A clothing customization shopping guide method based on non-measured intelligent size matching

Abstract

The present invention relates to the technical field of clothing matching and customization, and discloses a clothing customization shopping guide method based on non-measurement intelligent size matching, comprising the following steps: S101, obtaining non-measurement basic information input by a user; S102, calculating and correcting the user's net body size based on the non-measurement basic information using a preset physical parameter mapping model; S103, calling a dynamic addition and placement rule library based on clothing category attributes, and dynamically adjusting the net body size in combination with the user's selected style preferences to generate a first finished size; S104, verifying the generated first finished size, and if an abnormality is determined, prompting the user to input specific part adjustment parameters, and correcting the first finished size using a non-linear compensation algorithm to obtain a second finished size; S105, outputting the second finished size to a product shopping guide system and a customization processing system. The present invention reduces the complexity and cost of the measurement process and improves the intelligent level of clothing matching and customization.

Description

Technical Field

[0001] The present invention belongs to the technical field of clothing matching and customization, and in particular relates to a clothing customization shopping guide method based on non-measurement intelligent size matching. Background Art

[0002] With the development of intelligent clothing industry, consumers' demand for personalized customization and accurate size recommendations continues to grow. However, the current clothing shopping guide and customization methods in the market still have many shortcomings, resulting in inaccurate size matching and cumbersome customization processes, which affect the user experience.

[0003] Existing clothing shopping guides primarily rely on fixed size recommendations or analysis of historical purchase data. However, these methods fail to account for individual differences in physical characteristics and brand sizing, making it difficult to accurately match the needs of different users. Furthermore, clothing customization typically relies on manual measurements, which are cumbersome, costly, and inefficient in responding to market demand, hindering both efficiency and widespread adoption. Summary of the Invention

[0004] The present invention provides a clothing customization shopping guide method based on non-measured intelligent size matching, which solves the technical problems in related technologies such as low size recommendation accuracy, difficulty in meeting personalized clothing needs, complex customization and measurement processes, high remote operation costs, separation of ready-made clothing shopping guide and customized production, and difficulty in achieving integrated application.

[0005] The present invention provides a clothing customization shopping guide method based on non-measured intelligent size matching, comprising the following steps:

[0006] S101, obtaining non-physical basic information input by the user, including: height, weight, age and physical parameters, the physical parameters including: waist circumference, shoulder width, chest circumference and arm span;

[0007] S102: Based on the user's non-physical basic information, the user's predicted physical signs are calculated using a preset physical sign parameter mapping model, and the physical sign parameters are corrected and missing parts are supplemented to obtain the net body size. The physical sign parameter mapping model is generated through machine learning training, and the predicted physical signs, net body size, and physical sign parameters are dimensionally consistent.

[0008] S103, calling a dynamic addition rule library based on clothing category attributes, and dynamically adjusting the net body size in combination with the style preference selected by the user to generate a first finished product size;

[0009] S104: Checking the rationality of the generated first finished product size and issuing an early warning based on a first preset judgment rule. If an abnormality is detected, adjusting parameters based on the specific part input by the user and correcting the first finished product size using a nonlinear compensation algorithm to obtain a second finished product size.

[0010] S105, outputting the second finished product size to a target application end, wherein the target application end includes a product shopping guide system and a customized processing system.

[0011] Furthermore, the physical sign parameter mapping model is constructed based on a deep neural network, which includes: an implicit body shape inference module, a gating network structure, and a sub-model coordination mechanism;

[0012] The implicit body shape inference module extracts the user's body mass index based on non-measured basic information. It then implicitly categorizes the user into different body shape categories using a pre-trained clustering algorithm combined with distance inversion, generating preliminary probability vectors for each body shape category. The body shape categories include high fat density, low muscle density, and standard balanced. The gated network structure consists of an input layer, hidden layers, and an output layer.

[0013] The sub-model collaboration mechanism means that each body type category corresponds to an independent deep neural network sub-model, and each sub-model outputs a predicted net body size based on the user's non-measuring basic information.

