CN108829958B - Automatic clothing plate making method and system - Google Patents
Automatic clothing plate making method and system Download PDFInfo
- Publication number
- CN108829958B CN108829958B CN201810558725.0A CN201810558725A CN108829958B CN 108829958 B CN108829958 B CN 108829958B CN 201810558725 A CN201810558725 A CN 201810558725A CN 108829958 B CN108829958 B CN 108829958B
- Authority
- CN
- China
- Prior art keywords
- clothing
- style
- clothes
- collar
- component
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to an automatic plate making method and system for clothing, wherein the plate making method comprises the following steps: firstly, inputting a clothing style design drawing, determining the style of the clothing by automatic identification, then automatically searching and determining the component composition of the clothing style design drawing from a clothing layout prototype database according to the style of the clothing, then determining the types of all the components in the clothing style design drawing one by automatic identification, determining the parameter actual values of all the components by combining the clothing size and an expert knowledge base, further obtaining the effect drawings of all the components by the parameter actual values of all the components, and finally combining the effect drawings of all the components to generate a clothing layout effect drawing. The system using this method is composed of a plurality of modules that enable the method to be smoothly implemented. The pattern supporting parts automatically generated by the clothing plate-making system are automatically decomposed, and the decomposed parts can be independently manufactured and assembled into the pattern clothing, so that the method is suitable for industrial batch production, and the production efficiency of clothing enterprises is greatly improved by using the method of the system.
Description
Technical Field
The invention belongs to the field of garment plate making, and relates to an automatic garment plate making method and system.
Background
The clothing plate-making system is a software system which is popular and mature in clothing CAD systems in the market at present, through an interactive operation interface, a user inputs clothing requirements into a computer, the system can match the most appropriate basic paper pattern from a basic paper pattern library according to the requirements of the user, and then the user only needs to perform the conversion of moving, folding, cutting and unfolding and the like on the basic paper pattern or perform appropriate modification of loose distribution, chest skimming, waist line alignment, edge processing and the like, so that a required paper pattern effect picture is obtained.
Although the existing clothing platemaking system can meet the basic requirements of clothing platemakers, the clothing platemaking system can not meet the requirements of the clothing platemakers more and more along with the development of the clothing towards diversification, upgrading and individuation, and has the following defects:
the software marked with an automatic plate making function on the market is used for locally optimizing an interactive plate making process, and is used for storing a plate making process of a certain style in a computer by using a storage function of the computer, and a user repeatedly executes the process and performs corresponding adjustment when needed to realize semi-automatic generation of a fixed style. This increases the amount of repetitive work for the garment platemaker; and the clothing platemaking system has strong dependence on the experience and professional knowledge of operators and needs frequent repeated operation, so that the clothing production efficiency is not high, and the real automatic output of the paper pattern cannot be realized.
Second, women's clothing needs account for a large proportion of all the clothing, and the style is more. The existing clothing plate-making system can not carry out professional rapid plate-making on women's clothing, and can not help the processing industry to shorten the production period and improve the production efficiency well.
Therefore, the development of a scheme for automatically generating the garment layout, which is suitable for various garment styles and designs and has high production efficiency, is a problem to be solved urgently.
Disclosure of Invention
The invention aims to overcome the problems in the prior art and provides an automatic clothing plate making method and an automatic clothing plate making system which are suitable for various clothing style designs and have high production efficiency. The automatic plate making method and the system for the clothes can be used for designing various clothes styles, have high production efficiency and can better meet the requirements of the market on the type of the women's clothes.
In order to achieve the purpose, the invention adopts the technical scheme that:
an automatic plate making method for clothes includes firstly inputting a clothes style design drawing, determining clothes style types through automatic identification, then automatically searching and determining component composition of the clothes style design drawing from a clothes format prototype database according to the clothes style types, then determining types of all components in the clothes style design drawing one by one through automatic identification, determining parameter actual values of all the components by combining clothes sizes and an expert knowledge base, further obtaining effect drawings of all the components according to the parameter actual values of all the components, and finally combining the effect drawings of all the components to generate a format effect drawing; the combination means that the component types of the effect graphs of each component are identified, whether the types of the components can be completely included in the type effect graphs is judged based on an expert knowledge base, and if the types of the components can be completely included in the type effect graphs, the effect graphs of each component are combined.
For example, a design drawing of an actual shirt is composed of a panel, a collar, and sleeves. Drawing a cut piece pattern effect diagram, a collar pattern effect diagram and a sleeve pattern effect diagram according to the identification of the part types and the actually measured parameters. The 3 component effect maps constitute the final shirt pattern effect map. The combination here does not limit the order of the front and back, and whether the design drawing includes (most) these components is identified in order of the constituent modules. Meanwhile, because the existing books with the standard styles cannot be completely covered, an expert knowledge base is required to be applied.
