Learning system of jacket paper pattern

The learning system optimizes fabric distribution in outerwear patterns using machine learning and area ratios, automating the creation process to achieve precise fits without extensive data or human intervention.

JP2024169919AActive Publication Date: 2024-12-06YPSILON JAPAN CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
JP2023086781
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-26
Publication Date
2024-12-06
Estimated Expiration
2043-05-26

AI Technical Summary

Technical Problem

Existing methods for creating outerwear patterns require significant human intervention and data collection, relying on craftsmen's intuition and experience, and struggle to ensure balanced fit and fabric balance when worn, especially for garments like suits and coats.

Method used

A learning system that uses machine learning to optimize fabric distribution based on weight and area ratios, dividing patterns into specific parts like front and back sections, and adjusting these areas to fit individual body shapes without extensive data collection or human correction.

Benefits of technology

Enables automated creation of patterns that fit the wearer's body shape accurately, reducing time and reliance on human expertise while ensuring balanced fabric distribution and fit.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024169919000001_ABST
    Figure 2024169919000001_ABST
Patent Text Reader

Abstract

To provide a learning system of a jacket paper pattern, which can automatically generate a paper pattern adjusted to a body type of a wearer without requiring collection of a large amount of data or the help of a craftsman.SOLUTION: A learning system of a jacket paper pattern includes: means for storing reference paper pattern data; object person body data acquisition means; standard paper pattern data generation means for generating standard paper pattern data based on the reference paper pattern data and the object person body data; means for storing corrected paper pattern data obtained by correcting the standard paper pattern data to fit to the body type of an object person; means for generating learning data based on the object person body data, the standard paper pattern data and the corrected paper pattern data; means for performing machine learning; and means for executing processing by a machine learning model by using a learned model. The standard paper pattern data generation means acquires the standard paper pattern data on which a machine learning result has already been reflected in cooperation with the means for executing the processing by the machine learning model.SELECTED DRAWING: Figure 5
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a learning system for patterns of outerwear such as suits and coats, and in particular to a technology for automatically creating patterns that fit the individual body shape of a wearer through machine learning using artificial intelligence. [Background technology]

[0002] Conventionally, to create a pattern based on the wearer's body type, the wearer visits a manufacturer's store, has a technician take their measurements, and then the pattern is created. That is, each manufacturer and store has its own unique pattern template, and the pattern template is graded, that is, enlarged or reduced, to fit the wearer's body size, to roughly match the size.

[0003] The shape of the human torso varies from person to person. For example, some people have a developed and thick chest, while others have a hunched back, and each person has a different body type. In order to create a pattern that fits each individual body type, a technician uses a tape measure to measure the detailed dimensions of the wearer's body, and based on the measured dimensions, corrects the original pattern held by the manufacturer or store to create a pattern that matches the wearer's body type.

[0004] This procedure for creating a paper pattern requires the wearer to be present at the store and for a technician to be present at the store. However, in this day and age of advanced network technology, a method that requires two parties to be in the same place at the same time is inefficient, and a more user-friendly service is desired.

[0005] In addition, the adjustments to the pattern based on the original pattern depend to a large extent on the craftsman's intuition and experience, and it cannot be said that there is a method based on a specific theory or methodology. Furthermore, the pattern is graded and then sewn on a trial basis, after which the craftsman makes adjustments and then sews the garment, which is a lot of work.

[0006] In response to this, technology such as that described in Patent Document 1 has been proposed for using machine learning with artificial intelligence to model clothing patterns and automatically create clothing that fits the wearer's body (Patent Document 1). [Prior art documents] [Patent documents]

[0007] [Patent Document 1] JP 2018-109998 A Summary of the Invention [Problem to be solved by the invention]

[0008] According to the technology disclosed in Patent Document 1, by generating a learning model based on the dimensions of specific parts of the wearer's actual body, it is possible to generate a paper pattern that resembles the wearer's physical characteristics to a certain extent.

[0009] However, the technology disclosed in Patent Document 1 learns only based on the individual physical characteristics of the wearer, and a huge amount of data is required to automatically create a balanced pattern that fits snugly to the body of the wearer, whose characteristics are different from one another. Therefore, after a pattern that fits to a certain extent is automatically created using a learning model and a trial stitch is performed, a craftsman must make final corrections and perform the final stitching, which is no different from the conventional method of labor and yet undesirable because it depends on the craftsman's intuition and experience.

[0010] In particular, outerwear such as suits, jackets, coats, and dress shirts do not have separate parts such as the front and back, but are connected together to form a single outerwear. Therefore, even if the dimensions of a specific part are adjusted to fit the shape of the body, the balance of the fabric is often not maintained when actually worn, making it difficult to achieve a snug fit.

[0011] In view of these problems, the present invention aims to provide an upper garment pattern learning system that can automatically create a pattern suited to the wearer's body type without the need to collect a huge amount of data or rely on a craftsman. [Means for solving the problem]

[0012] As a result of extensive research, the inventor noticed that the area of ​​each part of fabric is proportional to its weight, and discovered that by taking into account the balance of the area of ​​the fabric based on a specific point, it is possible to optimize the weight distribution of the fabric and create a pattern that fits the wearer's body shape.He then performed machine learning based on a theory that corrects the paper pattern for an outerwear garment based on the weight distribution of each part of the fabric.

[0013] The present invention provides the following solutions.

[0014] The upper garment pattern learning system according to the first feature includes an upper front body portion defined by a front neck line, which is the front line of the neck circumference, a front center line, which defines the center edge of the body, a bust line, which defines the chest circumference, an arm front edge, which defines the front edge of the armhole, and a front shoulder ridge line, with the shoulder front base point, which is the boundary between the front body portion and the back body portion, as an apex, and a back neck line, which is the back line of the neck circumference, and a back edge of the body, with the shoulder back base point, which is the boundary between the front body portion and the back body portion as an apex. a back centerline defining a back circumference, a bustline defining a chest circumference, an arm rear edge line defining a rear edge of an armhole, and a rear shoulder ridge line, wherein the upper front body portion is divided into a front center portion and a front side portion by a front shoulder perpendicular line that hangs down vertically downward from a front shoulder base point, and the upper back body portion is divided into a back center portion and a back side portion by a back shoulder perpendicular line that hangs down vertically downward from the back shoulder base point.

[0015] The upper garment pattern learning system of the first feature comprises a master data storage means for storing reference pattern data that serves as a basis for creating a pattern pattern according to an individual's body type, a subject physical data acquisition means for acquiring physical data of a subject for whom an upper garment pattern is to be created, a basic pattern data generation means for generating basic pattern data based on the reference pattern data and the subject physical data, a modified pattern data storage means for storing modified pattern data obtained by modifying the basic pattern data to fit the subject's body type, a machine learning data generation means for generating learning data required for machine learning based on information regarding the subject's physical data, information regarding the basic pattern data, and information regarding the modified pattern data, a machine learning means for learning a machine learning model using the learning data generated by the machine learning data generation means, and a machine learning model execution means for executing processing by the machine learning model using the machine learning model learned by the machine learning means, and the basic pattern data generation means cooperates with the machine learning model execution means to acquire basic pattern data reflecting the machine learning results by reflecting the machine learning model on the basic pattern data.

[0016] According to the invention relating to the first feature, an outerwear pattern is divided into a front center and front sides, and a back center and back sides, and machine learning is carried out using modified pattern data that has been modified to fit the subject's body shape. The results of the machine learning are then reflected in the basic pattern data, making it possible to carry out machine learning with a small amount of data, even for outerwear patterns that are difficult to balance in terms of fabric.