[0014] Furthermore, the K-Means clustering algorithm combined with distance inversion is used to implicitly divide users into body shape categories. The specific steps include:

[0015] S301, set the number of clusters to 3, and randomly select three center points as body shape categories;

[0016] S302, calculating the Euclidean distance between each user data point and the center point;

[0017] S303: Assign the user to the nearest cluster, update the center point, and recalculate the mean of each cluster;

[0018] S304, repeating steps S302 to S303 until the position variation range of the center point is smaller than a preset threshold;

[0019] S305: Calculate the Euclidean distance between the user's non-body-size basic information and the three cluster centers, and use the distance inversion method to calculate the preliminary probability vector of the user belonging to each body type category. The calculation formula of the distance inversion method is: ,in, represents the probability that the user belongs to the jth cluster center, represents the Euclidean distance from the user to the jth cluster center, and j and k both represent the index of the cluster center.

[0020] Furthermore, the input layer of the gated network structure is used to receive the preliminary probability vector and body mass index of the user belonging to each body type category; the hidden layer includes two fully connected layers; and the output layer is used to generate a weight vector containing the weights of each body type category through an activation function, and the sum of the weights of each body type category is 1.

[0021] Furthermore, the predicted net body size output by each sub-model is combined with the weight vector output by the gated network structure for weighted summation to obtain the predicted physical sign. The calculation formula of the predicted physical sign includes:

[0022] ;

[0023] Among them, CBS represents the predicted physical sign, i represents the body shape index, HFD, LMD and STD represent high fat density type, low muscle density type and standard balanced type respectively. represents the weight of body type i, It represents the predicted value when the body size is i, which is output by the sub-model coordination mechanism.

[0024] Furthermore, the fields of the dynamic addition rule library include: clothing category code, addition part code, style preference code, basic addition coefficient, BMI correction coefficient, age correction coefficient and gender applicability. The calculation formula of the first finished product size is:

[0025] ;

[0026] in, Indicates the first finished product size, Indicates the net body size, Indicates the basic addition coefficient, represents the BMI correction factor, Indicates the preset standard body mass index, represents the age correction coefficient, A represents age, Indicates the preset standard age.

[0027] Furthermore, the specific steps of S104 include:

[0028] S401, when the first finished product size is detected to be abnormal, prompting the user to input the adjustment amount of the adjustment part;

[0029] S402: When the adjustment amount does not exceed the preset adjustment threshold, other parts are corrected based on a first empirical formula, where the first empirical formula is: ,in, Indicates the adjustment amount of the part a that needs to be corrected, Indicates the adjustment amount of the adjustment part b input by the user, Represents the human body scale coefficient, a and b are the index of the adjustment part input by the user and the index of the part to be corrected respectively;

[0030] S403: When the adjustment amount exceeds a preset adjustment threshold, a nonlinear exponential formula is used for correction, wherein the nonlinear exponential formula is: ,in, Indicates the second finished product size, Indicates the maximum compensation limit, represents the exponential convergence coefficient, Indicates adjusting the threshold.

[0031] Furthermore, the product shopping guide system matches market standard clothing sizes based on a brand size normalization model and generates cross-brand size recommendation results.

[0032] Furthermore, the customized processing system converts the second finished product size into clothing pattern data and completes the customized clothing production through automatic cutting equipment.

[0033] The beneficial effects of the present invention are as follows: the present invention only requires users to input non-physical data such as height, weight, and age, which do not require professional tools. A physical sign parameter mapping model is constructed through a deep neural network and clustering algorithm to achieve accurate prediction of net body dimensions. The model uses an implicit body shape inference method to classify body shapes such as high fat density and low muscle density. Combined with the prediction results of multiple sub-models using a gated network with dynamic weighted fusion, this solves the problem of poor adaptability of traditional single models to different body shapes. Compared with methods that rely on manual measurement, the user operation threshold is significantly lowered.

[0034] The present invention calls a dynamic addition rule library based on clothing categories and style preferences, dynamically adjusts the net body size through a formula, and introduces BMI and age correction coefficients to make the version adapt to the physical characteristics and wearing habits of different users. When the user manually adjusts, a nonlinear compensation algorithm is adopted: conventional adjustments correct related parts through the linkage of the human body proportion coefficient, and when the threshold is exceeded, the exponential function is used to suppress the size out of control, taking into account personalized needs and ergonomic rationality. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 The present invention is a flowchart of a clothing customization shopping guide method based on non-measured intelligent size matching. DETAILED DESCRIPTION

[0036] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed solely to enable those skilled in the art to better understand and implement the subject matter described herein, and that the functions and arrangements of the elements discussed may be varied without departing from the scope of this specification. Various examples may omit, substitute, or add various processes or components as needed. In addition, features described with respect to some examples may also be combined in other examples.

