KR20220072729A - Method for measuring body size based on artificial intelligence - Google Patents

Abstract

The present invention relates to an artificial intelligence-based method for measuring a user's body size. An artificial intelligence-based user body size measurement method includes the steps of generating a training data set, training a body size estimation model using the training data set, a first image captured by the user in a first direction, and the first Providing the body size estimation model with a second image captured in a second direction different from the direction, and estimating at least one body size of the user using the first image and the second image. , The generating of the training data set includes: extracting a first three-dimensional modeling shape from three-dimensional human body data measured in advance and stored; and using the first three-dimensional modeling shape, first training images captured in different directions and generating a second training image.

Description

Method for measuring body size based on artificial intelligence

The present invention describes an artificial intelligence-based method for measuring a user's body size. Specifically, the present invention describes a method by which the user can measure the user's body size using a captured image.

Recently, the amount of online purchases of clothes and the like is increasing explosively. In online shopping, consumers cannot try on clothes directly, so it is necessary to select the correct size to purchase. To this end, the consumer directly measures his or her body size and purchases clothes of the matching size, or purchases clothes of the size recommended by the online shopping platform by entering the measured body size on the online shopping platform. method is being used.

However, since the general consumer does not know exactly how to measure the body size, the body position measured by the consumer may not be accurate or the accuracy of body size measurement may be deteriorated due to the wrong use of a measuring device. When the size of clothes is determined using the wrongly measured body size, satisfaction with fit decreases and this leads to exchange or return, which incurs time and money costs.

In order to reduce such cost, it is necessary to provide a method for accurately and simply measuring the body size, even if the user does not know exactly how to measure the body size.

An object of the present invention is to provide a method of measuring at least one body size using an image captured by a user.

The objects of the present invention are not limited to the above-mentioned objects, and other objects and advantages of the present invention not mentioned may be understood by the following description, and will be more clearly understood by the examples of the present invention. It will also be readily apparent that the objects and advantages of the present invention may be realized by the means and combinations thereof indicated in the appended claims.

An artificial intelligence-based user body size measurement method according to some embodiments of the present invention for solving the above problems includes the steps of generating a training data set, training a body size estimation model using the training data set, and providing a first image captured in a first direction and a second image captured in a second direction different from the first direction to the body size estimation model, and using the first image and the second image, The step of estimating at least one body size of the user, wherein the generating of the training data set includes: extracting a first three-dimensional modeling shape from the three-dimensional body data measured in advance and stored; and the first three-dimensional modeling shape. and generating a first training image and a second training image captured in different directions by using the same.

In some embodiments, the generating of the training data set includes: extracting first dimension information related to the first three-dimensional modeling shape from the three-dimensional human body data; The method may further include generating a second three-dimensional modeling shape by wearing the first virtual garment on the first three-dimensional modeling shape.

In some embodiments, the generating of the first training image and the second training image may include generating the first training image by capturing the second three-dimensional modeling shape in the first direction, and the second It may include generating the second training image by imaging the three-dimensional modeling shape in the second direction.

In some embodiments, the estimating of at least one body size of the user by using the first image and the second image includes performing a preprocessing operation on the first image and the second image to obtain an effective body width It may include generating data and estimating at least one body size of the user by providing the effective body width data to the trained body size estimation model.

In some embodiments, performing the preprocessing operation on the first image and the second image includes extracting a first feature point from the first image, and extracting a second feature point based on the first feature point and extracting a region of interest by using the second feature point, dividing the region of interest into a plurality of measurement regions, and measuring a body width for each of the plurality of measurement regions. .

In some embodiments, performing the preprocessing operation on the first image and the second image may further include determining at least one of the plurality of measured body widths as effective body width data.

In some embodiments, the estimating of at least one body size of the user may include performing a pre-processing operation on the first image and the second image by using reference information about the user.

In some embodiments, the pre-processing operation for the first image and the second image includes the effective width for the first image and the second image using the reference information about the user and a regression formula to determine effective width data It may include determining data.

In the artificial intelligence-based method of measuring body size of a user of the present invention, since the user measures body size using a captured image, accurate and easy body size measurement is possible regardless of the user's body size measurement proficiency.

In addition, the artificial intelligence-based user body size measurement method of the present invention can accurately and easily measure the body size only with a user terminal such as a smart phone without the need for a separate body size measurement device.