[0017] The second feature of the learning system for outerwear pattern data is the first feature of the learning system for outerwear pattern data, and further includes a modeling means for generating the reference pattern data, the basic pattern data, and the modified pattern data based on a common model.

[0018] According to the second feature of the invention, the basic paper-pattern data and modified paper-pattern data, which are the basis for generating learning data, are generated based on a common model, so that learning can be performed quickly even with a small amount of data on the general outline.

[0019] The learning system for outerwear pattern sheets relating to the third feature is a learning system for outerwear pattern sheets relating to the second feature, and the modeling means includes an area ratio calculation means for calculating an area ratio of the central chest portion and the side chest portion when the chest above the bustline is divided into the central chest portion and the side chest portion, based on the subject's physical data acquired by the subject's physical data acquisition means, with a line extending downward from the point where the frontal plane of the body and the line around the neck intersect as the boundary, and an area ratio adjustment means for adjusting a front area ratio, which is the area ratio of the front central portion and the front armpit portion, and a back area ratio, which is the area ratio of the back center portion and the back armpit portion, based on the area ratio.

[0020] According to the third aspect of the invention, the front area ratio and the back area ratio of the original paper pattern are adjusted based on the area ratio of the center and sides of the chest of the wearer, so that the fabric can be rotated around the front and back shoulder base points appropriately to fit the body shape of the wearer. Therefore, by simply obtaining the body measurements of the wearer, the correction of the paper pattern that can fit snugly to the wearer's body can be incorporated into the learning.

[0021] The fourth feature of the learning system for outerwear paper patterns is a learning system for outerwear paper patterns related to the third feature, which adjusts the areas of the front center and front armpits so that the front area ratio is approximately equal to the area ratio of the chest center and chest side areas.

[0022] According to the fourth feature of the invention, the areas of the front center and front armpits are adjusted so that the front area ratio is approximately equal to the area ratio of the center of the chest and the sides of the chest, so that correction that fully takes into account the wearer's body balance can be incorporated into the learning. Effect of the Invention

[0023] According to the present invention, it is possible to provide a learning system for upper garment pattern that can automatically create a pattern suited to the wearer's body type without the need to collect a huge amount of data or rely on a craftsman. [Brief description of the drawings]

[0024] [Figure 1] FIG. 1 is a development showing the left half of the body part of a paper pattern for applying this embodiment. [Diagram 2] FIG. 2 is a schematic diagram showing the shape of the wearer's upper left body created based on scan data. [Diagram 3] FIG. 3 is a schematic diagram showing a case where the area adjustment step of the method for correcting the outerwear paper pattern in the common model is applied to the anti-body type. [Figure 4] FIG. 4 is a schematic diagram showing a change in weight distribution when the method for correcting the outerwear paper pattern in the common model is applied to the anti-body type. [Diagram 5] FIG. 5 is a block diagram for explaining the hardware configuration and software functions of the upper garment paper pattern learning system according to the present embodiment. [Figure 6] FIG. 6 is a flow chart showing a pattern learning method using the upper garment pattern learning system according to this embodiment. [Figure 7] FIG. 7 is a flow chart showing a method for creating a pattern using the upper garment pattern learning system according to this embodiment. [Figure 8] FIG. 8 is a schematic diagram of master data used in the learning system for upper garment paper patterns according to this embodiment. [Figure 9] 9A and 9B are schematic diagrams of subject body data used in the upper garment paper pattern learning system according to the present embodiment, where (a) shows point cloud data and (b) shows mesh data. [Figure 10] FIG. 10 is a schematic diagram of three-dimensional basic pattern data obtained by using the upper garment pattern learning system according to the present embodiment. [Figure 11] FIG. 11 is a schematic diagram of flat basic pattern data (before the machine learning results are reflected) obtained by using the upper garment pattern learning system according to the present embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0025] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Note that this is merely an example, and the technical scope of the present invention is not limited to this example.

[0026] In the present invention, the center side refers to the side of the midline of the body when viewed from the front or back, and the armpit side refers to the side of the body. Furthermore, in the present invention, outerwear refers to the entire clothing worn on the upper body, including suits, jackets, coats, shirts, etc., regardless of whether it is for men or women.

[0027] [Common Model] The common model in the present invention is a common model for forming master data, basic paper pattern data, and modified paper pattern data, which will be described later in the present invention, and is created based on balance theory.

[0028] The balance theory in the present invention is a theory that, when creating an upper garment pattern, it is possible to optimize the weight distribution of the fabric by considering the proportional relationship between the area and weight of each part of the fabric and the balance of the fabric area with a specific point as the base point, and to create a pattern that fits the wearer's body type. Based on the idea that the distribution of weight among the parts that make up the garment is important for the garment to fit snugly around the torso, the balance theory focuses on the rotational movement of the fabric around the front and back shoulder base points that support the weight of the entire garment, and on the fact that the rotational movement creates an appropriate fit.

[0029] The common model created based on the balance theory of the present invention is created by dividing the chest above the bustline into a central chest area and a lateral chest area, with a line extending downward from the point where the frontal plane of the body and the neck line intersect as the boundary, calculating the area ratio of the central chest area to the lateral chest area, and adjusting the front area ratio, which is the area ratio of the front central area to the front armpit area, and the back area ratio, which is the area ratio of the back central area to the back armpit area, based on the calculated area ratio.

[0030] [Pattern prototype] The original pattern for the common model will be explained using Figure 1. Figure 1 shows a schematic diagram of the original pattern for the left half of the body. The original pattern for the right half of the body is left-right reversed, so its explanation is omitted. Note that the original pattern is the source of patterns unique to each brand or manufacturer for suits, jackets, coats, etc., and corrections are made based on the original pattern according to the wearer's body shape, which will be described later.

[0031] The paper pattern X in the common model has an upper front body part FB and an upper back body part BB.

[0032] <Upper front body part> In the upper front body part FB, the front neck line NL1, which is the front line of the neck circumference, and the front shoulder ridge HL1 intersect at the position where the front shoulder base point N intersects. The front shoulder base point N is the base point for balancing the weight of the entire jacket when it is worn, and is the boundary between the front body part and the back body part. The front shoulder base point N is located near the depression of the sternocleidomastoid muscle of the human body.

[0033] An armhole AH is formed in the upper side of the upper front body part FB and the upper back body part BB. The armhole AH is defined by an arm leading edge line AL1 in the upper front body part FB and an arm trailing edge line AL2 in the upper back body part BB.

[0034] The bust line BL is a line that defines the chest circumference CC, and is formed horizontally through the upper front body part FB and the upper back body part BB. The front center end of the bust line BL at the upper front body part FB is point F, and the side end is point G.

[0035] Point H on the bust line BL is a point that is a specified distance away from point F, which is the front end point of the bust line BL, and in the common model shown in Figure 1, it is a point that is a distance of 0.37 times the measurement of adding 6 cm to the chest circumference CC (i.e., (CC + 6 cm) x 0.37) toward the armpit side. Point E is the point where the front center line CL that extends vertically upward from point F, which is the front end point of the bust line BL, intersects with the front neck line NL1.

[0036] Further, point P on the bust line BL is the point where the shoulder front perpendicular line PL, which is suspended vertically downward from the shoulder front base point N in the upper front body part FB, intersects with the bust line BL.

[0037] Point I is the rear end point of the front shoulder ridge line HL1, and is the point where the front shoulder ridge line HL1 intersects with the arm front edge line AL1 that defines the armhole AH.