[0037] like Figure 1 As shown, a clothing customization shopping guide method based on non-measured intelligent size matching includes the following steps:

[0038] S101, obtaining non-physical basic information input by the user, including: height, weight, age and physical parameters, the physical parameters including: waist circumference, shoulder width, chest circumference and arm span;

[0039] S102: Based on the user's non-physical basic information, the user's predicted physical signs are calculated using a preset physical sign parameter mapping model, and the physical sign parameters are corrected and missing parts are supplemented to obtain the net body size. The physical sign parameter mapping model is generated through machine learning training, and the predicted physical signs, net body size, and physical sign parameters are dimensionally consistent.

[0040] S103, calling a dynamic addition rule library based on clothing category attributes, and dynamically adjusting the net body size in combination with the style preference selected by the user to generate a first finished product size;

[0041] S104: Checking the rationality of the generated first finished product size and issuing an early warning based on a first preset judgment rule. If an abnormality is detected, adjusting parameters based on the specific part input by the user and correcting the first finished product size using a nonlinear compensation algorithm to obtain a second finished product size.

[0042] S105, outputting the second finished product size to a target application end, wherein the target application end includes a product shopping guide system and a customized processing system.

[0043] In one embodiment of the present invention, the user's non-physical basic information is collected through a user interaction interface, and the non-physical basic information includes: height, weight, age, waist circumference, shoulder width, chest circumference and arm span.

[0044] In one embodiment of the present invention, after receiving the user's non-physical basic information, the system pre-processes it. The specific steps include:

[0045] S201, when missing values ​​are detected, prompt the user to complete them;

[0046] S202: prompt and reject input that does not conform to conventional human dimensions, for example, height 50 cm, weight 200 kg;

[0047] S203: normalize the non-body-measurement basic information input by the user.

[0048] In one embodiment of the present invention, a physical sign parameter mapping model is constructed based on a deep neural network, which includes: an implicit body shape inference module, a gated network structure, and a sub-model coordination mechanism;

[0049] The implicit body shape estimation module is used to extract the body mass index based on the user's non-measured basic information. It then implicitly divides the user into different body shape categories using a pre-trained clustering algorithm combined with distance inversion, obtaining a preliminary probability vector for the user belonging to each body shape category. The body shape categories include high fat density, low muscle density, and standard balanced. Specifically, the body mass index is calculated as follows: , where BMI stands for body mass index, W stands for weight, and H stands for height;

[0050] The gated network structure includes: input layer, hidden layer and output layer;

[0051] The sub-model collaboration mechanism means that each body type category corresponds to an independent deep neural network sub-model. Each sub-model outputs a predicted net body size based on the user's non-measured basic information. Each sub-model uses the sample closest to the cluster center during training. For example, the high fat density sub-model is specifically optimized for prediction accuracy for people with a high BMI.

[0052] In one embodiment of the present invention, the K-Means clustering algorithm is combined with distance inversion to implicitly divide users into body shape categories. The specific steps include:

[0053] S301, set the number of clusters to 3, and randomly select three center points as body shape categories;

[0054] S302, calculating the Euclidean distance between each user data point and the center point;

[0055] S303: Assign the user to the nearest cluster, update the center point, and recalculate the mean of each cluster;

[0056] S304, repeating steps S302 to S303 until the position variation range of the center point is smaller than a preset threshold;

[0057] S305: Calculate the Euclidean distance between the user's non-body-size basic information and the three cluster centers, and use the distance inversion method to calculate the preliminary probability vector of the user belonging to each body type category. The calculation formula of the distance inversion method is: ,in, represents the probability that the user belongs to the jth cluster center, represents the Euclidean distance from the user to the jth cluster center, and j and k both represent the index of the cluster center.

[0058] In one embodiment of the present invention, the input layer of the gated network structure is used to receive the preliminary probability vector of the user belonging to each body type category and the body mass index; the hidden layer includes two fully connected layers; the output layer is used to generate a weight vector containing the weight of each body type category through an activation function, and the sum of the weights of each body type category is 1, for example, the weight vector , it means that the weights of the user's high fat density type, low muscle density type and standard balanced type are 0.2, 0.1 and 0.7 respectively, and the user is closer to the standard balanced type.