The clothing size is obtained by calling a clothing style key point identification model corresponding to the clothing style type to determine clothing style key points in a clothing style design drawing, measuring the distance between the key points, carrying out proportion conversion and analyzing by combining an expert knowledge base;
the method for determining the actual value of the parameter comprises the following steps: calling a corresponding clothing part key point identification model according to the types of all parts to determine clothing part key points in a clothing style design drawing, measuring the distance between the key points and carrying out proportion conversion to obtain a parameter measurement value, and finally judging whether the parameter measurement value exceeds an allowable adjustment range, wherein if the parameter measurement value exceeds the allowable adjustment range, a parameter reference value is taken as a parameter actual value, and if the parameter measurement value does not exceed the allowable adjustment range, the parameter measurement value is taken as the parameter actual value;
the allowable adjustment range and the parameter reference value are obtained by being extracted from an expert knowledge base according to the clothing size;
an example of the determination of the actual values of the parameters is as follows:
taking the V-shaped collar as an example, the size data is as follows:
| v-shaped collar (cm) | |
| Depth of collar | 15 |
| Width of collar | 6 |
| Wide collar | 4 |
Corresponding key points are obtained from the garment component key point identification model, and the above parameters are automatically modified by calculating distances, such as actual values which may correspond to 16cm, 6.5cm and 4.cm 8. Drawing formulas of the collar in the expert knowledge base can be drawn according to the new parameters, and whether the values are within the range allowed by the expert knowledge is monitored;
all the key point identification models are obtained by training a hourglass network model by taking the garment style design drawing and the determined key points as input and output; all the key point identification models are established based on a Convolutional Neural Network (CNN), and the structure of the Convolutional Neural Network (CNN) mainly comprises the following components: input layer → buildup layer → pooling layer → … … → pooling layer → connecting layer → output layer; the hourglass network model is a special neural network for identifying key points based on a convolutional neural network;
the clothing model prototype database comprises different clothing style types and components corresponding to the clothing style types, and the expert knowledge base comprises reference values of different clothing sizes and allowable adjustment ranges and parameter reference values of the corresponding components.
In the prior art, the type of clothes is judged manually and then is input into a system. The method for automatically identifying the category and the size adopts an automatic program, and aims to construct a full-automatic identification process, reduce manual operation and improve efficiency.
As a preferred technical scheme:
the automatic plate making method for the clothes is realized by automatically identifying and determining the clothes style type based on a clothes style classification model, wherein the clothes style classification model is obtained by training a convolutional neural network by taking a clothes style design drawing and the determined clothes style type as input and output;
the method is characterized in that the step of automatically identifying and determining the types of all the parts one by one is realized on the basis of a garment part type classification model, and the garment part type classification model is obtained by training a convolutional neural network by taking a garment style design drawing and the determined garment part types as input and output. A training set of garment component type classification models, such as pocket type classification models, is a plurality of garment design drawings containing pocket components.
The method for automatically making the plate of the clothing, wherein the method for constructing the clothing style key point identification model comprises the following steps:
(1) initializing parameters of a hourglass network model;
(2) taking a plurality of garment style design drawings with the same garment style type and calibrated garment style key points I as a training set, and sequentially inputting the garment style design drawings into a hourglass network model;
(3) extracting a clothing style key point II from the clothing style design drawing by the hourglass network model through convolution, relay supervision and deconvolution operations;
(4) calculating error function values of the clothing style key points II and the clothing style key points I;
(5) judging whether the error function value is less than or equal to 1px, if so, outputting the hourglass network model, namely the clothing style key point identification model, otherwise, adjusting the parameters of the hourglass network model by using an optimization function rmsprop and returning to the step (2);
the specific training steps and parameters are as follows: the initial size of the clothing style design drawing in the training set is 256 × 3 (length × width channels), a response drawing 1 with the size of 64 × 3 is obtained through one convolution operation of 7 × 7 convolution kernels and the step length is 2 and one down-sampling (pooling), the response drawing 1 is input into a hourglass network model (the default channel parameter is 256), k key points of the response drawing 2(64 × k) are output, and an effect drawing (256 × k) is obtained through one up-sampling (the inverse process of pooling) and one deconvolution operation of 7 × 7 convolution kernels and the step length is 2.
The optimization function in the network model is rmsprop (learning rate of 2.5 e-4). The loss function is MSE (mean square error), wherein MSE represents the error between the network estimated value and the real calibrated value, and the MSE is ensured to be less than or equal to 1 px.
The method for constructing the clothing component key point identification model comprises the following steps:
(a) initializing parameters of a hourglass network model;
(b) taking a plurality of garment style design drawings with the same garment style type, the same garment component type and calibrated garment component key points III as a training set, and sequentially inputting the garment style design drawings into a hourglass network model;
(c) extracting a clothing style key point IV from the clothing style design drawing by the hourglass network model through convolution, relay supervision and deconvolution operations;
(d) calculating error function values of the clothing style key points IV and the clothing style key points III;
(e) and (b) judging whether the error function value is less than or equal to 1px, if so, outputting the hourglass network model, namely the clothing component key point identification model, otherwise, adjusting the parameters of the hourglass network model by using an optimization function rmsprop and returning to the step (b).
The specific parameters of the garment part key point identification model training are similar to the garment style key point identification model.
The method for automatically making the plate of the garment comprises the following steps: and measuring the distance between the key points, carrying out proportion conversion to obtain a dimension measurement value, comparing the dimension measurement value with a reference value of the garment dimension in an expert knowledge base, wherein the garment dimension with the reference value closest to the dimension measurement value is the garment dimension corresponding to the garment style design drawing. The key points of different styles of clothes are different, and the importance degree of the key points is also different, and only a certain example is listed here for explanation, which is as follows:
taking a shirt as an example, the size data is as follows:
the key points comprise shoulder width points, chest circumference points and waist circumference points, the main degrees of the three key points of the shirt are the chest circumference point, the waist circumference point and the shoulder width point in sequence, firstly, the chest circumference is judged, if the chest circumference value obtained by measurement is 87, the corresponding clothing size is close to the close-fitting size, if the chest circumference value obtained by measurement is 88, the waist circumference is continuously judged, if the chest circumference value obtained by measurement is 71, then, the shoulder width is judged, the chest circumference value obtained by measurement is 39.5, the corresponding clothing size is closest to the fit size, namely, the corresponding clothing size is fit.