[0038] In this way, the fabric above the bustline BL in the upper front body portion FB is defined by the front neck line N1, which is the front line of the neck, the front center line CL, which defines the central edge of the body, the bust line BL, which defines the chest circumference, the arm front edge line AL1, which defines the front edge of the armhole, and the front shoulder ridge line HL1, with the shoulder front base point N as the vertex., that is, by point E-line CL-point F-point P-point H-point G-line AL1-point I-line HL1-point N-line NL1-point E.

[0039] In this case, the front center portion FB1 of the upper front body portion FB is defined by point N-line NL1-point E-line CL-point F-point P-line PL-point N, and the front side portion FB2 is defined by point N-line PL-point P-point H-point G-line AL1-point I-line HL1-point N.

[0040] <Upper back body part> In the upper back body part BB, the back neck line NL2, which is the back line of the neck circumference, and the back shoulder ridge HL2 intersect at the position where the back shoulder base point M is formed. The back shoulder base point M is the base point for balancing the weight of the entire garment when it is worn, and is the boundary between the front body part and the back body part.

[0041] In the upper back body part BB, a bust line BL is formed on an extension of the bust line BL of the front body part FB, and a point K is a point corresponding to the side end of the bust line BL.

[0042] Furthermore, point Q on the bust line BL is the point where the shoulder back perpendicular line QL, which is suspended vertically downward from the shoulder back base point M at the upper back body part BB, intersects with the bust line BL.

[0043] The waistline WL is a line that defines the abdominal circumference WC, and the point at the center end of the back side of the waistline WL at the upper back body part BB, that is, the point that is the boundary between the left and right sides, is designated as point S.

[0044] A line extending vertically upward from point S is defined as a provisional back center line SL1, and a point where the provisional back center line SL1 and the posterior neck line NL2 intersect is defined as a provisional back base point O1.

[0045] Here, since the actual back of the human body is not straight but curved, the original back base point determined in consideration of the allowance for this is called the original back base point O2. The original back base point O2 is formed at a position where the posterior neck line NL2 is further extended a certain distance from the provisional back base point O1, and this distance is defined as L1 x 0.1, where L1 is the length of the perpendicular line drawn from the provisional back base point O1 to the posterior shoulder perpendicular line QL. In addition, the line that connects the original back base point O2 and point S, which is the posterior center end of the waistline WL, in a gently curved manner is called the original back center line SL2.

[0046] The point where the original back center line SL2 and the bust line BL intersect is defined as point L. Point L is the point at the center of the back side of the bust line BL.

[0047] In this way, the fabric above the bustline BL in the upper back part BB is defined by the back neck line NL2, which is the back line of the neck, the back center line SL2, which defines the dorsal edge of the body, the bustline BL, which defines the chest circumference, the back arm edge line AL2, which defines the back edge of the armhole, and the back shoulder ridge line HL2, that is, by point O2-line SL2-point L-point Q-point K-line AL2-point J-line HL2-point M-line NL2-point O1-point O2.

[0048] In this case, the back center BB1 of the upper back body part BB is defined by point M-line NL2-point O1-point O2-line SL2-point L-point Q-line QL-point M, and the back armpit part BB2 is defined by point M-line QL-point Q-point K-line AL2-point J-line HL2-point M.

[0049] In this manner, a paper prototype of the common model is formed.

[0050] The area of ​​each part in the original pattern is predetermined. For example, in this embodiment, the front center area A of the front center part FB1 is 21,224.5 mm2, the front armpit area B of the front armpit part FB2 is 22,869.0 mm2, the back center area C of the back center part BB1 is 23,457.0 mm2, and the back armpit area D of the back armpit part BB2 is 33,781.9 mm2.

[0051] [How to correct the pattern in the common model] Next, a method for correcting an upper garment paper pattern in a common model according to this embodiment will be described with reference to FIGS.

[0052] Figure 2 is a schematic diagram showing the shape of the wearer's upper left body created based on scan data, Figure 3 is a schematic diagram showing the area adjustment step of the outerwear paper pattern correction method in a common model when applied to an anti-body type, and Figure 4 is a schematic diagram showing the change in weight distribution when the outerwear paper pattern correction method in a common model is applied to an anti-body type.

[0053] <Step S110: Preparation of paper pattern> First, a paper pattern master as shown in FIG. 1 is prepared (step S110).

[0054] The pattern template prepared in step S110 has a front neck line NL1, which is the front line of the neckline, a front center line CL, which defines the center edge of the body, a bust line BL, which defines the chest circumference, an arm front edge line AL1, which defines the front edge of the armhole, and a front shoulder ridge line HL1, which defines the upper front body part FB, and a back neck line NL1, which is the back line of the neckline, a front center line CL, which defines the center edge of the body, a bust line BL, which defines the chest circumference, and a front shoulder ridge line HL1, which defines the back edge of the body, and a back shoulder base point M, which is the border between the front body part and the back body part. The upper back body part BB is defined by a back centerline SL2 which defines the chest circumference, a bustline BL which defines the chest circumference, an arm rear edge line AL2 which defines the rear edge of the armhole, and a rear shoulder ridge line HL2, and the upper front body part FB is divided into a front center part FB1 and a front armpit part FB2 by a shoulder front perpendicular line PL which hangs vertically downward from the shoulder front base point N, and the upper back body part BB is divided into a back center part BB1 and a back armpit part BB2 by a shoulder back perpendicular line QL which hangs vertically downward from the shoulder rear base point M.

[0055] The original paper pattern is not limited to one formed to a predetermined size, and it is also possible to use one that has been adjusted according to the size of each part of the wearer's body measured in step S120 described below (in which case, the order of steps S110 and S120 is reversed).

[0056] <Step S120: Measuring the dimensions of each body part> Next, the dimensions of each part of the wearer's body are measured (Step S120).

[0057] In step S120, a 3D body scanner capable of measuring the human body in three dimensions is used to obtain height, bust height, waist height, hip height, inseam height, neck circumference, chest circumference, underbust circumference, and waist circumference. Based on each of the obtained data, the neck circumference, chest circumference, underbust circumference, and waist circumference are divided horizontally, and vertically into 12 parts by the center line of the body passing through the navel and the shoulder line. A schematic diagram showing the shape of the upper left half of the body obtained in this manner is shown in Figure 2.

[0058] Here, point Y is defined where the frontal plane, which divides the body into front and back, intersects with the neck line of the body. This point Y corresponds to the position of the depression beside the sternocleidomastoid muscle, and is also the point corresponding to the front shoulder base point N and the rear shoulder base point M in the original pattern mentioned above. Then, line XL extending downward from point Y is used as the boundary to divide the chest above the bust line into a central chest area CH1 and a lateral chest area CH2.

[0059] <Step S130: Calculation of Area> Based on the dimensions of each part measured in step S120, the area of ​​the wearer's body corresponding to each part of the pattern is calculated (step S130). In particular, the area of ​​the chest center CH1 and the area of ​​the chest side CH2 are calculated. This allows the area ratio of the wearer's chest center area to the chest side area to be calculated. The area ratio may be calculated separately for the left and right sides. This is because the human body is not necessarily symmetrical, and different corrections may be made for the right and left parts.

[0060] <Step S140: Adjustment of the area of ​​the paper pattern> Based on the area ratio of the wearer's central chest area and side chest area calculated in step S130, a front area ratio, which is the area ratio of the front central area FB1 to the front armpit area FB2 in the pattern template X, and a back area ratio, which is the area ratio of the back central area BB1 to the back armpit area BB2 in the pattern template X, are adjusted (step S140).

[0061] The method of adjusting the area ratio in step S140 will be described in detail with reference to FIG.