[0059] In one embodiment of the present invention, the predicted net body size output by each sub-model is combined with the weight vector output by the gated network structure for weighted summation to obtain the predicted vital signs. The calculation formula of the predicted vital signs includes:

[0060] ;

[0061] Among them, CBS represents the predicted physical sign, i represents the body shape index, HFD, LMD and STD represent high fat density type, low muscle density type and standard balanced type respectively. represents the weight of body type i, It represents the predicted value when the body size is i, which is output by the sub-model coordination mechanism.

[0062] In one embodiment of the present invention, a sub-model collaboration mechanism is adopted to construct independent deep neural network sub-models for different body types, and weighted fusion is performed based on a gated network structure to calculate the net body size. Compared with the traditional single model method, this mechanism can dynamically adapt to users of different body types, such as high fat density, low muscle density, and standard balanced types, to ensure that each user obtains a more accurate net body size prediction.

[0063] In one embodiment of the present invention, the present invention compares the physical sign parameters input by the user with the predicted physical signs obtained by the physical sign parameter mapping model. If the absolute value of the deviation exceeds 10%, it is determined to be abnormal, and the physical sign parameters input by the user are corrected to the predicted physical signs and expressed by the net body size.

[0064] In one embodiment of the present invention, the fields of the dynamic addition rule library include: clothing category code, addition part code, style preference code, basic addition coefficient, BMI correction coefficient, age correction coefficient and gender applicability. Specifically, the clothing category code is used to indicate the category of clothing, for example, 001 indicates formal wear, 002 indicates casual wear, 003 indicates sportswear, 004 indicates outdoor wear, etc.; the addition part code is used to indicate the specific addition part, and the addition parts include: chest circumference, waist circumference, shoulder width, arm span, etc.; the style preference code is used to indicate the user's clothing style preference, and the style preferences include: slim, standard and loose; the basic addition coefficient indicates the basic addition ratio under the clothing category, addition part and style preference; BMI indicates the correction coefficient of BMI affecting the part; the age correction coefficient indicates the correction coefficient of age affecting the part; and gender applicability indicates that the data is applicable to men, women or general use; the calculation formula for the first finished size is:

[0065] ;

[0066] in, Indicates the first finished product size, Indicates the net body size, It represents the basic addition coefficient obtained from the dynamic addition rule library based on the clothing category and style preferences selected by the user. Indicates the BMI correction factor, which is used to indicate the degree of influence of BMI on the added area. BMI stands for body mass index. Indicates the preset standard body mass index, such as 22, Indicates the age correction coefficient, which is used to indicate the degree of influence of age on the added part. A indicates age. Indicates the preset standard age, such as 30 years old.

[0067] In one embodiment of the present invention, when calculating the size of the first finished product, not only the attributes of the clothing category are taken into consideration, but the size is also dynamically adjusted in combination with the user's personalized style preferences. By adjusting the addition and subtraction coefficients, the clothing pattern is made more in line with the user's wearing habits. Compared with traditional static size recommendations, this method can flexibly adapt to the personalized needs of different users, reduce the discomfort caused by fixed addition and subtraction rules, improve matching accuracy, and provide users with more accurate and personalized clothing recommendations and customization services.

[0068] In one embodiment of the present invention, the specific steps of S104 include:

[0069] S401: When the first finished product size is determined to be abnormal based on a first preset determination rule, prompting the user to input an adjustment amount for the adjustment part. Specifically, the first preset determination rule includes: comparing the first finished product size with the net body size item by item, and performing a threshold determination based on the deviation value of each physical dimension. When the deviation exceeds the preset threshold range of each item, it is determined to be an abnormal adjustment. The present invention will issue a warning prompt to the user, mark the abnormal part, and guide the user to input the adjustment amount for the adjustment part.

[0070] S402: When the adjustment amount does not exceed the preset adjustment threshold, other parts are corrected based on a first empirical formula, where the first empirical formula is: ,in, Indicates the adjustment amount of the part a that needs to be corrected, Indicates the adjustment amount of the adjustment part b input by the user, Represents the human body scale coefficient, which is extracted from large-scale measurement data. a and b are the index of the adjustment part input by the user and the index of the part that needs to be corrected, respectively;

[0071] S403: When the adjustment amount exceeds a preset adjustment threshold, a nonlinear exponential formula is used for correction, wherein the nonlinear exponential formula is: ,in, Indicates the second finished product size, Indicates the maximum compensation limit, used to prevent clothing size from getting out of control. Represents the exponential convergence coefficient, which is used to adjust the nonlinear convergence speed. Indicates the adjustment threshold, for example, 5 cm.