In the automatic plate making method for the clothes, the expert knowledge base comprises drawing empirical formulas of all parts.
According to the automatic plate making method for the clothes, the step of obtaining the effect graph of each component from the actual parameter values of each component means that the effect graph of each component is generated by correspondingly inputting the actual parameter values of each component to the drawing empirical formula of each component in sequence.
The following examples of the effect graphs of the components obtained from the parameter actual values of the components are as follows: taking a collar as an example, the initial parameters of a collar are respectively 15, 6 and 4, the actual values of the actually measured parameters are 16, 6.5 and 4.8, and a drawing empirical formula (composed of mathematical parameter equations, and stored in an expert knowledge base, one parameter may correspond to a plurality of mathematical parameter equations) of the collar is taken to generate an effect diagram of the collar.
An automatic platemaking method for clothes as described above, wherein the clothes styles comprise shirts, suits, coats, jackets, sweaters, one-piece dress, dresses, trousers, down jackets, sweaters and coats;
the garment size comprises a fit, a fit and a loose;
each part comprises a front fly, a folding door, a collar shape, a pocket, cuffs, sleeve bodies, sleeve cuffs, a shoulder shape, a lower hem, a male main line division, a knife back seam division, a waist line division, a front-back restoring, a front-back raising, a front pleat, a back pleat, four clothes bodies, six clothes bodies and eight clothes bodies;
the front fly comprises a hidden front fly, a flat front fly, a visible front fly and a biased front fly;
the collar type includes wing collar, rabbet collar, stand collar, lapel and collar next to the shin, wherein: the wing collar is divided into a standard wing collar, a big wing collar and a corner wiping wing collar; the rabbet is divided into a standard rabbet, a sharp-angled rabbet, a right-angled rabbet, an obtuse-angled rabbet and a matchangle rabbet; the stand collar is divided into a common stand collar, a button-free stand collar, a turn-up collar and a cheongsam collar; the lapels are divided into common lapels, lapels with collars, lapels without collars, small lapels, large lapels, T-shirt lapels and round-angle lapels; the lapel is divided into a common lapel, a gun lapel, a Chinese olive lapel and a hairtail lapel; the close-fitting collar is divided into a copper basin collar, a navy collar and a one-piece collar;
the pocket includes bevel pocket and plain end pocket, wherein: the bevel pocket is divided into a bevel angle pocket, a bevel round pocket and a bevel angle smearing pocket; the plain pocket is divided into a plain closed angle pocket, a plain round angle pocket and a plain fillet pocket;
the sleeve body comprises original sleeves, one-piece sleeves, raglan sleeves and angle-inserting sleeves;
the sleeve cuff comprises a double-layer cuff and a single-layer cuff, wherein: the double-layer kaffir is divided into a double-layer right-angle kaffir, a double-layer corner wiping kaffir and a double-layer corner folding kaffir; the single-layer snap comprises a single-layer snap, an adjusting single-layer snap and two single-layer snap, wherein the single-layer snap is divided into a single-layer single-right-angle snap, a single-layer single-button angle wiping snap and a single-layer single-angle folding snap;
the backsplash includes no pleats, double pleats, and shirrs.
The key points of the garment parts differ for different parts, such as for the collar of a shirt:
and (3) turning over a collar: collar width, collar depth, front collar width, front side collar width and back collar height;
a V-shaped collar: the collar depth, the collar width and the collar surface are wide;
a round collar: collar width, collar depth, front collar width, front side collar width, back collar height and entrance guard width;
a business suit collar: collar width, collar depth, back collar height, suit collar corner length;
a butterfly collar: collar width, collar depth, collar height, bow tie length.
The invention also provides a system adopting the automatic plate making method for the clothes, which comprises the following steps:
the clothing style design drawing input module is used for receiving a clothing style design drawing input by a user and is used for subsequent plate making operation;
the clothing style type identification module is used for automatically identifying the clothing style type from the clothing style design drawing;
the clothing component composition extraction module is used for automatically searching and determining the component composition of the clothing style design drawing from the clothing style prototype database according to the clothing style type, and a user can also adjust the component type according to the requirement;
the clothing style prototype database is used for storing different clothing styles and corresponding component compositions;
the clothing style size recognition module is used for calling a clothing style key point recognition model according to the clothing style type and analyzing and determining clothing size corresponding to the clothing style design drawing by combining an expert knowledge base;
the clothing style key point identification model library is used for storing clothing style key point identification models corresponding to different clothing styles;
the clothing component type identification module is used for automatically identifying the type of each component from the clothing design drawing;
the clothing component parameter determining module is used for calling a corresponding clothing component key point identification model according to the types of all the components and determining the actual parameter values of all the components in the clothing style design drawing by combining the clothing size and the expert knowledge base; through the module, a user can appropriately modify parameters of each part of the garment model so that a subsequently generated model effect graph is more in line with the expectation of the user, and if the modification parameter range of the user exceeds the preset range of expert knowledge, the system automatically provides a reasonable modification range for the user to fine tune;
the clothing component key point identification model library is used for storing clothing component key point identification models corresponding to different component types;
the clothing pattern generating module is used for automatically generating a pattern effect diagram according to the actual parameter values of all the parts by combining with the expert knowledge base for the reference of a user and storing the pattern effect diagram to a pattern database;
the expert knowledge base is used for storing reference values of different garment sizes and allowable adjustment ranges and parameter reference values of corresponding parts;
the clothes style design drawing input module is respectively connected with the clothes style type identification module, the clothes style size identification module, the clothes part type identification module and the clothes part parameter determination module, the clothes style type identification module is connected with the clothes part composition extraction module and the clothes style size identification module, the clothes style size identification module is connected with the clothes style key point identification model library and the expert knowledge library, the clothes part composition extraction module is connected with the clothes style prototype database and the clothes part type identification module, the clothes part type identification module is connected with the clothes part parameter determination module, the clothes part parameter determination module is connected with the clothes part key point identification model library, the clothes paper pattern generation module, the expert knowledge library and the clothes style size identification module, and the clothes paper pattern generation module is connected with the expert knowledge library.