[0062] 3 is a schematic diagram of the upper front body part FB and the upper back body part BB when the area adjustment step of the method for correcting the outerwear paper pattern in the common model is applied to an anti-body type. The anti-body type is a body type in which the upper body is bent backward.

[0063] As shown in Fig. 3, the size of the chest at the upper front body part FB can be changed by making a cut from point F, which is the front center end of the bust line BL that defines the lower end of the upper front body part FB, along the bust line BL to point H, and rotating the upper front body part FB around point H. Also, the size of the back at the upper back body part BB can be changed by making a cut from point L, which is the back center end of the bust line BL at the upper back body part BB, along the bust line BL to point K, which is the armpit end, and rotating the upper back body part BB around point K. In Fig. 3, the original shape before correction is shown by a dashed line, and the shape after rotation, which is the corrected state, is shown by a solid line.

[0064] For example, in the case of a body type that is bent backwards, such as the anti-body type shown in Figure 3, if the fabric is made by enlarging or reducing the original pattern, there will not be enough fabric for the chest area, and the chest will look tight. Therefore, as shown in Figure 3, the upper front body part FB is rotated so that the bust line BL opens upward around point H. Conversely, the upper back body part BB is rotated so that the bust line closes downward around point K.

[0065] <Change in area ratio of upper front body part FB> With the rotation around point H, the shoulder front base point N moves to shoulder front base point N', and point I moves to point I'. Also, the bust line BL itself remains unchanged, and since the shoulder front perpendicular line is a perpendicular line extending vertically downward from the shoulder front base point N', the shoulder front perpendicular line PL moves to the shoulder front perpendicular line PL'. Furthermore, point P moves to point P' where the shoulder front perpendicular line PL' and the bust line BL intersect.

[0066] Furthermore, since the front centerline CL, which is a perpendicular line that defines the front edge of the upper front body portion FB, remains unchanged, point F moves vertically upward to point F', and point E moves vertically upward to point E' on an extension of the front centerline CL, as the paper pattern is rotated. As a result, the front neckline NL1 moves to the front neckline NL1', the front shoulder ridgeline HL1 moves to the front shoulder ridgeline HL1', and the arm front edge portion AL1 moves to the arm front edge portion AL1'.

[0067] Since the front center line CL and the bust line BL remain unchanged, the front center portion FB1' after rotation correction is defined as point N'-line NL1'-point E'-line CL-point F-point P'-line PL'-point N'.

[0068] Similarly, the front armpit FB2' after the rotation correction is defined by point N'-line PL'-point P'-point H-point G-line AL1'-point I'-line HL1'-point N'.

[0069] In other words, a new front center part FB1' after rotation correction is obtained by cutting the bust line BL to a point H that is a predetermined distance away from the front end point F and rotating the upper front body part FB around point H that is a predetermined distance away, and is obtained by the front center line CL before rotation, the bust line BL before rotation, the shoulder front perpendicular line PL' after rotation, and the front neck line NL1' after rotation.

[0070] Further, the new front armpit FB2' after the rotation correction is obtained by the bust line BL before rotation, the shoulder front perpendicular line PL' after rotation, the front shoulder ridge line NL1' after rotation, and the arm front edge line AL1' after rotation.

[0071] By rotating and correcting the bust line BL of the upper front body part FB to open upward in this way, the area A of the front center part FB1 before correction increases, and the area B of the front armpit part FB2 before correction decreases. The area A of the front center part FB1 increases because the shoulder front perpendicular line PL' moves to the armpit side even though the front center line CL remains unchanged. This is also due to the increase in the approximately triangular part connecting points F'-F-P' by the opening.

[0072] In addition, the area B of the front armpit FB2 is reduced by rotating the bust line BL upward to open it up, because the arm front edge AL1 is close to the point H, which is the center of rotation, so the amount of movement of the line is small and the corresponding increase in area is small, and because the shoulder front perpendicular line PL' has moved toward the armpit.

[0073] In other words, by rotating and correcting the bust line BL so that it opens upward, the area of ​​the front center portion FB1 in the upper front body portion FB can be enlarged and the area of ​​the front armpit portion FB2 can be reduced, thereby changing the front area ratio, which is the area ratio of the front center portion area A:front armpit area B.

[0074] When adjusting the balance of the pattern using the pattern correction method based on the balance theory of the common model in this embodiment, the amount of movement from point F to point F' is determined so that the front area ratio of the new front center area A' at the front center area FB1' after rotation correction to the new front armpit area FB2' at the front armpit area FB2' is approximately equal to the ratio of the central chest area to the side chest area calculated in step S3 above based on the wearer's measurement results.

[0075] For example, if the area A of the front center FB1 in the original pattern is 21,224.5 mm2 and the area B of the front armpit FB2 is 22,869.0 mm2, but the ratio of the area of ​​the chest center CH1 to the area of ​​the chest side CH2 calculated based on the wearer's measurements is CH1:CH2 = 1.145:1, point F is moved 1 cm upward. As a result, the area A' of the new front center FB1' is 24,474.1 mm2, and the area B' of the new front armpit FB2' is 21,376.2 mm2. As a result, the ratio of the area A' of the new front center FB1' to the area B' of the new front armpit FB2' is A':B' = 1.145:1, which matches the area ratio calculated based on the measurements.

[0076] <Change in area ratio of upper back body BB> On the other hand, the upper back body part BB is rotated in the opposite direction to the rotation of the upper front body part FB, so as to close the bust line BL downward about the point K.

[0077] At this time, in Figure 3, with the rotation around point K, the back shoulder base point M moves to back shoulder base point M', point O2 moves to point O2', and point J moves to point J'. Also, since the bust line BL remains unchanged and the back shoulder perpendicular line QL is a perpendicular line extending vertically downward from the back shoulder base point M', the back shoulder perpendicular line QL moves to the back shoulder perpendicular line QL'. Furthermore, point Q moves to point Q' where the back shoulder perpendicular line QL' and the bust line BL intersect.

[0078] As a result, the arm rear edge AL2 moves to the arm front edge AL2', the rear shoulder ridge HL2 moves to the rear shoulder ridge HL2', and the rear neck line NL2 moves to the rear neck line NL2'. Also, although point O2 moves to point O2', point S at the center end of the waistline WL remains unchanged, so the real back centerline SL2 moves to the real back centerline SL2'.

[0079] And, since the bust line BL remains unchanged, the point where the bust line BL intersects with the actual back center line SL2' moves horizontally from point L to point L'.

[0080] Since the bust line BL remains unchanged, the new back center point BB1' after rotation correction is defined by point M'-line NL2'-point O2'-line SL2'-point L'-point Q'-line QL'-point M'.

[0081] Similarly, the new back side BB2' after the rotation correction is defined by point M'-line QL'-point Q'-point K-line AL2'-point J'-line HL2'-point M'.

[0082] In other words, a new back center line BB1' obtained by cutting into the bust line BL and rotating the upper back body part BB around point K at the back center end of the bust line BL is obtained by the bust line BL before rotation, the back shoulder perpendicular line QL' after rotation, the back neck line NL2' after rotation, and the main back center line SL2' after rotation.

[0083] In addition, the new back armpit BB2' after rotation is obtained by the bust line BL before rotation, the arm rear edge line AL2' after rotation, the rear shoulder ridge line HL2' after rotation, and the rear shoulder perpendicular line QL' after rotation.

[0084] In this way, by rotating and correcting upper back body part BB so that the part above bustline BL in upper back body part BB is closed downward, the back center area C of back center part BB1 before correction is reduced, and the back side area D of back side part BB2 before correction is enlarged. The reason why the back center area C of back center part BB1 is reduced is because the movement of back back perpendicular line QL' toward the center is greater than the movement of back center line SL2'. This is because back back perpendicular line QL' is a vertical line, while back center line SL2' is a gentle curve extending to point S. This is also due to the reduction in the area of ​​the approximately triangular part that is closed.