[0072] In one embodiment of the present invention, step S103 is used to dynamically adjust the user's net body dimensions based on the characteristics of the clothing category and the overall style preferences selected by the user, generating a preliminary first finished size. Step S104 is used to further utilize a nonlinear compensation algorithm based on the first finished size to optimize the user's personalized adjustment parameters based on the user's personalized local or overall fine-tuning needs, ensuring that the adjusted dimensions still conform to human body proportions. A first empirical formula is used to automatically correct relevant parts to prevent local adjustments from causing overall imbalance. When the adjustment exceeds a threshold, an exponential correction formula is utilized to avoid overcompensation and improve wearing comfort. Compared to traditional linear adjustment methods, the present invention can dynamically adapt to user adjustment needs, ensuring that the finished size not only meets personalized preferences but also maintains the structural rationality and ergonomic coordination of the garment.

[0073] In one embodiment of the present invention, the product shopping guide system is configured to receive the second finished product size and match the standard sizes of different brands on the market using a brand size normalization model to generate a cross-brand size recommendation result. The specific steps include:

[0074] S501: Establish a brand size normalization database. Specifically, collect size data from multiple brands and create a brand size conversion relationship table. This database stores clothing size data from different brands and styles, and optimizes the conversion relationship between brand sizes based on large-scale user purchase feedback.

[0075] S502, inputting the second finished product size into the product shopping guide system;

[0076] S503: Based on the brand size normalization model, convert the second finished product size into the recommended size of each brand. Specifically, a size conversion method based on multiple linear regression is used to fit the second finished product size with the historical size data of each brand.

[0077] S504: Recommend appropriate size options to the user based on the normalized mapping relationship between sizes of different brands. For example, if the dimensions of the second finished product are calculated as: waist circumference 82cm, shoulder width 45cm, chest circumference 96cm, and arm span 170cm, the system may recommend: Brand A: size L, Brand B: size M, and Brand C: size L.

[0078] In one embodiment of the present invention, the customized processing system converts the second finished product size into clothing pattern data and completes the customized clothing production through automatic cutting equipment.

[0079] In one embodiment of the present invention, the customized processing system converts the second finished product size into clothing pattern data and completes the customized clothing production through the automatic cutting equipment. The specific steps include:

[0080] S601: Establishing a clothing pattern database. Specifically, pre-stored clothing templates of different patterns, including slim fit, loose fit, standard fit, and other different pattern structures. Each pattern contains basic structural parameters of key parts and can be personalized according to the user's second finished product size.

[0081] S602, performing garment pattern conversion based on the second finished product size, specifically, inputting the second finished product size into a garment pattern conversion model and converting it into a CAD garment pattern making file of the corresponding pattern;

[0082] S603, automatic cutting data is generated by CAD conversion software, and the automatic cutting device reads the cutting data to complete the cutting of the fabric;

[0083] S604, the cut fabric is transferred to the sewing process through the production system for garment production.

[0084] The above describes the embodiments of the present invention, but the present invention is not limited to the above specific implementation methods. The above specific implementation methods are merely illustrative and not restrictive. Ordinary technicians in this field can also make many forms based on the inspiration of this embodiment, all of which are protected by this embodiment.