As a preferred technical scheme:
the system as described above, further comprising:
the clothing style classification model library is used for storing clothing style classification models;
the clothing component type classification model library is used for storing clothing component type classification models corresponding to different clothing styles;
the pattern database is used for storing the pattern effect diagram generated by the clothing pattern generating module;
the clothing style classification model library is connected with the clothing style type identification module, the clothing component type classification model library is connected with the clothing component type identification module, and the pattern database is connected with the clothing pattern generation module.
Has the advantages that:
(1) the system has the function of automatically making the plate of the clothing style, and only needs to input the clothing style design drawing by a user, and the system can automatically identify the basic information (data such as type, size and the like) in the clothing style design drawing. The automatic generation of the female garment layout can be realized by combining an expert knowledge base, and the generated female garment layout paper pattern can conform to most basic style types and is suitable for professional and non-professional garment plate-making personnel;
(2) the invention supports the automatic characteristic parameter identification function of the clothing style design drawing, and the function utilizes the image identification and machine learning technology to achieve the effect of automatically identifying the style and the size of the clothing. The realization of the function can greatly reduce the mapping time of a garment plate-making professional in the traditional garment plate-making process, and the automatic operation can reduce the error and error rate caused by manual operation;
(3) all fine-tuning means in the invention are based on the expert knowledge base, when the operation of the user exceeds the limited range of the expert knowledge, the system automatically cancels the operation and provides a reasonable interval range, so that the clothing pattern paper pattern is more accurate, conforms to the clothing plate-making principle and is not limited to the personal experience of the clothing plate-making operator;
(4) in the invention, the paper pattern automatically generated by the system supports the function of automatically decomposing the parts, and each decomposed part can be independently manufactured and assembled into the sample garment, so that the method is suitable for industrialized mass production, the production period of products is greatly shortened, and the industrial efficiency of garment enterprises is improved;
(5) the expert knowledge base of the invention supports the continuous expansion of expert knowledge, improves the relevant problems caused by the knowledge lag in the general expert knowledge base, and simultaneously improves the precision rate of the automatic identification module and the efficiency of the paper sample automatic generation module.
Drawings
FIG. 1 is an exemplary illustration of a key point distribution for a shirt of the present invention;
fig. 2 is a connection relationship diagram of modules in the automatic plate making system for clothing of the present invention.
Detailed Description
The invention will be further illustrated with reference to specific embodiments. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and such equivalents may fall within the scope of the present invention as defined in the appended claims.
The invention relates to an automatic clothing plate making method, which comprises the following specific steps:
(1) inputting a clothing style design drawing, and determining the clothing style type through automatic identification, wherein the clothing style type is determined through automatic identification and is realized on the basis of a clothing style classification model, and the clothing style classification model is obtained by training a convolutional neural network by taking the clothing style design drawing and the determined clothing style type as input and output;
(2) automatically searching and determining the component composition of the clothing style design drawing from a clothing style prototype database comprising different clothing style types and corresponding components according to the clothing style types, wherein the clothing style prototype database comprises different clothing style types and corresponding components;
(3) the method comprises the following steps of determining the types of all parts in a clothing style design drawing one by one through automatic identification, and determining the actual parameter values of all parts by combining clothing sizes and an expert knowledge base, wherein the determination of the types of all parts one by one through automatic identification is realized on the basis of a clothing part type classification model, the clothing part type classification model is obtained by training a convolutional neural network by taking the clothing style design drawing and the determined clothing part types as input and output, the expert knowledge base comprises reference values of different clothing sizes, allowable adjustment ranges and parameter reference values of all parts corresponding to the reference values, and drawing experience formulas of all parts, and the process is as follows:
(3.1) determining the types of all the parts in the clothing style design drawing one by one through automatic identification;
(3.2) determining the garment size, which is as follows:
(3.2.1) establishing a clothing style key point identification model;
(a) initializing parameters of a hourglass network model;
(b) taking a plurality of garment style design drawings with the same garment style type and calibrated garment style key points I as a training set, and sequentially inputting the garment style design drawings into a hourglass network model;
(c) extracting a clothing style key point II from the clothing style design drawing by the hourglass network model through convolution, relay supervision and deconvolution operations;
(d) calculating error function values of the clothing style key points II and the clothing style key points I;
(e) judging whether the error function value is less than or equal to 1px, if so, outputting the hourglass network model, namely the clothing style key point identification model, otherwise, adjusting the parameters of the hourglass network model by using an optimization function rmsprop and returning to the step (b);