[0085] In addition, by performing a rotational correction so as to close the portion above the bust line BL downward, the position of the rear shoulder ridge line HL2 moves downward, but the rear armpit area D of the rear armpit area BB2 increases because the amount of movement of the rear shoulder perpendicular line QL' toward the center is greater than the amount of movement of the rear arm edge line AL2.

[0086] Here, the amount of correction for the back body part BB is the same as the amount of correction for the front body part FB. In other words, if the amount of upward movement from point F to point F' is 1 cm, the amount of downward movement from point L to point L' is also 1 cm.

[0087] For example, in the original pattern, the back center area C of the back center part BB1 was 23,457.0 mm2, and the back underarm area D of the back underarm part BB2 was 33,781.9 mm2, but point F on the front body part FB was moved 1 cm upward, and point L2 on the back body part BB was moved 1 cm downward. As a result, the back center area C' of the new back center part BB1' was 21,528.4 mm2, and the back underarm area D' of the new back underarm part BB2' was 36,375.1 mm2.

[0088] In this way, based on the balance theory that forms a common model, by performing rotational correction around the bust line BL, it is possible to adjust the front area ratio, which is the area ratio between the front center portion FB1 and the front armpit portion FB2, and the back area ratio, which is the area ratio between the back center portion BB1 and the back armpit portion BB2.

[0089] Here, the change in fabric movement when rotation correction is performed will be described for the case of an anti-body type with reference to Figure 4. As mentioned above, an anti-body type is a body type with a curved chest, and if a pattern is created with the area ratio of the original paper pattern, the chest (i.e., the front center FB1 of the upper front body part FB) will be stretched and pulled up, and the fabric on the back will be pulled down too far. When there are a mixture of cramped and loose parts like this, it is difficult to wear a garment that fits the body type.

[0090] Therefore, as shown in FIG. 4, in the upper front body part FB, the area above the bust line BL is rotated upwardly around point H, and in the upper back body part BB, the area above the bust line BL is rotated downwardly around point K, and closed.

[0091] 3, in the upper front body part FB, the front center area A' of the front center part FB1' becomes larger and the front armpit area B' of the front armpit part FB2' becomes smaller. Conversely, in the upper back body part BB, the back center area C' of the back center part BB1' becomes smaller and the back armpit area D' of the back armpit part BB2' becomes larger.

[0092] As a result, in the upper front body portion FB, the weight of the front center portion FB1' becomes heavier and the weight of the front armpit portion FB2' becomes lighter, resulting in a rotational motion in which the front center portion FB1' drops downward and the front armpit portion FB2' rises upward around the shoulder front base point N' (arrow AR1 in the figure).

[0093] On the other hand, in the upper back body part BB, the weight of the back center part BB1' becomes lighter and the weight of the back side part BB2' becomes heavier, but because the right and left halves of the body are sewn together at the back center part BB1', it is considered as the combined weight of the right and left halves of the body and not the weight of one side. Therefore, even if the weight of the back center part BB1' becomes lighter, the direction of the rotational movement in which the back center part BB1' moves downward and the back side part BB2' moves upward remains the same as before the correction; it is just that the rotational force is weakened (arrow AR2 in the figure).

[0094] In this way, in the case of the opposite body type, by opening the upper front body part FB above the bust line BL and increasing the weight of the front center part FB1, the front center part FB'1, which tends to be tight, can be lowered downward, and by closing the upper back body part BB above the bust line BL and decreasing the weight of the back center part BB1, the back center part BB1', which tends to be loose, can be fitted so that it does not drop downward. Also, the area R of the part enclosed between the temporary back center line SL1, the real back center line SL2, and the bust line BL increases.

[0095] In this way, in the common model based on balance theory, the front center FB1' and the front side FB2' rotate around the shoulder front base point N' like a seesaw to try to balance, forming a pattern that fits the wearer's body. Similarly, the back center BB1' and the back side BB2' rotate around the shoulder rear base points M, M' like a seesaw to try to balance, forming a pattern that fits the wearer's body.

[0096] In this way, common models using balance theory are characterized by creating patterns that fit the wearer's body by balancing the rotational movement that accompanies changes in the weight of the fabric.

[0097] [Tops Pattern Learning System 1] Next, the upper garment pattern learning system 1 according to this embodiment will be described with reference to FIG.

[0098] FIG. 5 is a block diagram for explaining the hardware configuration and software functions of the upper garment paper pattern learning system 1 according to this embodiment.

[0099] As shown in Figure 5, the upper garment paper pattern learning system 1 is composed of a control means 10 that controls data, a communication means 20 that communicates between each means, a memory means 30 that stores data, and a subject's physical data acquisition means 40 that acquires the subject's physical data.

[0100] The control means 10 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), and the like.

[0101] The control means 10 reads a predetermined program and cooperates with the communication means 20 as necessary to realize a master data creation means 11, a basic pattern data generation means 12, a modified pattern data generation means 13, a machine learning data generation means 14, a machine learning means 15, a machine learning model execution means 16, and a modeling means 17.

[0102] The communication means 20 includes a device for enabling communication with other devices, for example, a Wi-Fi (Wireless Fidelity) compatible device conforming to IEEE802.11.

[0103] The storage means 30 is a device for storing data and files, and includes a data storage unit such as a hard disk, a semiconductor memory, a recording medium, a memory card, etc. The storage means 30 also includes a master data storage means 31, a subject body data storage means 32, a basic paper pattern data storage means 33, a corrected paper pattern data storage means 34, a machine learning data storage means 35, a machine learning completed model storage means 36, and ancillary data storage means 37.

[0104] The subject's body data acquisition means 40 acquires the subject's physical characteristic data for creating an upper garment paper pattern. The subject's body data acquisition means 40 is, for example, configured with a 3D body scanner capable of measuring the subject's detailed body dimensions in a non-contact manner. The subject's physical characteristic data acquired by the subject's body data acquisition means 40 is stored in the subject's body data storage means 32 via the communication means 20.

[0105] [Master data creation method 11] The master data creating means 11 creates reference paper pattern data that serves as a reference when creating a paper pattern according to an individual body type. The reference paper pattern data is created by correcting the paper pattern prototype data to match the reference body type data, which is the reference body type data. In other words, the reference paper pattern data is created by matching the paper pattern prototype X shown in FIG. 1 to the reference body type and performing the correction as shown in FIG. 3. Here, the reference body type data, which is the reference body type data, is acquired by the subject body data acquiring means 40. In addition, the reference paper pattern data created at this time uses a common model based on the balance theory described above. The reference paper pattern data consists of planar reference paper pattern data, which is two-dimensional paper pattern data, and three-dimensional reference paper pattern data obtained by converting the planar reference paper pattern data into three-dimensional data.

[0106] [Master data storage means 31] The master data storage means 31 stores master data consisting of reference body type data and reference paper pattern data created by the master data creation means 11 based on the reference body type data.

[0107] [Basic pattern data generating means 12] The basic pattern data generating means 12 automatically generates basic pattern data prior to the final modified pattern data, and is composed of a two-dimensional basic pattern data generating section 12a which generates two-dimensional basic pattern data, and a three-dimensional basic pattern data generating section 12b which creates three-dimensional basic pattern data.

[0108] In this embodiment, the three-dimensional basic pattern data is created in the three-dimensional basic pattern data generating unit 12b by grading the three-dimensional reference pattern data stored in the master data storage means 31 within an approximate range based on the difference between the reference body type data stored in the master data storage means 31 and the subject's body data obtained by the subject's body data acquisition means 40.

