram is established according to the following formula: in, represents the preference histogram, represents the number of the keywords, Respectively represent The probability of a keyword appearing.

5. The clothing design method according to claim 1, characterized in that: The step of extracting features of clothing parts of clothing worn by a person in the image according to the image includes: Using a semantic segmentation algorithm, extracting key points of clothes and human body from the image, and calculating vectors between the key points to obtain semantic vectors; The semantic vector is transformed to generate a feature vector in a high-dimensional space, and the feature vector is integrated to obtain clothing component features.

6. The clothing design method according to claim 5, characterized in that: The method uses a semantic segmentation algorithm to extract key points of clothes and human body from the image, and calculates vectors between the key points to obtain semantic vectors, including: Using a semantic segmentation algorithm, extracting key points of clothes and human body in the image, wherein the key points include corner point coordinates and contour line point coordinates; Using the Euclidean distance measurement method to calculate the vector between the key points, to obtain the semantic vector; The semantic vector is calculated according to the following formula: in, represents the semantic vector, represents the horizontal coordinate of the first key point, represents the horizontal coordinate of the second key point, represents the ordinate of the first key point, Indicates the ordinate of the second key point.

7. The clothing design method according to claim 5, characterized in that: The step of transforming the semantic vector to generate a feature vector in a high-dimensional space includes: Combining all the semantic vectors to form a feature matrix; Calculating the covariance of the feature matrix to obtain a covariance matrix; Performing eigenvalue decomposition on the covariance matrix to obtain the eigenvector; Wherein, the covariance matrix is ​​calculated according to the following formula: in, represents the covariance matrix, represents the number of semantic vectors, represents the feature matrix, Represents the mean matrix of the feature matrix.

8. The clothing design method according to claim 3, characterized in that: The method of using a matching algorithm to calculate the user preference features and the clothing component features to obtain a matching degree value, and selecting standard clothing components from a clothing database according to the matching degree value to obtain an optimal matching solution includes: Using the Euclidean distance metric to calculate the distance between the user preference feature and the clothing component feature to obtain a matching degree value; The matching degree value is multiplied and matched with the preference histogram, and standard clothing parts are selected from the clothing database to obtain the best matching solution; The matching degree value is calculated according to the following formula: in, represents the matching degree value, represents the user preference characteristics, represents the characteristics of the clothing component, Indicates the number of clothing categories, Indicates the current Kind of clothing.

9. The clothing design method according to claim 1 or 5, characterized in that: The clothing component features include clothing morphological features and human body morphological features; wherein the clothing morphological features include the style, texture and color of the clothing; and the human body morphological features include the body shape, skin color, face shape and body part proportions of the human body.

10. A clothing design system, characterized in that: include: A database building module, used for building a clothing database containing standard clothing parts; A user preference feature learning module, which is used to obtain keywords and images that users prefer, and use a machine learning algorithm to learn the keywords to obtain user preference features; A clothing component feature extraction module, used to extract clothing component features of clothing worn by a person in the image based on the image; The best matching solution calculation module is used to calculate the user preference characteristics and the clothing component characteristics using a matching algorithm to obtain a matching degree value, and select standard clothing components from the clothing database according to the matching degree value to obtain the best matching solution; The data display module is used to apply the optimal matching solution to a preset model model to obtain a clothing model model, and return the clothing model model to the user.

Images