In addition to the above, the specific effects of the present invention will be described together while describing specific details for carrying out the present invention.

1 is a diagram schematically illustrating the configuration of a system for implementing an artificial intelligence-based body size measurement method according to some embodiments of the present invention.
2 is a view for explaining the configuration of a body size estimation model according to some embodiments of the present invention.
3 is a view for explaining the overall flow of an artificial intelligence-based user body size measurement method according to some embodiments of the present invention.
4 is a diagram for explaining the configuration of a training module of a body size estimation model according to some embodiments of the present invention.
5 is a diagram for explaining a process of generating a training data set by a training module according to some embodiments of the present invention.
6 to 10 are diagrams for exemplarily explaining a process of generating the training data set of FIG. 5 .
11 is a diagram for explaining a process of optimizing the body size algorithm by the training module 111 according to some embodiments of the present invention.
12 is a view for explaining a process of pre-processing a training image according to some embodiments of the present invention.
13 to 15 are diagrams for exemplarily explaining a pre-processing process of the training image of FIG. 12 .
16 is a diagram for explaining a configuration of an estimation module according to some embodiments of the present invention.
17 is a diagram for explaining an operation of an estimation module according to some embodiments of the present invention.
18 is a diagram for explaining the overall flow of an artificial intelligence-based user body size measurement method according to some other embodiments of the present invention.
19 is a diagram for explaining the configuration of a training module of a body size estimation model according to some other embodiments of the present invention.
20 is a diagram for explaining the configuration of an estimation module of a body size estimation model according to some embodiments of the present invention.
21 is a diagram for explaining a process of pre-processing an image according to another exemplary embodiment of the present invention.

Terms or words used in this specification and claims should not be construed as being limited to a general or dictionary meaning. In accordance with the principle that the inventor can define a term or concept of a word in order to best describe his/her invention, it should be interpreted as meaning and concept consistent with the technical idea of the present invention. In addition, since the embodiments described in this specification and the configurations shown in the drawings are only one embodiment in which the present invention is realized, and do not represent all the technical spirit of the present invention, they can be substituted at the time of the present application. It should be understood that there may be various equivalents and modifications and applicable examples.

Terms such as first, second, A, B, etc. used in this specification and claims may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, a first component may be referred to as a second component, and similarly, a second component may also be referred to as a first component. The term 'and/or' includes a combination of a plurality of related listed items or any of a plurality of related listed items.

Terms used in this specification and claims are used only to describe specific embodiments, and are not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. It should be understood that terms such as “comprise” or “have” in the present application do not preclude the possibility of addition or existence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification in advance. .

Unless defined otherwise, all terms used herein, including technical and scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

In addition, each configuration, process, process or method included in each embodiment of the present invention may be shared within a range that does not technically contradict each other.

Hereinafter, an artificial intelligence-based body size measurement method according to some embodiments of the present invention will be described with reference to FIGS. 1 to 17 .

1 is a diagram schematically illustrating the configuration of a system for implementing an artificial intelligence-based body size measurement method according to some embodiments of the present invention.

Referring to FIG. 1 , a system for implementing an artificial intelligence-based body size measurement method according to some embodiments of the present invention may include a server 100 , a user terminal 200 , and a communication network 300 .

The server 100 may receive, from the user terminal 200 through the communication network 300 , the first image Img_1 and the second image Img_2 captured by the user, and reference information about the user. The first image Img_1 , the second image Img_2 , and reference information about the user may be provided to the body size estimation model 110 in the server 100 . The first image Img_1 may be an image captured by the user in the first direction. Meanwhile, the second image Img_2 may be an image captured by the user in a second direction different from the first direction. For example, the first image Img_1 may be an image captured by the user from the front, and the second image Img_2 may be an image captured by the user from the side. The reference information may include at least one of a user's gender, age, height, and weight. Reference information about the user may be used in a pre-processing process for the image, and a detailed description will be given later. The body size estimation model 110 may estimate at least one body size of the user by using the received first image Img_1 , the second image Img_2 , and/or reference information about the user. For example, the body size estimation model 110 uses the received first image (Img_1), the second image (Img_2) and/or reference information about the user, the user's shoulder width, chest circumference, waist circumference, At least one of the user's body dimensions, such as a hip circumference, may be estimated.