[0109] The flat basic pattern data generating unit 12a creates flat pattern data by converting the three-dimensional basic pattern data created by the three-dimensional basic pattern data generating unit 12b into two-dimensional pattern data. In this embodiment, the results of experience and experiments are reflected in the data, and adjustments are made to the places where straight lines are to be kept, the places where darts are to be opened, and the degree to which they are to be opened.

[0110] Furthermore, the basic pattern data generating means 12 writes out the difference between the converted two-dimensional basic pattern data and the two-dimensional reference pattern data. The written out difference is reflected in the results of machine learning executed by the machine-learned model executing means 16 described later, and the result is transformed from the reference pattern data stored in the master data storage means 31 by the amount of the difference to obtain flat basic pattern data reflected in the machine learning results. In other words, the basic pattern data generating means 12 cooperates with the machine-learned model executing means 16 to obtain basic pattern data reflected in the machine learning results, which is obtained by reflecting the machine learning model on the basic pattern data.

[0111] The two-dimensional paper pattern data and the three-dimensional paper pattern data thus generated are used as a set to create basic paper pattern data. The basic paper pattern data generated by the basic paper pattern data generating means 12 is stored in the basic paper pattern data storage means 33 via the communication means 20.

[0112] [Modified pattern data generating means 13] The modified pattern data generating means 13 modifies the basic pattern data generated by the basic pattern data generating means 12 so that it fits the subject's body type. The basic pattern data is modified using a common model based on the balance theory described above. In this embodiment, the modified pattern data generating means 13 is incorporated into the learning system, but the basic pattern data may be modified manually. However, even when modified manually, the modification is still based on the balance theory.

[0113] [Modified paper pattern data storage means 34] The modified pattern data storage means 34 stores modified pattern data obtained by modifying the basic pattern data generated by the basic pattern data generating means 12 so as to fit the subject's body type. The basic pattern data is modified using a common model based on the balance theory described above.

[0114] [Methods for generating data for machine learning 14] The machine learning data generating means 14 generates learning data necessary for machine learning of the AI ​​model based on information on the subject's body data, information on the basic paper pattern data, and information on the corrected paper pattern data. In particular, here, the learning data is generated by linking the correction amount, which is the difference between the corrected paper pattern data and the basic paper pattern data, to the subject's body data.

[0115] In addition, the learning data may be generated based on the area of ​​each part of the subject's body divided as shown in Figure 2, bust size, waist size, neck size, etc., acquired by the subject's physical data acquisition means 40.

[0116] [Machine learning data storage means 35] The machine learning data storage means 35 stores the learning data generated by the machine learning data generation means 14 in association with the subject's physical data, basic pattern data, and modified pattern data that were the source of the learning data.

[0117] [Machine learning method 15] The machine learning means 15 causes the machine learning model to perform machine learning using the learning data generated by the machine learning data generation means 14. A large amount of machine learning data generated by the machine learning data generation means 14 is used for machine learning. The machine learning means 15 performs a convolution operation on the input machine learning data using, for example, a convolutional neural network, which is one type of machine learning, and extracts its feature amount. Note that when performing machine learning, means other than the convolutional neural network may also be used.

[0118] [Means for storing machine-learned models 36] The machine-learned model storage means 36 stores the machine-learned model that has been machine-learned by the machine-learning means 15.

[0119] [Methods for running machine-learned models 16] The machine-learned model executing means 16 executes processing by the machine-learned model using the machine-learned model learned by the machine-learning means 15. The processing by the machine-learned model executing means 16 is reflected in the generation of planar basic paper pattern data in the basic paper pattern data generating means 12. That is, the basic paper pattern data generated by the basic paper pattern data generating means 12 reflects the processing by the machine-learned model executed by the machine-learned model executing means 16 in addition to the reference body type data, reference paper pattern data, and subject body data.

[0120] [Modeling Method 17] The modeling means 17 generates data based on a common model for the master data, the basic paper pattern data, and the modified paper pattern data. The modeling means has an area ratio calculation section 17a and an area ratio adjustment section 17b.

[0121] Based on the subject's physical data acquired by the subject's physical data acquisition means 40, the area ratio calculation unit 17a calculates the area ratio of the central chest region CH1 to the lateral chest region CH2 when the chest above the bustline is divided into the central chest region CH1 and the lateral chest region CH2, with the boundary being a line XL extending downward from point Y where the frontal plane of the body and the neck line shown in Figure 2 intersect.

[0122] The area ratio adjustment unit 17b adjusts the front area ratio, which is the area ratio between the front center portion FB1 and the front armpit portion FB2 shown in Figure 1, and the back area ratio, which is the area ratio between the back center portion BB1 and the back armpit portion BB2, based on the area ratio between the subject's chest center portion CH1 and the chest lateral portion CH2 calculated by the area ratio calculation unit 17a.

[0123] [Attachment data storage means 37] The accessory data storage means 37 stores accessory data relating to accessory parts of a jacket such as a collar and pockets.

[0124] [Learning method using the upper garment paper pattern learning system 1] Next, a learning method using the upper garment paper pattern learning system 1 configured as above will be described with reference to FIG.

[0125] 6 is a flowchart showing how to learn an outerwear pattern using the outerwear pattern learning system 1 according to this embodiment. The learning is performed based on data registered in each data storage unit when the outerwear pattern is created, as described below.

[0126] <Step S200: Reading out subject's physical data> First, the subject's physical data stored in the subject's physical data storage means 32 is read out (step S200).

[0127] <Step S210: Reading out flat basic paper pattern data> Next, flat basic paper pattern data is read out from the basic paper pattern data stored in the basic paper pattern data storage means 33 (step S210).

[0128] <Step S220: Reading out flat corrected paper pattern data> Furthermore, the plane modified paper pattern data is read out from the modified paper pattern data stored in the modified paper pattern data storage means 34 (step S220).

[0129] <Step S230: Generation of machine learning data> Then, the machine learning data generating means 14 generates machine learning data based on the subject's body data read in step S200, the flat basic paper pattern data read in step S210, and the flat corrected paper pattern data read in step S220 (step S230). In particular, here, the machine learning data is generated by linking the correction amount, which is the difference between the flat corrected paper pattern data and the flat basic paper pattern data, to the subject's body data.

[0130] <Step S240: Executing machine learning> The machine learning means 15 performs machine learning using the data for machine learning generated in step S230 (step S240). The result of the machine learning is reflected in the machine learning model stored in the machine-learned model storage means 36.

[0131] The above-described machine learning is performed each time an upper garment pattern is created, and the results of the machine learning are gradually accumulated and used to generate flat basic pattern data, which will be described later.

[0132] [Creating outerwear pattern data using outerwear pattern learning system 1] Next, a method for creating a pattern using the upper garment pattern learning system 1 configured as described above will be described with reference to Figures 7 to 11. At this time, it is assumed that the reference pattern data, which is the basis for creating a pattern according to an individual body type, and the reference body type data, which is the reference body type data, have already been registered in the master data storage means 31 as master data.

[0133] Figure 7 is a flowchart for creating an outerwear pattern using the outerwear pattern learning system 1 of this embodiment, Figure 8 is a schematic diagram showing an example of master data, Figure 9 is a schematic diagram showing an example of subject body data, Figure 10 is a schematic diagram showing an example of three-dimensional basic pattern data, and Figure 11 is a schematic diagram showing an example of two-dimensional basic pattern data (before the machine learning results are reflected).