(3.2.2) determining clothing style key points in the clothing style design drawing by calling a clothing style key point identification model corresponding to the clothing style type, measuring the distance between the key points and carrying out proportion conversion to obtain a dimension measurement value, comparing the dimension measurement value with a reference value of clothing dimensions in an expert knowledge base, wherein the clothing dimension with the reference value closest to the dimension measurement value is the clothing dimension corresponding to the clothing style design drawing;
as shown in fig. 1, the size data of the shirt is as follows:
the key points comprise shoulder width points, chest circumference points and waist circumference points, the main degrees of the three key points of the shirt are the chest circumference point, the waist circumference point and the shoulder width point in sequence, firstly, the chest circumference is judged, if the chest circumference value obtained by measurement is 87, the corresponding clothing size is close to the close-fitting size, if the chest circumference value obtained by measurement is 88, the waist circumference is continuously judged, if the chest circumference value obtained by measurement is 71, then, the shoulder width is judged, the chest circumference value obtained by measurement is 39.5, the corresponding clothing size is close to the fitting size, namely, the corresponding clothing size is fit;
(3.3) determining the allowable adjustment range of the actual values of the parameters of each part, which is as follows:
(3.3.1) establishing a garment component key point identification model;
(a) initializing parameters of a hourglass network model;
(b) taking a plurality of garment style design drawings with the same garment style type, the same garment component type and calibrated garment component key points III as a training set, and sequentially inputting the garment style design drawings into a hourglass network model;
(c) extracting a clothing style key point IV from the clothing style design drawing by the hourglass network model through convolution, relay supervision and deconvolution operations;
(d) calculating error function values of the clothing style key points IV and the clothing style key points III;
(e) judging whether the error function value is less than or equal to 1px, if so, outputting the hourglass network model, namely the clothing part key point identification model, otherwise, adjusting the parameters of the hourglass network model by using an optimization function rmsprop and returning to the step (b);
(3.3.2) calling the corresponding clothing component key point identification models of each component to obtain the allowable adjustment range of the parameter actual value of each component from the expert knowledge base according to the clothing size;
(3.4) obtaining the parameter actual values of each component by combining the allowable adjustment range adjustment of the parameter actual values, specifically:
calling a corresponding clothing part key point identification model according to the types of all parts to determine clothing part key points in a clothing style design drawing, measuring the distance between the key points and carrying out proportion conversion to obtain a parameter measurement value, and finally judging whether the parameter measurement value exceeds an allowable adjustment range, wherein if the parameter measurement value exceeds the allowable adjustment range, a parameter reference value is taken as a parameter actual value, and if the parameter measurement value does not exceed the allowable adjustment range, the parameter measurement value is taken as the parameter actual value;
(4) sequentially correspondingly inputting the actual parameter values of each component into a drawing empirical formula of each component to generate an effect graph of each component;
(5) combining the effect graphs of all the parts to generate a layout effect graph;
the types of clothes in the invention comprise shirts, suits, coats, jackets, sweaters, one-piece dress, short skirts, trousers, down jackets, sweaters and coats;
the garment size comprises body shaping, fit and loose;
each part comprises a front fly, a folding door, a collar shape, a pocket, cuffs, sleeve bodies, sleeve cuffs, a shoulder shape, a hem, a male main line division, a knife back seam division, a waist line division, a front-back recovery gesture, a front-back raising, a front pleat, a back pleat, four clothes bodies, six clothes bodies and eight clothes bodies;
the front fly comprises a hidden front fly, a flat front fly, a visible front fly and a biased front fly;
the collar type comprises a wing collar, an rabbet collar, a stand collar, a lapel collar and a close-fitting collar; the wing collar is divided into a standard wing collar, a big wing collar and a corner wiping wing collar; the rabbet is divided into a standard rabbet, a sharp-angled rabbet, a right-angled rabbet, an obtuse-angled rabbet and a matchangle rabbet; the stand collar is divided into a common stand collar, a button-free stand collar, a turn-up collar and a cheongsam collar; the lapels are divided into common lapels, lapels with collars, lapels without collars, small lapels, large lapels, T-shirt lapels and round-angle lapels; the lapel is divided into a common lapel, a gun lapel, a Chinese olive lapel and a hairtail lapel; the close-fitting collar is divided into a copper basin collar, a navy collar and a one-piece collar;
the pocket includes bevel connection pocket and flat mouthful of pocket, wherein: the bevel pocket is divided into a bevel angle pocket, a bevel round pocket and a bevel angle smearing pocket; the plain pocket is divided into a plain closed angle pocket, a plain round angle pocket and a plain fillet pocket;
the sleeve body comprises original sleeves, one-piece sleeves, raglan sleeves and angle sleeves;
the sleeve cuff comprises a double-layer cuff and a single-layer cuff, wherein: the double-layer kaffir is divided into a double-layer right-angle kaffir, a double-layer corner wiping kaffir and a double-layer corner folding kaffir; the single-layer snap comprises a single-layer snap, an adjusting single-layer snap and two single-layer snap, wherein the single-layer snap is divided into a single-layer single-right-angle snap, a single-layer single-button angle wiping snap and a single-layer single-angle folding snap;
backsheets include no pleats, double pleats and shirrs.