Although only one server 100 is illustrated in the drawing, this does not mean that the server 100 is configured with only one server. For example, the server 100 includes a plurality of servers, and at least some steps of the artificial intelligence-based user body size estimation method described below may be performed on different servers. In other words, the server 100 may refer to one or more servers that perform the artificial intelligence-based user body size estimation method according to some embodiments of the present invention. In order to describe the body size estimation model 110 , further reference is made to FIG. 2 .

2 is a view for explaining the configuration of a body size estimation model according to some embodiments of the present invention.

Referring to FIG. 2 , the body size estimation model 110 may include a training module 111 and an estimation module 112 . Hereinafter, the term 'module' refers to a component that performs a specific function. Accordingly, different 'modules' refer to components that perform different functions, and do not necessarily mean different hardware. For example, the training module 111 may perform a function of optimizing an estimation algorithm using a training data set, and the estimation module 112 may perform a function of estimating a body size of a user. On the other hand, the training module 111 and the estimation module 112 may be configured with different hardware, or may be configured with at least some of the same hardware.

The training module 111 may be trained to optimize parameters of the user body size estimation algorithm used in the estimation module 112 . The training module 111 may optimize the parameters of the user body size estimation algorithm through supervised learning or unsupervised learning by using the training data set. A detailed description will be given later.

The estimation module 112 may design a user body size estimation algorithm by using the parameters optimized in the training module 111 , and may estimate the user body size through this. For example, the estimation module 112 may input the user's first image Img_1 and the second image Img_2 into an optimized user body size estimation algorithm to estimate at least one user body size. A detailed description will be given later.

Referring back to FIG. 1 , the user terminal 200 refers to a communication terminal capable of operating an application in a wired/wireless communication environment. The user terminal 200 may be applied without limitation to a device capable of operating an application. For example, the user terminal 200 may include various types of electronic devices such as a personal computer (PC), a notebook computer, a tablet, a mobile phone, a smart phone, and a wearable device.

In addition, although only one user terminal 200 is illustrated in the drawing, the present invention is not limited thereto, and the server 100 may operate in conjunction with a plurality of user terminals 200 .

As used herein, the term 'application' may mean any series of steps or software executed in a computing device. For example, the application may include a mobile application, a web application, and a web page.

Additionally, the user terminal 200 includes an input unit for receiving a user's input, a display unit for displaying visual information, a communication unit for transmitting and receiving signals with the outside, a camera unit for photographing the user's face, and the user's voice as digital data. It may include a microphone unit that converts, and a control unit that processes data, controls each unit inside the user terminal 200, and controls data transmission/reception between the units.

Meanwhile, the communication network 300 serves to connect the server 100 and the user terminal 200 . That is, the communication network 300 refers to a communication network that provides an access path so that the user terminal 200 can transmit and receive data after accessing the server 100 . The communication network 300 is, for example, a wired network such as LANs (Local Area Networks), WANs (Wide Area Networks), MANs (Metropolitan Area Networks), ISDNs (Integrated Service Digital Networks), etc., wireless LANs, CDMA, Bluetooth, satellite communication, etc. may cover a wireless network, but the scope of the present invention is not limited thereto.

3 is a diagram for explaining the overall flow of an artificial intelligence-based user body size measurement method according to some embodiments of the present invention.

1 to 3 , the training module 111 of the body size estimation model 110 may generate a training data set ( S100 ). The training module 111 may generate a training data set by processing pre-stored three-dimensional body data. For a detailed description, further reference is made to FIGS. 4 to 10 .

4 is a diagram for explaining the configuration of a training module of a body size estimation model according to some embodiments of the present invention. 5 is a diagram for explaining a process of generating a training data set by a training module according to some embodiments of the present invention. 6 to 10 are diagrams for exemplarily explaining a process of generating the training data set of FIG. 5 .

4 and 5 , the training module 111 may include a training data set generation module (DGM), a first preprocessing module (PPM_1), and a body size estimation algorithm optimization module (AOM).

The training data set generating module DGM may receive the three-dimensional body data 3D_d (S101). The three-dimensional body data 3D_d may include a first three-dimensional modeling shape 3DM_1 and related first body dimension information Info_size. The first three-dimensional modeling shape 3DM_1 may be a modeling shape obtained by three-dimensionally scanning the human body. The first body size information Info_size may include at least one piece of information about body dimensions, such as a chest circumference, a waist circumference, and a hip circumference, of the first three-dimensional modeling shape 3DM_1 .