[0134] <Step S300: Reading Master Data> First, reference paper pattern data, which is a reference for creating a paper pattern according to an individual body type and reference body type data, which are created in advance by the master data creating means 11 and stored in the master data storage means 31, are read out as master data (step S300). The reference paper pattern data, which is the master data, is created based on balance theory and is, for example, as shown in Fig. 8. The reference paper pattern data to be read out includes not only planar reference paper pattern data, which is two-dimensional paper pattern data, but also three-dimensional reference paper pattern data, which is obtained by converting the planar reference paper pattern data into three-dimensional data.

[0135] <Step S310: Acquisition and registration of subject's physical data> Next, the subject's body data acquiring means 40 is used to measure the dimensions of each part of the subject's body for which an upper garment pattern is to be created, and the subject's body data is acquired (step S310).

[0136] In step S310, the subject's body shape is measured using a 3D body scanner capable of measuring the human body in three dimensions, which is an example of the subject's body data acquisition means 40. The data obtained here is point cloud data in which each point on the surface of the body is projected onto a three-dimensional coordinate system, and the point cloud data thus obtained is, for example, as shown in FIG.

[0137] Next, the point cloud data obtained by three-dimensional measurement using a 3D body scanner is converted into mesh-shaped body mesh data. When converting the point cloud data into body mesh data, three adjacent points are connected to form a triangular mesh, and the subject's body shape is represented by a collection of small triangular meshes. The subject's body mesh data obtained in this way is shown in Figure 9(b).

[0138] From the obtained body mesh data, the subject's height, bust height, waist height, crotch height, inseam height, neck circumference, chest circumference, underbust circumference, and waist circumference are obtained. Based on each obtained data, as shown in FIG. 2, the neck circumference, chest circumference, underbust circumference, and waist circumference are divided horizontally, and vertically divided into a total of 12 parts by the center line of the body passing through the navel and the shoulder line. In this way, the subject's body data in the present invention, which is composed of the body mesh data and the dimensional data of each part, is obtained. The obtained subject's body data is stored in the subject's body data storage means 32 via the communication means 20. Note that the conversion from the point cloud data to the body mesh data and the acquisition of each dimensional data from the body mesh data may be performed within the subject's body data acquisition means 40, or may be performed using a calculation means separately provided within the control means 10.

[0139] <Step S320: Generation of 3D basic pattern data> Based on the subject's body data obtained in step S310, three-dimensional basic pattern data is generated (step S320). The three-dimensional basic pattern data is generated using the three-dimensional basic pattern generating unit 12b in the basic pattern data generating means 12, and is generated by grading the three-dimensional three-dimensional reference pattern read from the master data storage means 31 in step S300 in an approximate range based on the difference between the reference body type data read in step S300 and the subject's body data acquired in step S310. The three-dimensional pattern represented by the three-dimensional basic pattern data generated in step S320 is as shown in FIG. 10. The three-dimensional basic pattern data generated here is obtained by roughly grading the reference pattern data based on the difference between the reference body type data registered as master data and the subject's body type data, so it is not highly accurate and is created with some margin for actual wearing.

[0140] <Step S330: Generation of flat basic paper pattern data> Based on the three-dimensional basic pattern data generated in step S320, two-dimensional basic pattern data is generated (step S330). The two-dimensional basic pattern data is generated by using the two-dimensional basic pattern generating unit 12a in the basic pattern data generating means 12, and first, the three-dimensional basic pattern data is converted into two-dimensional pattern data. When converting the three-dimensional basic pattern data into two-dimensional pattern data, for example, the relationship between the three-dimensional pattern data and the two-dimensional pattern data stored in the database is referenced, or the modeling between the three-dimensional pattern data and the two-dimensional pattern data is performed using a function. At that time, in this embodiment, adjustments are made to the parts that maintain straight lines, the parts and the degree to which darts are opened, etc., by reflecting the results of experience and experiments. The two-dimensional basic pattern data generated in this way is as shown in FIG. 11.

[0141] <Step S340: Writing out the difference between the master and the flat basic pattern data> Next, the difference between the flat reference paper pattern data, which is the flat paper pattern data of the reference paper pattern data in the master data read out in step S300, and the flat basic paper pattern data generated in step S330 is written (step S340).

[0142] <Step S350: Acquiring flat basic paper pattern data reflecting the machine learning results> The difference between the two-dimensional reference paper pattern data and the two-dimensional basic paper pattern data written out in step S340 is reflected with the results of the machine learning model execution means 16, and new difference data is added to the two-dimensional reference paper pattern data of the master data to obtain two-dimensional basic paper pattern data reflecting the machine learning model (step S350). That is, in this step, the difference between the two-dimensional reference paper pattern data and the two-dimensional basic paper pattern data written out in step S340 is reflected with the results of the machine learning executed by the machine learning model execution means 16, and the result is transformed from the reference paper pattern data stored in the master data storage means 31 to obtain two-dimensional basic paper pattern data reflecting the machine learning result. That is, the basic paper pattern data generation means 12 cooperates with the machine learning model execution means 16 to obtain two-dimensional basic paper pattern data reflecting the machine learning result by reflecting the machine learning model on the two-dimensional basic paper pattern data.

[0143] As explained in Fig. 6, the machine learning in the machine learning means 15 links the amount of correction made to actually fit the subject's body to the flat basic pattern data mechanically and automatically generated by the flat basic pattern data generating means 12a, and learns data on the amount of correction and the part of correction, such as the basic pattern data was like this for this body type, but the actual corrected pattern data was like this. By learning the amount of correction and the part of correction linked to the body data, the accuracy of the flat basic pattern data generated in step S330 is improved.

[0144] <Step S360: Registration of basic pattern data> The two-dimensional basic pattern data reflecting the machine learning results generated in step S350 and the three-dimensional basic pattern data generated in step S320 are registered in the basic pattern data storage means 33 (step S360).

[0145] <Step S370: Generation of flat corrected paper pattern data> The flat basic pattern data, which has been created in step S350 and reflects the results of the machine learning, is modified to fit the subject's body type, and modified flat pattern data is generated (step S370). The modified flat pattern data is generated using the modified pattern data generating means 13, according to the balance theory, as described above with reference to Fig. 3. The modification in step S370 completes the flat pattern data, and the creation of the pattern data to be delivered to the subject is now complete.

[0146] <Step S380: Storing flat corrected paper pattern data> The flat corrected paper pattern data generated in step S370 is stored in the corrected paper pattern data storage means 34 (step S380).

[0147] <Step S390: Incorporating accessory data> The accessory data relating to the collar, pockets, etc. stored in the accessory data storage means 37 is incorporated into the flat corrected paper pattern data generated in step S370 (step S390). This completes the flat data for delivery.

[0148] In this manner, a pattern created by the pattern creating method using the jacket pattern learning system according to this embodiment is completed.

[0149] The present invention can be summarized as follows.

[0150] The upper garment pattern learning system of the present invention defines an upper front body part defined by a front neck line, which is the front line of the neck circumference, a front center line, which defines the center edge of the body, a bust line, which defines the chest circumference, an arm front edge, which defines the front edge of the armhole, and a front shoulder ridge line, with a front shoulder base point, which is the boundary between the front body part and the back body part, as an apex, and a back neck line, which is the back line of the neck circumference, and a back edge of the body, with a back shoulder base point, which is the boundary between the front body part and the back body part as an apex. The upper back body portion is defined by a back centerline that defines the chest circumference, a bustline that defines the chest circumference, an arm rear edge line that defines the rear edge of the armhole, and a rear shoulder ridge line, and the upper front body portion is divided into a front center portion and a front side portion by a shoulder front perpendicular line that hangs down vertically downward from the shoulder front base point, and the upper back body portion is divided into a back center portion and a back side portion by a shoulder rear perpendicular line that hangs down vertically downward from the shoulder rear base point.