The system adopting the automatic plate making method for the clothes comprises the following steps:
the clothing style design drawing input module is used for receiving a clothing style design drawing input by a user;
the clothing style type identification module is used for automatically identifying the clothing style type from the clothing style design drawing;
the clothing part composition extraction module is used for automatically searching and determining the component composition of a clothing style design drawing from a clothing style prototype database according to the clothing style type;
the clothing style prototype database is used for storing different clothing styles and corresponding component compositions;
the clothing style size recognition module is used for calling a clothing style key point recognition model according to the clothing style type and analyzing and determining clothing size corresponding to the clothing style design drawing by combining an expert knowledge base;
the clothing style key point identification model library is used for storing clothing style key point identification models corresponding to different clothing styles;
the clothing component type identification module is used for automatically identifying the type of each component from the clothing design drawing;
the clothing component parameter determining module is used for calling a corresponding clothing component key point identification model according to the types of all the components and determining the actual parameter values of all the components in the clothing style design drawing by combining the clothing size and the expert knowledge base;
the clothing component key point identification model library is used for storing clothing component key point identification models corresponding to different component types;
the clothing pattern generating module is used for automatically generating a format effect diagram according to the actual parameter values of all the parts by combining with an expert knowledge base;
the expert knowledge base is used for storing reference values of different garment sizes and allowable adjustment ranges and parameter reference values of corresponding parts;
wherein the connection relation diagram of each module is shown in figure 2, the garment style design drawing entry module is respectively connected with the garment style type identification module, the garment style size identification module, the garment component type identification module and the garment component parameter determination module, the garment style type identification module is connected with the garment component composition extraction module and the garment style size identification module, the garment style size identification module is connected with the garment style key point identification model library, the clothing component composition extraction module is connected with the clothing model prototype database and the clothing component type identification module, the clothing component type identification module is connected with the clothing component parameter determination module, the clothing component parameter determination module is connected with the clothing component key point identification model base, the clothing pattern generation module, the expert knowledge base and the clothing style size identification module, and the clothing pattern generation module is connected with the expert knowledge base.
The system, still include:
the clothing style classification model library is used for storing clothing style classification models;
the clothing component type classification model library is used for storing clothing component type classification models corresponding to different clothing styles;
the pattern database is used for storing the pattern effect diagram generated by the clothing pattern generating module;
the clothing style classification model library is connected with the clothing style type identification module, the clothing component type classification model library is connected with the clothing component type identification module, and the paper pattern database is connected with the clothing paper pattern generation module.
Claims (8)
1. An automatic plate making method for clothes is characterized in that: firstly, inputting a clothing style design drawing, determining the style of the clothing by automatic identification, then automatically searching and determining the component composition of the clothing style design drawing from a clothing layout prototype database according to the style of the clothing, determining the types of all components in the clothing style design drawing one by automatic identification, determining the parameter actual values of all the components by combining the clothing size and an expert knowledge base, further obtaining the effect drawings of all the components by the parameter actual values of all the components, and finally combining the effect drawings of all the components to generate a layout effect drawing;
the clothing size is obtained by calling a clothing style key point identification model corresponding to the clothing style type to determine clothing style key points in a clothing style design drawing, measuring the distance between the key points, carrying out proportion conversion and analyzing by combining an expert knowledge base;
the method for determining the actual value of the parameter comprises the following steps: calling a corresponding clothing part key point identification model according to the types of all parts to determine clothing part key points in a clothing style design drawing, measuring the distance between the key points and carrying out proportion conversion to obtain a parameter measurement value, and finally judging whether the parameter measurement value exceeds an allowable adjustment range, wherein if the parameter measurement value exceeds the allowable adjustment range, a parameter reference value is taken as a parameter actual value, and if the parameter measurement value does not exceed the allowable adjustment range, the parameter measurement value is taken as the parameter actual value;
the allowable adjustment range and the parameter reference value are obtained by being extracted from an expert knowledge base according to the clothing size;
all the key point identification models are obtained by training a hourglass network model by taking the garment style design drawing and the determined key points as input and output;
the clothing model prototype database comprises different clothing style types and corresponding parts, and the expert knowledge base comprises reference values of different clothing sizes and allowable adjustment ranges and parameter reference values of corresponding parts;
the types of clothes comprise shirts, suits, coats, jackets, sweaters, one-piece dress, short skirts, trousers, down jackets, sweaters and coats;
the garment size comprises a fit, a fit and a loose;
each part comprises a front fly, a folding door, a collar shape, a pocket, cuffs, sleeve bodies, sleeve cuffs, a shoulder shape, a lower hem, a male main line division, a knife back seam division, a waist line division, a front-back restoring, a front-back raising, a front pleat, a back pleat, four clothes bodies, six clothes bodies and eight clothes bodies;
the front fly comprises a hidden front fly, a flat front fly, a visible front fly and a biased front fly;
the collar type includes wing collar, rabbet collar, stand collar, lapel and collar next to the shin, wherein: the wing collar is divided into a standard wing collar, a big wing collar and a corner wiping wing collar; the rabbet is divided into a standard rabbet, a sharp-angled rabbet, a right-angled rabbet, an obtuse-angled rabbet and a matchangle rabbet; the stand collar is divided into a common stand collar, a button-free stand collar, a turn-up collar and a cheongsam collar; the lapels are divided into common lapels, lapels with collars, lapels without collars, small lapels, large lapels, T-shirt lapels and round-angle lapels; the lapel is divided into a common lapel, a gun lapel, a Chinese olive lapel and a hairtail lapel; the close-fitting collar is divided into a copper basin collar, a navy collar and a one-piece collar;
the pocket includes bevel pocket and plain end pocket, wherein: the bevel pocket is divided into a bevel angle pocket, a bevel round pocket and a bevel angle smearing pocket; the plain pocket is divided into a plain closed angle pocket, a plain round angle pocket and a plain fillet pocket;
the sleeve body comprises original sleeves, one-piece sleeves, raglan sleeves and angle-inserting sleeves;
the sleeve cuff comprises a double-layer cuff and a single-layer cuff, wherein: the double-layer kaffir is divided into a double-layer right-angle kaffir, a double-layer corner wiping kaffir and a double-layer corner folding kaffir; the single-layer snap comprises a single-layer snap, an adjusting single-layer snap and two single-layer snap, wherein the single-layer snap is divided into a single-layer single-right-angle snap, a single-layer single-button angle wiping snap and a single-layer single-angle folding snap;
the backsplash includes no pleats, double pleats, and shirrs.