As shown in FIG. 6 , the training data set generating module DGM may extract the first three-dimensional modeling shape 3DM_1 from the three-dimensional body data 3D_d ( S102 ). According to some embodiments, since the first three-dimensional modeling shape 3DM_1 is a modeling shape obtained by scanning an actual human body, it may be a nude shape or a shape wearing only underwear.

The training data set generating module DGM may extract the first body size information Info_size of the first three-dimensional modeling shape 3DM_1 from the three-dimensional body data 3D_d ( S103 ).

As shown in FIG. 7 , the training data set generating module DGM may generate a first virtual garment CTH_1 by using the first body size information Info_size ( S104 ). Even when the first virtual garment CTH_1 is worn on the first three-dimensional modeling shape 3DM_1, the first virtual garment CTH_1 may be created to be tight so that body flexure and body shape information of the first three-dimensional modeling shape 3DM_1 can be expressed.

As shown in FIGS. 8 and 9 , the training data set generation module (DGM) generates a second three-dimensional modeling shape (3DM_2) using a first three-dimensional modeling shape (3DM_1) and a first virtual garment (CTH_1). It can be (S105). The training data set generation module DGM may wear the first virtual garment CTH_1 on the first three-dimensional modeling shape 3DM_1. A first three-dimensional modeling shape 3DM_1 wearing the first virtual garment CTH_1 is defined as a second three-dimensional modeling shape 3DM_2 for convenience.

As shown in FIG. 9 , the training data set generating module DGM may generate a first training image tImg_1 by imaging the second three-dimensional modeling shape 3DM_2 in the first direction ( S106 ). The first training image tImg_1 may be a two-dimensional image. For example, the training data set generating module DGM may capture the second three-dimensional modeling shape 3DM_2 from the front direction and generate it as the first training image tImg_1 .

As shown in FIG. 10 , the training data set generating module DGM may generate a second training image tImg_2 by imaging the second three-dimensional modeling shape 3DM_2 in a second direction different from the first direction. (S107). Similarly, the second training image tImg_2 may be a two-dimensional image. For example, the training data set generating module DGM may image the second three-dimensional modeling shape 3DM_2 in the lateral direction, and may generate it as the second training image tImg_2.

The training data set generating module DGM may store the first training image tImg_1 , the second training image tImg_2 , and the first body size information Info_size in the training data set ( S108 ).

According to some embodiments, the first three-dimensional modeling shape 3DM_1 may be a nude shape or a shape wearing only underwear. However, as will be described later, the user body size estimation model 110 uses an image captured by the user as input data. If the first three-dimensional modeling shape 3DM_1 is used as the training data as it is, the user's body size estimation model 110 is trained in a state of being naked or wearing only underwear. In this case, in order to measure the user's body size, the user must take an image of himself or herself in a state of being naked or wearing only underwear, and there is a risk that the sensitive image of the user is leaked to the outside by an external attack such as hacking. .

However, the training data set generation module (DGM) according to some embodiments of the present invention generates the first virtual clothing (CTH_1) and wears it on the first three-dimensional modeling shape (3DM_1), so that the first state in which the clothes are worn A two-dimensional modeling shape (3DM_2) can be created. The training data set generation module (DGM) generates a training data set by imaging the second three-dimensional modeling shape 3DM_2, and the body size estimation model 110 trains through this, so the user measures the body size while wearing clothes. The estimation model 110 may be used.

According to some embodiments, the three-dimensional body data 3D_d may be a three-dimensional modeling shape that is illustrated by scanning an actual human body, and data obtained by measuring and storing dimensional information about the corresponding human body in advance. In other words, the three-dimensional body data 3D_d may be a database that is measured and stored in advance according to a policy request, a statistical need, or a research purpose. For example, the three-dimensional body data 3D_d may be three-dimensional body data of Size Korea (a Korean body size research project). That is, since the process of generating a training data set according to some embodiments of the present invention processes and utilizes a previously measured and stored database, there is no need for a process of newly collecting training data to prepare a training data set. It can dramatically reduce the cost and time.

Referring back to FIG. 3 , the training module 111 may optimize the body size estimation algorithm using the training data set ( S110 ). In order to describe the optimization process of the body size estimation algorithm, further reference is made to FIG. 11 .