Claims

1. A clothing customization shopping guide method based on non-measured intelligent size matching, characterized in that: The following steps are involved: S101, obtaining non-physical basic information input by the user, including: height, weight, age and physical parameters, the physical parameters including: waist circumference, shoulder width, chest circumference and arm span; S102: Based on the user's non-physical basic information, the user's predicted physical signs are calculated using a preset physical sign parameter mapping model, and the physical sign parameters are corrected and missing parts are supplemented to obtain the net body size. The physical sign parameter mapping model is generated through machine learning training, and the predicted physical signs, net body size, and physical sign parameters are dimensionally consistent. S103, calling a dynamic addition rule library based on clothing category attributes, and dynamically adjusting the net body size in combination with the style preference selected by the user to generate a first finished product size; The fields of the dynamic addition rule library include: clothing category code, addition part code, style preference code, basic addition coefficient, BMI correction coefficient, age correction coefficient, and gender applicability; The calculation formula for the first finished product size is: ； in, Indicates the first finished product size, Indicates the net body size, It represents the basic addition coefficient obtained from the dynamic addition rule library based on the clothing category and style preferences selected by the user. represents the BMI correction factor, BMI represents body mass index, Indicates the preset standard body mass index, represents the age correction coefficient, A represents age, Indicates the preset standard age; S104: Check the rationality of the generated first finished product size and issue an early warning based on the first preset judgment rule. If an abnormality is detected, adjust the parameters according to the specific part input by the user and correct the first finished product size using a nonlinear compensation algorithm to obtain a second finished product size. The specific steps include: S401, when the first finished product size is detected to be abnormal, prompting the user to input the adjustment amount of the adjustment part; S402: When the adjustment amount does not exceed the preset adjustment threshold, the first empirical formula is used to perform linkage correction on the relevant parts based on the human body proportion relationship, wherein the first empirical formula is: ,in, Indicates the adjustment amount of the part a that needs to be corrected, Indicates the adjustment amount of the adjustment part b input by the user, Represents the human body scale coefficient, a and b are the index of the adjustment part input by the user and the index of the part to be corrected respectively; S403: When the adjustment amount exceeds a preset adjustment threshold, a nonlinear exponential formula is used for correction, wherein the nonlinear exponential formula is: ,in, Indicates the second finished product size, Indicates the maximum compensation limit, represents the exponential convergence coefficient, Indicates adjustment threshold; S105, outputting the second finished product size to a target application end, wherein the target application end includes a product shopping guide system and a customized processing system.

2. A clothing customization and shopping guide method based on non-measured intelligent size matching according to claim 1, characterized in that: The physical sign parameter mapping model is built based on a deep neural network, which includes: an implicit body shape inference module, a gating network structure, and a sub-model coordination mechanism; The implicit body shape inference module extracts the user's body mass index based on non-measured basic information. It then implicitly categorizes the user into different body shape categories using a pre-trained clustering algorithm combined with distance inversion, generating preliminary probability vectors for each body shape category. The body shape categories include high fat density, low muscle density, and standard balanced. The gated network structure consists of an input layer, hidden layers, and an output layer. The sub-model collaboration mechanism means that each body type category corresponds to an independent deep neural network sub-model, and each sub-model outputs a predicted net body size based on the user's non-measuring basic information.

3. The clothing customization and shopping guide method based on non-measured intelligent size matching according to claim 2 is characterized in that: The K-Means clustering algorithm combined with distance inversion is used to implicitly divide users into body shape categories. The specific steps include: S301, set the number of clusters to 3, and randomly select three center points as body shape categories; S302, calculating the Euclidean distance between each user data point and the center point; S303: Assign the user to the nearest cluster, update the center point, and recalculate the mean of each cluster; S304, repeating steps S302 to S303 until the position variation range of the center point is smaller than a preset threshold; S305: Calculate the Euclidean distance between the user's non-body-weighted basic information and the three cluster centers, and use the distance inversion method to calculate the preliminary probability vector of the user belonging to each body type category. The distance inversion method exponentially processes the Euclidean distance from the user to the cluster center and combines it with a normalized calculation method to obtain the probability of the user belonging to each body type category.

4. The clothing customization and shopping guide method based on non-measured intelligent size matching according to claim 2, characterized in that: The input layer of the gated network structure is used to receive the preliminary probability vector and body mass index of the user belonging to each body type category; the hidden layer includes two fully connected layers; and the output layer is used to generate a weight vector containing the weights of each body type category through an activation function, and the sum of the weights of each body type category is 1.

5. The method for guiding clothing customization based on non-measured intelligent size matching according to claim 4, characterized in that: The predicted net body size output by each sub-model is combined with the weight vector output by the gated network structure for weighted summation to obtain the predicted physical signs. Among them, the predicted values ​​of each dimension of the predicted physical signs are output by the sub-models of different body shape categories respectively, and the predicted physical signs are calculated based on the weight ratio of the user in each body shape category.

6. The clothing customization and shopping guide method based on non-measured intelligent size matching according to claim 1, characterized in that: The product shopping guide system matches market standard clothing sizes based on a brand size normalization model and generates cross-brand size recommendation results.

7. The method for guiding clothing customization based on non-measured intelligent size matching according to claim 1, characterized in that: The customized processing system converts the second finished product size into clothing pattern data and completes the customized clothing production through automatic cutting equipment.

Images