2. The automatic plate making method for clothes according to claim 1, wherein the determination of the clothes style type through automatic identification is realized based on a clothes style classification model, and the clothes style classification model is obtained by training a convolutional neural network with a clothes style design drawing and the determined clothes style type as input and output;
the method is characterized in that the determination of the types of all parts in the clothing style design drawing one by one through automatic identification is realized on the basis of a clothing part type classification model, and the clothing part type classification model is obtained by training a convolutional neural network by taking the clothing style design drawing and the determined clothing part types as input and output.
3. The automatic plate making method for clothes according to claim 1, wherein the construction steps of the clothes style key point identification model are as follows:
(1) initializing parameters of a hourglass network model;
(2) taking a plurality of garment style design drawings with the same garment style type and calibrated garment style key points I as a training set, and sequentially inputting the garment style design drawings into a hourglass network model;
(3) extracting a clothing style key point II from the clothing style design drawing by the hourglass network model through convolution, relay supervision and deconvolution operations;
(4) calculating error function values of the clothing style key points II and the clothing style key points I;
(5) judging whether the error function value is less than or equal to 1px, if so, outputting the hourglass network model, namely the clothing style key point identification model, otherwise, adjusting the parameters of the hourglass network model by using an optimization function rmsprop and returning to the step (2);
the method for constructing the clothing component key point identification model comprises the following steps:
(a) initializing parameters of a hourglass network model;
(b) taking a plurality of garment style design drawings with the same garment style type, the same garment component type and calibrated garment component key points III as a training set, and sequentially inputting the garment style design drawings into a hourglass network model;
(c) extracting a clothing style key point IV from the clothing style design drawing by the hourglass network model through convolution, relay supervision and deconvolution operations;
(d) calculating error function values of the clothing style key points IV and the clothing style key points III;
(e) and (b) judging whether the error function value is less than or equal to 1px, if so, outputting the hourglass network model, namely the clothing component key point identification model, otherwise, adjusting the parameters of the hourglass network model by using an optimization function rmsprop and returning to the step (b).
4. The automatic plate making method for the clothes as claimed in claim 1, wherein the method for determining the clothes size is as follows: and measuring the distance between the key points, carrying out proportion conversion to obtain a dimension measurement value, comparing the dimension measurement value with a reference value of the garment dimension in an expert knowledge base, wherein the garment dimension with the reference value closest to the dimension measurement value is the garment dimension corresponding to the garment style design drawing.
5. The method of claim 1, wherein the expert knowledge base includes empirical formulas for drawing the components.
6. The automatic plate making method for the clothes according to claim 5, wherein the step of obtaining the effect graph of each component from the actual parameter values of each component is to generate the effect graph of each component by correspondingly inputting the actual parameter values of each component to a drawing empirical formula of each component in sequence.
7. The system for adopting the automatic plate making method for the clothes as claimed in any one of claims 1 to 6, which is characterized by comprising the following steps:
the clothing style design drawing input module is used for receiving a clothing style design drawing input by a user;
the clothing style type identification module is used for automatically identifying the clothing style type from the clothing style design drawing;
the clothing part composition extraction module is used for automatically searching and determining the component composition of a clothing style design drawing from a clothing style prototype database according to the clothing style type;
the clothing style prototype database is used for storing different clothing styles and corresponding component compositions;
the clothing style size recognition module is used for calling a clothing style key point recognition model according to the clothing style type and analyzing and determining clothing size corresponding to the clothing style design drawing by combining an expert knowledge base;
the clothing style key point identification model library is used for storing clothing style key point identification models corresponding to different clothing styles;
the clothing component type identification module is used for automatically identifying the type of each component from the clothing design drawing;
the clothing component parameter determining module is used for calling a corresponding clothing component key point identification model according to the types of all the components and determining the actual parameter values of all the components in the clothing style design drawing by combining the clothing size and the expert knowledge base;
the clothing component key point identification model library is used for storing clothing component key point identification models corresponding to different component types;
the clothing pattern generating module is used for automatically generating a format effect diagram according to the actual parameter values of all the parts by combining with an expert knowledge base;
the expert knowledge base is used for storing reference values of different garment sizes and allowable adjustment ranges and parameter reference values of corresponding parts;
the clothes style design drawing input module is respectively connected with the clothes style type identification module, the clothes style size identification module, the clothes part type identification module and the clothes part parameter determination module, the clothes style type identification module is connected with the clothes part composition extraction module and the clothes style size identification module, the clothes style size identification module is connected with the clothes style key point identification model library and the expert knowledge library, the clothes part composition extraction module is connected with the clothes style prototype database and the clothes part type identification module, the clothes part type identification module is connected with the clothes part parameter determination module, the clothes part parameter determination module is connected with the clothes part key point identification model library, the clothes paper pattern generation module, the expert knowledge library and the clothes style size identification module, and the clothes paper pattern generation module is connected with the expert knowledge library.