11 is a diagram for explaining a process of optimizing the body size algorithm by the training module 111 according to some embodiments of the present invention.

4 and 11 , the first preprocessing module PPM_1 may receive a first training image tImg_1 and a second training image tImg_2 from the training data set generation module DGM (S111).

The first pre-processing module PPM_1 may pre-process the first training image tImg_1 and the second training image tImg_2 to generate effective body width training data T_AW ( S112 ). In order to describe the preprocessing of the first training image tImg_1 and the second training image tImg_2, further reference is made to FIGS. 12 to 15 .

12 is a view for explaining a process of pre-processing a training image according to some embodiments of the present invention. 13 to 15 are diagrams for exemplarily explaining a pre-processing process of the training image of FIG. 12 .

4, 12, and 13 , the first pre-processing module PPM_1 may receive a first training image tImg_1 and a second training image tImg_2. The first preprocessing module PPM_1 may perform a preprocessing operation on the first training image tImg_1 and the second training image tImg_2. Hereinafter, for convenience of explanation, only the preprocessing process for the first training image tImg_1 will be described. The preprocessing process for the second training image tImg_2 may be the same as or similar to the preprocessing process for the first training image tImg_1 to be described below. The first pre-processing module PPM_1 may extract the first feature point FP_1 from the first training image tImg_1 ( S112_1 ). The first feature point FP_1 may be a point that can be relatively easily classified in the first training image tImg_1 . For example, the first feature point FP_1 may be at least one of a nose, a shoulder, a wrist, and a pelvis, but embodiments are not limited thereto. Extracting the first feature point FP_1 may use, for example, a pose estimation model, but embodiments are not limited thereto.

Referring further to FIG. 14 , the first preprocessing module PPM_1 may extract a second feature point FP_2 based on the first feature point FP_1 ( S112_2 ). The second feature point FP_2 may be a point that is relatively difficult to classify in the first training image tImg_1 . For example, the second feature point FP_2 may include at least one of an armpit point and a groin point. The first preprocessing module PPM_1 may extract the second feature point FP_2 by sampling a region around the first feature point FP_1 that is relatively easy to classify. For example, the first preprocessing module (PPM_1) extracts the positions of both shoulders as the first feature points (FP_1), samples the area around the positions of both shoulders, and extracts the underarm points as the second feature points (FP_2). have.

15 , the first pre-processing module PPM_1 may extract the region of interest IR using the second feature point FP_2 ( S112_3 ). The region of interest IR may be a region associated with a body size to be estimated. For example, the region of interest IR may be a region between an underarm point that is the second feature point FP_2 and a groin point that is another second feature point FP_2 .

The first preprocessing module PPM_1 may divide the region of interest IR into one or more measurement regions MR (S112_4). The measurement region MR may be a region in which the body width W associated with the body size is measured.

The first preprocessing module (PPM_1) measures the body width (W) for each of the one or more measurement regions (MR) (S112_5), and trains the effective body width among the body widths (W) for each of the one or more measurement regions (MR) The data T_AW may be determined (S112_6). The effective body width training data T_AW means a body width most related to a body size to be estimated. For example, among one or more body widths, since the body width between the underarm points has the highest correlation with the chest circumference, the effective body width training data T_AW for the chest circumference may be determined as the body width between the underarm points. The first pre-processing module PPM_1 may provide the determined effective body width training data T_AW to the body size estimation algorithm optimization module AOM ( S112_7 ).

According to some embodiments, determining the effective body width training data T_AW may be determined according to various algorithms. For example, the effective body width training data T_AW may be determined as the largest value among a plurality of body widths, or may be determined by a regression equation for determining the effective body width training data T_AW. A person of ordinary skill in the art may come up with a method for determining various effective body width training data T_AW without departing from the scope of the present invention.

According to some embodiments, the second feature point FP_2 may be determined based on the first feature point FP_1 . Since the second feature point FP_2 is difficult to distinguish from the background and the classification accuracy is low, the first feature point FP_1, which is relatively easy to classify, is first extracted, and the second feature point FP_2 can be extracted based on this. . Accordingly, according to some embodiments of the present invention, by performing the extraction of the second feature point FP_2 step by step, the accuracy of classification of the second feature point FP_2 may be increased.

Referring back to FIGS. 4 and 11 , the body size estimation algorithm optimization module AOM may receive the effective body width training data T_AW from the first preprocessing module PPM_1 . Also, the body size estimation algorithm optimization module AOM may receive the first body size information Info_size from the training data set generation module DGM. The body size estimation algorithm optimization module AOM may optimize the body size estimation algorithm by using the effective body width training data T_AW and the first body size information Info_size ( S113 ). The body size estimation algorithm may be an algorithm related to a relational expression between the effective body width training data T_AW and the first body size information Info_size. The body size estimation algorithm optimization module (AOM) calculates a loss function for the estimated body size derived from the effective body width training data (T_AW) and the first body size information (Info_size), and uses this to calculate the parameters for the algorithm. By making adjustments, the body size estimation algorithm can be optimized.

Referring back to FIG. 3 , the first image Img_1 and the second image Img_2 captured by the user may be provided to the estimation module 112 of the body size estimation model 110 ( S120 ). The first image Img_1 may be an image captured by the user in the first direction, and the second image Img_2 may be an image captured by the user in the second direction. As described above, the first image Img_1 and the second image Img_2 may be images captured by the user while wearing clothes.

The estimation module 112 of the body size estimation model 110 may estimate the body size of the user by using the received first image Img_1 and the second image Img_2 ( S130 ). For further explanation, reference is made further to FIGS. 16 and 17 .

16 is a diagram for explaining the configuration of an estimation module according to some embodiments of the present invention. 17 is a diagram for explaining an operation of an estimation module according to some embodiments of the present invention. For convenience of description, the same or similar contents as those described above will be omitted or simply described.

16 and 17 , the estimation module 112 may include a second preprocessing module PPM_2 and a body size estimation algorithm execution module SEM. The second pre-processing module PPM_2 may perform substantially the same function as the first pre-processing module PPM_1. The body size estimation algorithm execution module SEM may receive the algorithm parameter O_para optimized from the body size estimation algorithm optimization module AOM and use it.

The second pre-processing module PPM_2 may receive a first image Img_1 and a second image Img_2 captured by the user (S131). The second pre-processing module PPM_2 may pre-process the first image Img_1 and the second image Img_2 to determine at least one effective body width data AW ( S132 ). The second pre-processing module PPM_2 may provide the determined at least one effective body width data AW to the body size estimation algorithm execution module SEM ( S133 ). The body size estimation algorithm execution module SEM may estimate at least one body size of the user using the optimized algorithm parameter O_para and at least one effective body width data AW ( S134 ).

According to some embodiments, the user may measure a relatively accurate body size by simply capturing an image of the user and providing the image to the body size estimation model without a separate measuring device or separate knowledge. In addition, since the body size estimation model is trained using training data wearing virtual clothes, the user can take an image while wearing all the clothes, and there is no need to worry about leakage of sensitive images. Hereinafter, an artificial intelligence-based body size measurement method according to some other embodiments of the present invention will be described with reference to FIGS. 18 to 21 .

18 is a view for explaining the overall flow of an artificial intelligence-based user body size measurement method according to some other embodiments of the present invention. 19 is a diagram for explaining the configuration of a training module of a body size estimation model according to some embodiments of the present invention. 20 is a diagram for explaining the configuration of an estimation module of a body size estimation model according to some embodiments of the present invention. For convenience of description, the same or similar contents as those described above will be omitted or simply described.

18 to 19 , the training data set generating module (DGM) may generate a training data set (S200). The training module 111 may generate a training data set by processing the pre-stored three-dimensional body data 3D_d. The three-dimensional body data 3D_d includes a first three-dimensional modeling shape 3DM_1, first body dimension information Info_size related to the first three-dimensional modeling shape 3DM_1, and first reference information related to the first three-dimensional modeling shape 3DM_1 (Info_user) may be included. The first reference information Info_user may include the height, weight, gender, and age of the first three-dimensional modeling shape 3DM_1.

The training module 111 may train the body size estimation model 110 using the training data set generated by the training data set generation module DGM (S210). The training data set generating module DGM may provide the first training image tImg_1 , the second training image tImg_2 , and the first reference information Info_user to the first pre-processing module PPM_1 . Also, the training data set generation module DGM may provide the first body size information Info_size to the body size estimation algorithm optimization module AOM. The first preprocessing module PPM_1 may perform a preprocessing operation using the first training image tImg_1 , the second training image tImg_2 , and the first reference information Info_user.

The first image Img_1 and the second image Img_2 captured by the user may be provided to the body size estimation model 110 ( S220 ). In addition, the user's first reference information Info_user may also be provided to the body size estimation model 110 ( S230 ). The body size estimation model 110 may estimate the body size of the user by using the first image Img_1 , the second image Img_2 , and the first reference information Info_user ( S240 ).

Referring further to FIG. 20 , the second preprocessing module PPM_2 may receive the first image Img_1 , the second image Img_2 , and the first reference information Info_user. That is, the second preprocessing module PPM_2 may perform a preprocessing operation using the first image Img_1 , the second image Img_2 , and the first reference information Info_user.

That is, according to some embodiments, the first reference information Info_user may be additionally used to pre-process an image. For a detailed description, refer further to FIG. 21 .

21 is a view for explaining a process of pre-processing an image according to another exemplary embodiment of the present invention.

Referring to FIG. 21 , the pre-processing module (the first pre-processing module PPM_1 and the second pre-processing module PPM_2) may extract a first feature point from the image ( S300 ). The pre-processing module may extract the second feature point based on the first feature point (S310). The pre-processing module may extract a region of interest by using the extracted second feature point ( S320 ). The preprocessing module may divide the ROI into a plurality of measurement regions ( S330 ), and measure a body width for each of the plurality of measurement regions ( S340 ). The pre-processing module may determine the effective body width W by using the first reference information Info_user (S350).

When the pre-processing module determines the effective body width W, the first reference information Info_user may be used. According to some embodiments, a regression equation for determining the effective body width data AW may be derived through statistical data on the effective body width data AW by sex, age, height, weight, and the like. The preprocessing module may effectively determine the effective body width data AW of the user by using the effective body width data AW determination regression equation and the first reference information Info_user.

The above description is merely illustrative of the technical idea of this embodiment, and a person skilled in the art to which this embodiment belongs may make various modifications and variations without departing from the essential characteristics of the present embodiment. Accordingly, the present embodiment is intended to explain, not to limit the technical spirit of the present embodiment, and the scope of the technical spirit of the present embodiment is not limited by these embodiments. The protection scope of this embodiment should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be interpreted as being included in the scope of the present embodiment.

Claims (8)

generating a training data set;
training a body size estimation model using the training data set;
providing a first image captured in a first direction of a user and a second image captured in a second direction different from the first direction to the body size estimation model; and
Comprising the step of estimating at least one body size of the user by using the first image and the second image,
The step of generating the training data set comprises:
extracting a first three-dimensional modeling shape from the three-dimensional human body data measured in advance and stored; and
Using the first three-dimensional modeling shape, comprising the step of generating a first training image and a second training image captured in different directions
How to measure your body size.
According to claim 1,
The step of generating the training data set comprises:
extracting first dimension information related to the first three-dimensional modeling shape from the three-dimensional human body data;
generating a first virtual garment by using the first dimension information; and
Wearing the first virtual garment on the first three-dimensional modeling shape, the user body size measurement method further comprising the step of generating a second three-dimensional modeling shape.
3. The method of claim 2,
The step of generating the first training image and the second training image,
generating the first training image by imaging the second three-dimensional modeling shape in the first direction; and
and generating the second training image by capturing the second three-dimensional modeling shape in the second direction.
According to claim 1,
The step of estimating at least one body size of the user by using the first image and the second image,
generating effective body width data by performing a pre-processing operation on the first image and the second image; and
and estimating at least one body size of the user by providing the effective body width data to the trained body size estimation model.
5. The method of claim 4,
The step of performing a pre-processing operation on the first image and the second image,
extracting a first feature point from the first image;
extracting a second feature point based on the first feature point;
extracting a region of interest by using the second feature point;
dividing the region of interest into a plurality of measurement regions; and
and measuring a body width for each of the plurality of measurement areas.
6. The method of claim 5,
The step of performing a pre-processing operation on the first image and the second image,
The method further comprising determining at least one of the plurality of measured body widths as effective body width data.
According to claim 1,
The step of estimating at least one body size of the user,
A method of measuring a user's body size by performing a pre-processing operation on the first image and the second image by using the reference information about the user.
8. The method of claim 7,
The pre-processing operation for the first image and the second image,
and determining effective width data for the first image and the second image using reference information about the user and a regression equation for determining effective width data.

Images