[0151] The system further comprises a master data storage means for storing reference pattern data that serves as a standard for creating a pattern pattern according to an individual's body type, a subject physical data acquisition means for acquiring physical data of a subject for whom an upper garment pattern is to be created, a basic pattern data generation means for generating basic pattern data based on the reference pattern data and the subject physical data, a modified pattern data storage means for storing modified pattern data obtained by modifying the basic pattern data to fit the subject's body type, a machine learning data generation means for generating learning data required for machine learning based on information regarding the subject's physical data, information regarding the basic pattern data, and information regarding the modified pattern data, a machine learning means for training a machine learning model using the learning data generated by the machine learning data generation means, and a machine learning completed model execution means for executing processing by the machine learning model using the machine learning completed model learned by the machine learning means, and the basic pattern data generation means cooperates with the machine learning completed model execution means to acquire basic pattern data reflecting the machine learning results by reflecting the machine learning model on the basic pattern data.

[0152] According to this, by dividing an outerwear pattern into a front center and front sides, and a back center and back sides, and conducting machine learning using modified pattern data that has been modified to fit the subject's body shape, and reflecting the results of the machine learning in the basic pattern data, it is possible to conduct machine learning with a small amount of data, even for outerwear patterns that are difficult to balance in fabric.

[0153] The apparatus further comprises a modeling means for generating the reference paper pattern data, the basic paper pattern data and the modified paper pattern data based on a common model.

[0154] According to this method, the basic paper pattern data and the modified paper pattern data that are the basis for generating the learning data are generated based on a common model, so that the general outline can be learned quickly even with a small amount of data.

[0155] The modeling means further includes an area ratio calculation means for calculating an area ratio between the central chest portion and the side chest portion when the chest above the bustline is divided into the central chest portion and the side chest portion, with a line extending downward from the point where the frontal plane of the body and the line around the neck intersect as the boundary, based on the subject's physical data acquired by the subject's physical data acquisition means, and an area ratio adjustment means for adjusting a front area ratio, which is the area ratio between the front central portion and the front armpit portion, and a rear area ratio, which is the area ratio between the back center portion and the back armpit portion, based on the area ratio.

[0156] According to this, the front area ratio and the back area ratio of the original paper pattern are adjusted based on the area ratio of the center and sides of the chest of the wearer, so that the rotational movement of the fabric around the front and back shoulder base points can be appropriately induced so as to fit the wearer's body shape. Therefore, by simply obtaining the body measurements of the wearer, the correction of the paper pattern that can fit snugly to the wearer's body can be incorporated into the learning.

[0157] Furthermore, the areas of the front center and the front armpits are adjusted so that the front area ratio is approximately equal to the area ratio of the chest center and the chest lateral parts.

[0158] According to this, the areas of the front center and front armpits are adjusted so that the front area ratio is approximately equal to the area ratio of the center of the chest and the sides of the chest, so that corrections that fully take into account the wearer's body balance can be incorporated into the learning.

[0159] Although the embodiments of the present invention have been described above, the present invention is not limited to these embodiments. Furthermore, the effects described in the embodiments of the present invention are merely a list of the most preferable effects resulting from the present invention, and the effects of the present invention are not limited to those described in the embodiments of the present invention.

[0160] For example, the idea of ​​forming a paper pattern in consideration of the weight distribution of the fabric of each part in the present invention can be applied to not only outerwear such as jackets, but also pants and slacks, and can be theorized.

[0161] Furthermore, the above-described embodiment has been described in detail in order to clearly explain the present invention, and the present invention is not necessarily limited to having all of the configurations described. [Industrial Applicability]

[0162] The outerwear pattern learning system of the present invention can be applied not only to suit jackets, but also to general clothing worn on the upper body, such as coats and dress shirts. [Explanation of symbols]

[0163] N Front shoulder base M Posterior shoulder base BL Bustline CL anterior center line FB1 anterior center FB2 Front armpit BB1 heart of back BB2 Back side part PL Shoulder vertical line QL Shoulder vertical line

Claims

1. The upper front body part is defined by the front neck line, which is the front line of the neck circumference, the front center line, which defines the center edge of the body, the bust line, which defines the chest circumference, the arm front edge line, which defines the front edge of the armhole, and the front shoulder ridge line, with the front shoulder base point, which is the boundary between the front body part and the back body part, as the apex, and the back neck line, which is the rear line of the neck circumference, the back center line, which defines the back edge of the body, and the bust line, which defines the chest circumference, with the back shoulder base point, which is the boundary between the front body part and the back body part, as the apex. A learning system for an outerwear pattern having an upper back body portion defined by a back shoulder ridge line, an arm rear edge line defining the rear edge of an armhole, and a rear shoulder ridge line, the upper front body portion being divided into a front center portion and a front side portion by a shoulder front perpendicular line that hangs down vertically from the shoulder front base point, and the upper back body portion being divided into a back center portion and a back side portion by a shoulder rear perpendicular line that hangs down vertically from the shoulder rear base point, a master data storage means for storing reference paper pattern data that serves as a reference for creating paper pattern data according to individual body types; A subject's body data acquisition means for acquiring body data of a subject for which an upper garment pattern is to be created; A basic pattern data generating means for generating basic pattern data based on the reference pattern data and the subject's body data; a modified paper pattern data storage means for storing modified paper pattern data obtained by modifying the basic paper pattern data so as to fit the body shape of a subject; A machine learning data generating means for generating learning data necessary for machine learning based on information regarding the subject's body data, information regarding the basic paper pattern data, and information regarding the modified paper pattern data; a machine learning means for learning a machine learning model using the learning data generated by the machine learning data generation means; a machine-learned model execution means for executing a process based on the machine-learned model using the machine-learned model learned by the machine-learning means; The basic paper pattern data generating means cooperates with the machine learning model executing means to obtain basic paper pattern data on which a machine learning result has been reflected, the basic paper pattern data being subjected to a machine learning model. A learning system for outerwear paper patterns.

2. further comprising a modeling means for generating the reference paper pattern data, the basic paper pattern data and the modified paper pattern data based on a common model; 2. The upper garment paper pattern learning system according to claim 1.

3. The modeling means includes an area ratio calculation means for calculating an area ratio between the central chest area and the lateral chest area when the chest above the bustline is divided into the central chest area and the lateral chest area along a line extending downward from the intersection point of the frontal plane of the body and the neck line based on the subject's body data acquired by the subject's body data acquisition means; and and an area ratio adjusting means for adjusting a front area ratio, which is an area ratio between the front center portion and the front armpit portion, and a rear area ratio, which is an area ratio between the back center portion and the back armpit portion, based on the area ratio.

3. The upper garment paper pattern learning system according to claim 2.

4. The areas of the front center portion and the front armpit portion are adjusted so that the front side area ratio is approximately equal to the area ratio of the chest center portion and the chest side portion.

4. The upper garment paper pattern learning system according to claim 3.

5. An outerwear pattern obtained by using the outerwear pattern learning system according to any one of claims 1 to 4.

Citation Information

Patent Citations

  • How to correct outerwear pattern

    JP7130295B1

  • Grading system, method, and program

    JP7220324B1

  • Artificially intelligent unit-of-one pattern creation for bespoke garment fabrication

    US20210166068A1

  • Pattern automatic generation device, system, method, program, and recording medium

    WO2019021349A1

  • Information processing apparatus, information processing method, and program

    JP2018109998A