8. The system of claim 7, further comprising:
the clothing style classification model library is used for storing clothing style classification models;
the clothing component type classification model library is used for storing clothing component type classification models corresponding to different clothing styles;
the pattern database is used for storing the pattern effect diagram generated by the clothing pattern generating module;
the clothing style classification model library is connected with the clothing style type identification module, the clothing component type classification model library is connected with the clothing component type identification module, and the pattern database is connected with the clothing pattern generation module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810558725.0A CN108829958B (en) | 2018-06-01 | 2018-06-01 | Automatic clothing plate making method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810558725.0A CN108829958B (en) | 2018-06-01 | 2018-06-01 | Automatic clothing plate making method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108829958A CN108829958A (en) | 2018-11-16 |
| CN108829958B true CN108829958B (en) | 2021-07-13 |
Family
ID=64147078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810558725.0A Active CN108829958B (en) | 2018-06-01 | 2018-06-01 | Automatic clothing plate making method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108829958B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110335104A (en) * | 2019-05-09 | 2019-10-15 | 中山易裁剪网络科技有限公司 | A kind of quick ordering system and its method of customized clothing class order |
| CN112883445A (en) * | 2021-03-11 | 2021-06-01 | 上海工程技术大学 | Intelligent clothing board making system and method |
| CN117455625A (en) * | 2023-11-21 | 2024-01-26 | 波司登羽绒服装有限公司 | Garment style customization method and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1994173A (en) * | 2006-01-05 | 2007-07-11 | 北京服装学院 | Business suit sample sheet automatic generation system and method |
| CN102657400A (en) * | 2012-04-27 | 2012-09-12 | 厦门理工学院 | Positioned combined garment sewing template and preparation method thereof |
| CN103390085A (en) * | 2013-07-24 | 2013-11-13 | 深圳市华创振新科技发展有限公司 | Automatic patterning method of basic style of clothing |
| CN104146423A (en) * | 2014-08-19 | 2014-11-19 | 北京服装学院 | Automatic shirt paper pattern generating system and method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7657341B2 (en) * | 2006-01-31 | 2010-02-02 | Dragon & Phoenix Software, Inc. | System, apparatus and method for facilitating pattern-based clothing design activities |
| US9949519B2 (en) * | 2016-04-25 | 2018-04-24 | Original, Inc. | Methods and systems for customized garment design generation |
-
2018
- 2018-06-01 CN CN201810558725.0A patent/CN108829958B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1994173A (en) * | 2006-01-05 | 2007-07-11 | 北京服装学院 | Business suit sample sheet automatic generation system and method |
| CN102657400A (en) * | 2012-04-27 | 2012-09-12 | 厦门理工学院 | Positioned combined garment sewing template and preparation method thereof |
| CN103390085A (en) * | 2013-07-24 | 2013-11-13 | 深圳市华创振新科技发展有限公司 | Automatic patterning method of basic style of clothing |
| CN104146423A (en) * | 2014-08-19 | 2014-11-19 | 北京服装学院 | Automatic shirt paper pattern generating system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108829958A (en) | 2018-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108829958B (en) | Automatic clothing plate making method and system | |
| Huang et al. | Block pattern generation: From parameterizing human bodies to fit feature-aligned and flattenable 3D garments | |
| CN101017575B (en) | Method for automatically forming 3D virtual human body based on human component template and body profile | |
| TW202022782A (en) | Method and image matching method for neural network training and device thereof | |
| CN103106586A (en) | Three-dimensional virtual fitting system | |
| CN103390085A (en) | Automatic patterning method of basic style of clothing | |
| Sabina et al. | Virtual fitting–innovative technology for customize clothing design | |
| CN115272579A (en) | Single-image three-dimensional garment reconstruction method based on multi-feature fusion | |
| CN111125415A (en) | Clothing design method and device, computer equipment and storage medium | |
| CN115906650A (en) | Standard working hour prediction method and prediction system for short-flow modular processing procedure of clothes | |
| GB2503331A (en) | Aligning garment image with image of a person, locating an object in an image and searching for an image containing an object | |
| KR102260796B1 (en) | An apparatus for providing additional pattern automatically based on reference patern and methos thereof | |
| CN104585942B (en) | The variable that clothing shoulder chest is economized is than former figure tailoring method | |
| CN109034977B (en) | Customized garment design method and system | |
| Moniruzzaman et al. | An approach to design solutions for garments using a CAD system | |
| Slavinska et al. | Devising a method to parametrize the jacket style varieties through the modification of tіpological series structures | |
| CN104585943B (en) | The former figure tailoring method of clothes equivalent transfer closing waist | |
| CN104544668B (en) | The variable that clothes pleurobranch is economized is than former figure tailoring method | |
| CN114117568A (en) | Parameterized cheongsam plate making method based on arc fitting curve | |
| Lyu et al. | Identification method of dress pattern drawing based on machine vision algorithm | |
| TWI621405B (en) | 3d sculpturing garment manufacturing method and system thereof | |
| CN113487619A (en) | Data processing method, device, equipment and medium | |
| Yan et al. | Virtual technology of Made-to-Measure Men Shirt | |
| Nizam et al. | Investigation of garments fit for 20-30 aged Bangladeshi men’s | |
| CN106256276A (en) | The BN former number design cutting-out method of door skimmed by clothing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |