Body mass index estimation method, body mass index estimation system, body mass index estimation device, and body mass index estimation program

Abstract

To estimate indicators related to body size. [Solution] The body size index estimation method includes an acquisition step of acquiring a user side image taken from the side of the user; a region determination step of determining a specific region of the user's body in the user side image based on a specific position of the user in the user side image; and an estimation step of estimating an index related to the user's body size. The estimation step estimates the index related to the user's body size based on a numerical value indicating the specific region in the user side image and body size index information indicating the correlation between the numerical value indicating the specific region and the index related to body size.

Description

[Technical Field]

[0001] The present invention relates to a method for estimating body mass index. [Background technology]

[0002] There is a method for estimating the width and thickness of the frontal and lateral sections of the torso near the navel of a person being estimated. This method involves placing a pair of hands against the navel area to estimate the length (Patent Document 1). [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2018-198800 [Overview of the project] [Problems that the invention aims to solve]

[0004] Patent Document 1 estimated the length of the torso near the navel using frontal and side images of the subject, but there was room for improvement because it required images of the subject taken from multiple directions. Furthermore, it had the problem of not being able to address the need to grasp other indicators related to body size besides the length of the torso near the navel.

[0005] The present invention has been made in view of the above problems, and relates to a method for estimating body size indicators for estimating indicators related to a user's body size from a side image of the user. [Means for solving the problem]

[0006] The present invention relates to a method for estimating a body size index, comprising: an acquisition step of acquiring a user side image captured from the side of a user; a region determination step of determining a specific region of the user's body in the user side image based on a specific position of the user in the user side image; and an estimation step of estimating an index related to the user's body size, wherein the estimation step estimates the index related to the user's body size based on a numerical value indicating the specific region in the user side image and body size index information indicating a correlation between the numerical value indicating the specific region and the index related to body size.

[0007] Furthermore, the present invention relates to a body size index estimation system that uses a control device to estimate an index related to a user's body size, comprising: acquisition means for acquiring a user side image captured from the side of the user; region determination means for determining a specific region of the user's body in the user side image based on a specific position of the user in the user side image; and estimation means for estimating an index related to the user's body size, wherein the estimation means estimates the index related to the user's body size based on a numerical value indicating the specific region in the user side image and body size index information indicating a correlation between the numerical value indicating the specific region and the index related to body size.

[0008] Furthermore, the present invention relates to a body size index estimation device comprising: acquisition means for acquiring a user side image captured from the side of a user; region determination means for determining a specific region of the user's body in the user side image based on a specific position of the user in the user side image; and estimation means for estimating an index related to the user's physique, wherein the estimation means estimates the index related to the user's physique based on a numerical value indicating the specific region in the user side image and physique index information indicating a correlation between the numerical value indicating the specific region and the index related to physique.

[0009] Furthermore, the present invention relates to a body size index estimation program for estimating an index related to a user's physique, comprising: an acquisition step of acquiring a user side image captured from the side of the user; a region determination step of determining a specific region of the user's body in the user side image based on a specific position of the user in the user side image; an estimation step of estimating an index related to the user's physique, wherein the estimation step estimates the index related to the user's physique based on a numerical value indicating the specific region in the user side image and body size index information indicating the correlation between the numerical value indicating the specific region and the index related to physique, and the present invention relates to a body size index estimation program that causes the above steps to be executed. [Effects of the Invention]

[0010] According to the method provided by the present invention, it is possible to estimate indicators related to the user's physique from a side view image of the user. [Brief explanation of the drawing]

[0011] [Figure 1] (a) is an illustrative image of the image taken from the front of the user, and (b) is an illustrative image of the image taken from the side of the user. [Figure 2] (a) is a diagram showing the user's front view and specific location, and (b) is a diagram showing the user's side view and specific location. [Figure 3] This diagram shows the processing flow of the body mass index estimation method. [Figure 4] (a) is a diagram showing a front view of the user, (b) is an image showing the front view of the user after segmentation processing, and (c) is an image showing the side view of the user after segmentation processing. [Figure 5] (a) is a diagram showing a side view of the user, (b) is a diagram showing a specific location on the side view of the user, and (c) is a diagram showing a specific region in Case 1. [Figure 6] This figure shows the body size index information for Case 1. [Figure 7](a) is a diagram showing a user side image, (b) is a diagram showing a specific position in the user side image, and (c) is a diagram showing a specific area in Case 2. [Figure 8] It is a diagram showing physical index information for Case 2. [Figure 9] (a) is a diagram showing a specific area in Case 4, and (b) is a diagram showing physical index information for Case 4. [Figure 10] It is a diagram showing physical index information for Case 5. [Figure 11] It is a diagram showing physical index information for Case 6. [Figure 12] It is a diagram showing physical index information for Case 7. [Figure 13] It is a diagram showing physical index information for Case 8. [Figure 14] It is a diagram showing physical index information for Case 9. [Figure 15] It is a diagram showing physical index information for Case 10. [Figure 16] It is a diagram showing physical index information for Case 11. [Figure 17] It is a block diagram of a physical index estimation system. [Figure 18] It is a block diagram of a physical index estimation device.

Embodiments for Carrying Out the Invention

[0012] Hereinafter, examples of preferred embodiments of the present invention will be described with reference to the drawings. Note that the drawings of this embodiment are all for explaining the technical idea, configuration, and operation of the present invention, and do not specifically limit the configuration. Also, in all the drawings, the same reference numerals are given to the same components, and duplicate explanations are omitted as appropriate. Note that in this specification, a "system" is assumed to include one or more information processing devices. For example, a single information processing device can also constitute a system, and when a plurality of information processing devices cooperate to perform functions such as a web server, these information processing devices can also constitute a system.

[0013] An overview of the body mass index estimation method in this embodiment (hereinafter sometimes referred to as "this method") will be described below. The body size index estimation method of this embodiment includes an acquisition step of acquiring a user side image captured from the side of the user, a region determination step of determining a specific region of the user's body in the user side image based on a specific position of the user in the user side image, and an estimation step of estimating an index related to the user's body size, wherein the estimation step estimates the index related to the user's body size based on a numerical value indicating the specific region in the user side image and body size index information indicating the correlation between the numerical value indicating the specific region and the index related to body size.

[0014] Figure 1(a) shows an image of the user from the front, and Figure 1(b) shows an image of the user from the side. Additionally, Figure 2(a) shows an image of the user from the front, and Figure 2(b) shows an image of the user from the side. "User" can be any person regardless of gender, age, or height. "Front view" includes not only the direction in which the user's face is facing straight ahead, the direction in which the user can directly view the imaging device 20 while standing upright, the direction in which the user's navel is approximately parallel to the imaging device 20 while standing upright, and the direction in which the chest is located, but also the direction in which the back of the user's head is located and the direction in which the back of the user's head is located. "Side view" refers to the direction in which the user's front view is rotated approximately 90 degrees around the vertical axis, and includes both the right side (the user's right side) and the left side (the user's left side). "Image taken" means taking an image using the imaging device 20, which may be a general RGB camera, a monochrome camera, or a spectral camera, and its performance and specifications are not limited. The imaging device 20 includes video cameras, cameras built into smartphones, cameras built into tablet devices, and webcams that can be attached to personal computers, etc., by means of connection such as cables. "User side view image" refers to an image of the user's side taken by the imaging device 20 described above, and includes both images taken from the right side (the user's right side) and images taken from the left side (the user's left side).

[0015] Figure 3 shows the processing flow of the body mass index estimation method. "Acquiring" means acquiring an image, and the source of the user-side image is irrelevant, including acquiring it from a predetermined medium, acquiring it via a network, or acquiring it directly from the imaging device 20 (direct capture from the imaging device 20). The "acquisition process" is the process of acquiring a user side image, which is an image of the user's side as described above, and is step S100 shown in Figure 3.

[0016] "User's specific position" refers to a position in the user's side view image that satisfies predetermined conditions, and is uniquely determined by satisfying predetermined conditions such as the user's skeletal features, the line connecting the user's skeletal features to other features, and the height of the user's center of gravity in the direction of the user's height. "Specific area of ​​the user's body" refers to a predetermined range from a specific position of the user in a lateral image of the user, a predetermined range including the specific position, or a predetermined range based on the specific position, and is an area uniquely determined from the specific position of the user. The "domain determination process" is the process of determining a specific area of ​​the user's body as described above, and is step S200 shown in Figure 3.

[0017] "User body size indicators" refer to indicators related to body size, such as body characteristics, body type, and the level at which body size affects health. In particular, they refer to indicators related to body fat and visceral fat (influenced by body fat and visceral fat), specifically BMI (Body Mass Index), obesity level, weight, height, body fat percentage, muscle mass, and visceral fat mass. The "estimation process" is the process of estimating the indicators related to the user's physique, as described above, and is step S300 shown in Figure 3.

[0018] Here, as an example of a specific location of the user that satisfies predetermined conditions in the user's side view image, we will describe the positional relationship of multiple feature points on the user's body.

[0019] In this embodiment, the user stands in a predetermined position (for example, the state shown in Figure 1(a)), and while standing in the same position such that the position of the user and the imaging device 20 remain substantially unchanged, the user rotates around the vertical axis while capturing moving images. Figure 1(b) shows the state in which the user's side is approximately parallel to the imaging device 20 when the user is rotating while capturing images. Thus, it is preferable that the user's side image or user's front image is a moving image consisting of multiple frames.

[0020] Figure 2(a) shows the positions of multiple feature points on the user's body in a frontal view image of the user. "User frontal image" refers to an image of the user taken from the front of the user, and "front" means not only the direction in which the user's face is facing straight ahead, the direction in which the user can directly look at the imaging device 20 while standing upright, the direction in which the user's navel is approximately parallel to the imaging device 20 while standing upright, and the direction in which the chest is located, but also the direction in which the back of the user's head is located and the direction in which the back of the user is located. In this embodiment, as will be described in detail later, it typically includes a feature point identification step in which a user front image is acquired by capturing the user's front view, and the positions of multiple feature points of the user's body in the user front image are identified. That is, since the specific position is identified using the positions of multiple feature points of the user's body in the user front image, it is preferable that the acquisition step (step S100) acquires a front image in addition to a side image of the user, from the viewpoint of estimating an index related to the user's physique with higher accuracy.

[0021] "Multiple characteristic points on the user's body" refer to points (locations) that can identify specific positions on the body, such as skeletal points like the top of the head, neck, acromion, wrists, and ankles; points related to the position of bones such as joints; and points related to the position of body parts such as the corners of the eyes, the tip of the nose, and the fingertips. In this embodiment, for example, MediaPipe Pose is used to extract skeletal points that are used to determine height (e.g., crown of head, heels, etc.), skeletal points that are highly correlated with height (e.g., fingertips, shoulders, neck, pelvis, knees, heels, etc.), skeletal points that tend to differ depending on gender (e.g., pelvis, shoulders, rib cage, etc.), skeletal points related to the upper limbs (e.g., neck, shoulders, elbows, wrists, etc.), and skeletal points related to the lower limbs (e.g., buttocks, knees, ankles, heels, etc.). In this embodiment, skeletal points used to determine height (crown of head 10, heels 13), skeletal points related to the upper limbs (shoulders (acromion and shoulder joint) 11), and skeletal points related to the lower limbs (buttocks (anterior superior iliac spine (pelvis and hip joint)) 12) are extracted, and each of these extracted skeletal points is used as a feature point. The image coordinates (X (horizontal direction), Y (vertical direction)) of each extracted feature point are also identified. The locations of the feature points mentioned above are identified by the "feature point identification process." The "feature point identification process" should be performed after the acquisition process (step S100) in the processing flow shown in Figure 3, and before the region determination process (step S200).

[0022] The dotted lines from 10 to 13 in Figure 2(b) correspond to the dotted lines from 10 to 13 in Figure 2(a), respectively, and indicate the height positions corresponding to 10 to 13 in Figure 2(a). In known skeletal information acquisition techniques, skeletal points are difficult to extract from the user's side image alone. However, as in this embodiment, by capturing multiple frames of images and transitioning the position (vertical height position) of the skeletal points based on the skeletal points extracted from the front image, it becomes possible to identify the position (vertical height position) of feature points (skeletal points) in the side image. Thus, in the feature point identification process, it is preferable to identify feature points in the side image as well, from the viewpoint of estimating indicators related to the user's physique with higher accuracy.

[0023] From among the identified feature points, the specific location in the user's side view image is determined based on the positional relationship between the feature point and other feature points. "Identifying a specific location in a user's side view image based on the positional relationship between a feature point and other feature points" means that the location is identified by the positions of multiple feature points, and examples include interpolation positions between a feature point and other feature points, and extrapolation positions between a feature point and other feature points. Specifically, these include a predetermined position on a virtual line connecting a feature point and other feature points, a predetermined ratio position, a center position, and a position located a predetermined ratio of the distance from the feature point to the other feature points. In addition to interpolation and extrapolation positions, other examples include the intersection point of a virtual line at a predetermined angle from a feature point and a virtual line at a predetermined angle from another feature point. Alternatively, it may be a position on the perpendicular bisector of the virtual line connecting a feature point and other feature points, where the position is a predetermined multiple of the distance between the two points. Furthermore, in addition to the feature point and other feature points, there may be other feature points (three or more feature points), for example, the centroid position of all feature points or the average value obtained by assigning a predetermined weight to the coordinates of each feature point. Alternatively, the intersection point may be the point where a virtual line connecting a feature point (referred to as the first feature point) and another feature point (referred to as the second feature point) intersects with a virtual line connecting a feature point different from that feature point and the other feature point (referred to as the third feature point) and another feature point different from that feature point (referred to as the fourth feature point).

[0024] In this embodiment, for example, a feature point is a point corresponding to the user's shoulder joint (shoulder 11), and another feature point is a point corresponding to the user's hip joint (buttocks 12). The positional relationship between the point corresponding to the user's shoulder joint and the point corresponding to the user's hip joint is used to identify a specific position in the user's lateral image. Typically, as shown in Figure 2(a), a position 2 / 3 of the way from the shoulder 11 is determined relative to a virtual line 21 connecting the shoulder 11 and the hip 12, and this position is designated as the specific position 22. This is because, as a result of diligent research by the inventors of the present invention, it is important that the specific position used to determine the specific region in this embodiment is the user's abdominal position. This relationship was derived after extensive research into what positional relationships between various feature points extracted from images of multiple subjects constitute the abdominal position. In other words, the inventors identified multiple feature points correlated with the abdominal position and found a relationship that shows the relationship between these multiple feature points. Note that although a virtual line 21 is shown in Figure 2(a) for ease of explanation, it is not necessary to include the virtual line 21 as the coordinates of each acquired feature point are used. In the above example, a position 2 / 3 of the way from the shoulder 11 was determined relative to the imaginary line 21 connecting the shoulder 11 and the hip 12, and this position was designated as the specific position 22. However, the specific position 22 can be appropriately changed relative to the imaginary line 21 connecting the shoulder 11 and the hip 12, between 3 / 5 and 4 / 5 of the way from the shoulder 11, and it is preferable that it be between 2 / 3 and 3 / 4 of the way from the shoulder 11. Then, based on the specific position identified using the user's front image, the specific position in the user's side image is identified. For example, as shown in Figure 2(b), the specific position 22 (vertical height position) in Figure 2(a) is transitioned to identify the specific position 22 in the user's side image. It is preferable to have a identification step that identifies the specific position in the user's side image based on the positional relationship between the feature point and other feature points. Note that the "identification step" should be performed after the acquisition step (step S100) and the feature point identification step described above, and before the region determination step (step S200) in the processing flow shown in Figure 3.

[0025] Next, we will explain the estimation process (step S300). The estimation process estimates an index related to the user's physique based on numerical values ​​indicating specific regions in the user's side view image and physique index information showing the correlation between the numerical values ​​indicating the specific regions and numerical values ​​related to physique. "Numerical values ​​indicating a specific region" refers to numerical values ​​that express the specific region determined in the region determination process in a predetermined unit and according to a predetermined standard. These include "numerical values ​​that directly indicate a specific region in a user side view image," such as the area of ​​the specific region in the user side view image, the number of pixels within the specific region in the user side view image, and the perimeter of the specific region in the user side view image; "numerical values ​​that show the specific region in a user side view image as a ratio to the entire user side view image," such as the ratio of the area of ​​the specific region to the total area of ​​the user side view image, the ratio of the number of pixels in the specific region to the total number of pixels in the user side view image, the ratio of the perimeter of the specific region to the perimeter of the entire user side view image, and the ratio of the maximum vertical length to the maximum horizontal length of the specific region in the user side view image; and the area of ​​the specific region in the user side view image converted from imaging conditions such as the distance between the user and the imaging device 20. This refers to "numerical values ​​obtained by converting imaging conditions to numerical values ​​of a specific area in a user's side image," such as the area of ​​a specific area in the user's side image, the area of ​​the user's specific area calculated by converting imaging conditions such as the distance between the user and the imaging device 20 to the number of pixels in the specific area in the user's side image, and the perimeter of the user's specific area calculated by converting imaging conditions such as the distance between the user and the imaging device 20 to the perimeter of the specific area in the user's side image. It also refers to "numerical values ​​obtained by converting imaging conditions to numerical values ​​of a specific area in a user's side image," such as the area of ​​the user's specific area calculated by converting the area of ​​the specific area in the user's side image using a predetermined standard value, the area of ​​the user's specific area calculated by converting the number of pixels in the specific area in the user's side image using a predetermined standard value, and the perimeter of the user's specific area calculated by converting the perimeter of the specific area in the user's side image using a predetermined standard value.

[0026] "Physical size index information showing the correlation between numerical values ​​indicating a specific region and indicators related to physical size" refers to information statistically calculated from numerical values ​​indicating a specific region obtained from multiple subjects and indicators related to physical size obtained from multiple subjects, and the form of the information can be anything, such as a table, relational formula, or graph. In this embodiment, as an example, a relational formula (regression formula) obtained from numerical values ​​indicating a specific region obtained from multiple subjects and indicators related to physical size obtained from multiple subjects is used.

[0027] <Physical index estimation method> The estimation method of this embodiment will be explained. Figure 4(a) is a front view of the user, similar to Figure 2(a). Figure 4(b) shows the result of segmenting the user in the user front view image from Figure 4(a) using a known technique, such as Selfie Segmentation, a library for segmenting a predetermined object (e.g., a person) from an image. In the segmentation shown in Figure 4(b), the brightness values ​​of pixels representing the user are converted to white, and the brightness values ​​of pixels representing the background are converted to black. Note that in Figure 4(b), the background (which would normally be "black") is shown with dot hatching for ease of explanation. In Figure 4(b), "m" represents the length corresponding to the user's height on the coordinate system, calculated from the coordinates of the user's head 10 and heel 13 (hereinafter sometimes referred to as "height equivalent length"). The number of pixels per predetermined unit (1 cm in this embodiment) is calculated from the calculated height equivalent length m and the user's actual height. The number of pixels per predetermined unit is used in the processing described later.

[0028] Figure 4(c) shows the result of segmenting the user side image in the same way as in Figure 4(b), where the brightness values ​​of pixels representing the user are converted to white and the brightness values ​​of pixels representing the background are converted to black. In Figure 4(c), the background (which would normally be "black") is shown with dot hatching for ease of explanation. 22 in Figure 4(c) indicates a specific position. In this embodiment, as shown in Figure 4(a), the position 2 / 3 of the way from the shoulder 11 on the imaginary line connecting the user's characteristic point (shoulder 11) and other characteristic points (buttocks 12) is calculated based on the image in Figure 4(a), and this calculated position is designated as the specific position 22.

[0029] <Case 1> Figure 5(a) shows the vertical height position of the shoulder 11 and hip 12 relative to the user's side view image, where "n" represents the vertical distance between the shoulder 11 and hip 12. In this embodiment, unless otherwise specified, "vertical direction" and "height direction" refer to the same direction. Figure 5(b) shows a specific region, where "p1" indicates the distance between a specific position 22 and a position 50 pixels vertically upward (towards the top of the head) and a position 50 pixels downward (towards the heel).

[0030] Figure 5(c) shows the segmented result compared to Figure 5(b). In the figure, "p1" indicates the distance (distance in the height direction) between a specific position 22 and a position 50 pixels vertically upward (towards the top of the head) and a position 50 pixels downward (towards the heel). In Case 1, the area labeled "S1" in Figure 5(c) is extracted as a "specific area." Thus, the "specific area" is an area extracted from the user's side image that includes a specific position within a predetermined range in the direction of the user's height (in Case 1, the "S1" area within the range of "p1"). Then, the number of pixels in the S1 region is extracted, and the object quantity in the S1 region is calculated based on the user's height equivalent length m calculated from the user's coordinates (the coordinates of the user's head 10 and heel 13) and the user's actual height, which is calculated from the number of pixels per predetermined unit on the captured image.

[0031] Figure 6 shows a graph and regression equation illustrating the correlation between the object volume in the S1 region shown in Figure 5(c) for 31 subjects (ages 25-62) and the BMI (Body Mass Index) calculated from the measured height and weight of the 31 subjects. In other words, the object quantity in the S1 region corresponds to a "numerical value indicating a specific region," BMI corresponds to an "indicator related to body size," and the graph and regression equation correspond to "body size index information." The regression equation shown in Figure 6 uses BMI as the dependent variable and the object volume in the S1 region and the subject's actual height as independent variables. The coefficient of determination (R) in this case is 2 The value is 0.91, indicating a high correlation between the number of objects in the S1 region and the BMI.

[0032] <Case 2> Figure 7(a) is similar to Figure 5(a), showing the vertical height position of the shoulder 11 and hip 12 relative to the user's side view image, where "n" in the figure indicates the vertical distance between the shoulder 11 and hip 12. Figure 7(b) shows a specific region, where "p2" indicates the distance from a specific position 22 to a position 50 pixels vertically upward (towards the top of the head) and 100 pixels vertically downward (towards the heel). Figure 7(c) shows the segmented result compared to Figure 7(b). In the figure, "p2" indicates the distance from a specific position 22 to a position 50 pixels upward (towards the top of the head) and 100 pixels downward (towards the heel), similar to Figure 7(b). In Case 2, the "S2" region shown in Figure 7(c) is extracted as a "specific region." Thus, a "specific region" is a region extracted from the user's side image that includes a specific position and is within a predetermined range in the direction of the user's height (in Case 2, the "S2" region within the "p2" range). Furthermore, the specific region (S2) in Case 2 includes the area around the buttocks (the entire buttocks) compared to the specific region in Case 1 (compared to the S1 region in Figure 5(c)). Then, the number of pixels in the S2 region is extracted, and the object quantity in the S2 region is calculated based on the user's height equivalent length m calculated from the user's coordinates (the coordinates of the user's head 10 and heel 13) and the user's actual height, which is calculated from the number of pixels per predetermined unit on the captured image.

[0033] Figure 8 shows a graph and regression equation illustrating the correlation between the object volume in the S2 region shown in Figure 7(c) for 31 subjects similar to those in Case 1, and the BMI calculated from the measured height and weight of the 31 subjects. In other words, the number of objects in the S2 domain corresponds to a "numerical value indicating a specific domain," BMI corresponds to an "indicator related to body size," and the graph and regression equation correspond to "body size index information." The regression equation shown in Figure 8 uses BMI as the dependent variable and the object volume in the S2 region and the subject's actual height as independent variables. The coefficient of determination (R) in this case is 2 The coefficient of determination (R) is 0.92, indicating a high correlation between the number of objects in the S2 region and BMI. Furthermore, the coefficient of determination (R) is higher in Case 2 than in Case 1. 2 Because the value of ) is high, the S2 region can be said to be even more suitable as a specific region than the S1 region.

[0034] <Case 3> Under the same conditions as in Case 2, if the dependent variable is BMI and the independent variables are the object quantity in the S2 region and the height equivalent length m on the subject's coordinate system, the coefficient of determination (R 2 The value was 0.93. Therefore, when using the user's height as an explanatory variable, using the height equivalent length m obtained from the coordinates of the top of the head 10 and the heel 13 allows for a more accurate estimation than using the actual height value.

[0035] <Case 4> Figure 9(a) shows the segmented result of the user's front view image in Figure 7. In the figure, "p2" indicates the distance from a specific position 22 to a position 50 pixels upward (towards the top of the head) and 100 pixels downward (towards the heel), similar to Figure 7(b). In Case 4, the region labeled "S3" in Figure 9(a) is defined as the "specific region." While the specific region is typically a specific area of ​​the user's body in a side view image, in Case 4, based on the user's specific position, the specific region of the user's body in a front view image is defined as the "specific region." The number of pixels in the S3 region is extracted, and the object quantity in the S3 region is calculated based on the number of pixels per predetermined unit on the captured image, which is calculated from the user's height equivalent length m calculated from the coordinates of the user's head 10 and heel 13, and the user's actual height.

[0036] Figure 9(b) shows a graph and regression equation illustrating the correlation between the object volume in the S3 region shown in Figure 9(a) for 31 subjects similar to those in Case 1, and the BMI calculated from the measured height and weight of the 31 subjects. The regression equation shown in Figure 9(b) uses BMI as the dependent variable and the object volume in the S3 region and the subject's actual height as independent variables. The coefficient of determination (R) in this case is 2 The coefficient of determination (R) is 0.56, indicating a correlation between the amount of objects in the S3 region and BMI. However, compared to cases using specific regions of the user's body in the user's lateral image, as in Cases 1 to 3, the coefficient of determination (R) is lower. 2 Since the value of ) is remarkably low, it can be seen that using specific regions of the user's body in the user's side view image to estimate body size-related indicators is effective in improving estimation accuracy.

[0037] <Case 5> Figure 10 shows the results for Case 5, in which the "S2" region shown in Figure 7(c) is designated as the "specific region" and body weight is designated as the "indicator related to body size." The number of pixels in the S2 region is extracted, and the object quantity in the S2 region is calculated based on the number of pixels per predetermined unit on the captured image, which is calculated from the user's height equivalent length m calculated from the coordinates of the user's head 10 and heel 13, and the user's actual height. A graph and regression equation showing the correlation between the amount of objects in the S2 region of 31 subjects (similar to Case 2) and the measured weight of the 31 subjects are presented. The regression equation shown in Figure 10 uses body weight as the dependent variable and the object quantity in the S2 region and the subject's actual height as independent variables. The coefficient of determination (R) in this case is 2 The value is 0.90, indicating a high correlation between the amount of objects in the S2 domain and body weight.

[0038] Cases 6 to 11, in which the distance to the specific position 22 is varied, are described below. Cases 6 to 11 are in which "p1" in Figure 5(c) is replaced from "p6" to "p11", and the region of "S1" is replaced from "S6" to "S11".

[0039] <Case 6> "p6" is defined as the distance between a specific position 22 in the user's side view image, 60 pixels vertically upward (towards the top of the head), and 100 pixels vertically downward (towards the heel). In Case 6, for the user's side view image, the range of "p6" in the direction of the user's height, including the specific position 22, is defined as "S6" and designated as the specific region. Then, the number of pixels in the S6 region is extracted, and the object quantity in the S6 region is calculated based on the user's height equivalent length m calculated from the user's coordinates (the coordinates of the user's head 10 and heel 13) and the user's actual height, which is calculated from the number of pixels per predetermined unit on the captured image. Figure 11 shows a graph and regression equation illustrating the correlation between the object volume in the S6 region of 31 subjects, similar to Case 1, and the BMI calculated from the measured height and weight of the 31 subjects. The regression equation shown in Figure 11 uses BMI as the dependent variable and the object volume in the S6 region and the subject's actual height as independent variables. The coefficient of determination (R) in this case is 2 The ratio is 0.92, indicating a high correlation between the number of objects in the S6 domain and the BMI.

[0040] <Case 7> "p7" is defined as the distance between a specific position 22 in the user's side view image, 70 pixels vertically upward (towards the top of the head), and 100 pixels vertically downward (towards the heel). In Case 7, for the user's side image, the range of "p7" in the user's height direction including the specific position 22 is defined as "S7" and set as the specific region. Then, the number of pixels in the S7 region is extracted, and based on the pixel number per predetermined unit on the imaging image calculated from the length m corresponding to the user's height on the user's coordinates calculated from the coordinates of the user's head top 10 and the coordinates of the heel 13 and the user's actual height, the object quantity in the S7 region is calculated. Fig. 12 shows a graph and a regression equation indicating the correlation between the object quantity in the S7 region of 31 examinees similar to those in Case 1 and the BMI calculated from the actually measured height and actually measured weight of the 31 examinees. The regression equation shown in Fig. 12 has the BMI as the objective variable and uses the object quantity in the S7 region and the examinee's actual height as explanatory variables. The coefficient of determination (R 2 ) at this time is 0.93, indicating that there is a high correlation between the object quantity in the S7 region and the BMI.

[0041] <Case 8> "p8" is the distance between the position 70 pixels upward (toward the head top) and the position 120 pixels downward (toward the heel) in the vertical direction with respect to the specific position 22 in the user's side image. In Case 8, for the user's side image, the range of "p8" in the user's height direction including the specific position 22 is defined as "S8" and set as the specific region. Then, the number of pixels in the S8 region is extracted, and based on the pixel number per predetermined unit on the imaging image calculated from the length m corresponding to the user's height on the user's coordinates calculated from the coordinates of the user's head top 10 and the coordinates of the heel 13 and the user's actual height, the object quantity in the S8 region is calculated. Fig. 13 shows a graph and a regression equation indicating the correlation between the object quantity in the S8 region of 31 examinees similar to those in Case 1 and the BMI calculated from the actually measured height and actually measured weight of the 31 examinees. The regression equation shown in Fig. 13 has the BMI as the objective variable and uses the object quantity in the S8 region and the examinee's actual height as explanatory variables. The coefficient of determination (R 2 ) at this time is 0.93, indicating that there is a high correlation between the object quantity in the S8 region and the BMI.

[0042] <Case 9> "p9" is defined as the distance between a specific position 22 in the user's side view image, 70 pixels vertically upward (towards the top of the head), and 130 pixels vertically downward (towards the heel). In Case 9, for the user's side view image, the range of "p9" in the direction of the user's height, including the specific position 22, is defined as "S9" and designated as the specific region. Then, the number of pixels in the S9 region is extracted, and the object quantity in the S9 region is calculated based on the user's height equivalent length m calculated from the user's coordinates (the coordinates of the user's head 10 and heel 13) and the user's actual height, which is calculated from the number of pixels per predetermined unit on the captured image. Figure 14 shows a graph and regression equation illustrating the correlation between the object volume in the S9 region of 31 subjects, similar to those in Case 1, and the BMI calculated from the measured height and weight of the 31 subjects. The regression equation shown in Figure 14 uses BMI as the dependent variable and the object volume in the S9 region and the subject's actual height as independent variables. The coefficient of determination (R) in this case is 2 The ratio is 0.92, indicating a high correlation between the number of objects in the S9 domain and the BMI.

[0043] <Case 10> "p10" is defined as the distance between a specific position 22 in the user's side view image, 70 pixels upward (towards the top of the head), and 140 pixels downward (towards the heel). In Case 10, for the user's side view image, the range of "p10" in the direction of the user's height, including the specific position 22, is defined as "S10" and is considered a specific region. Then, the number of pixels in the S10 region is extracted, and the object quantity in the S10 region is calculated based on the user's height equivalent length m calculated from the user's coordinates (the coordinates of the user's head 10 and heel 13) and the user's actual height, which is calculated from the number of pixels per predetermined unit on the captured image. Figure 15 shows a graph and regression equation illustrating the correlation between the object volume in the S10 region of 31 subjects, similar to those in Case 1, and the BMI calculated from the measured height and weight of the 31 subjects. The regression equation shown in Figure 15 uses BMI as the dependent variable and the object volume in the S10 region and the subject's actual height as independent variables. The coefficient of determination (R) in this case is 2 The ratio is 0.92, indicating a high correlation between the number of objects in the S10 region and BMI.

[0044] <Case 11> "p11" is defined as the distance between a specific position 22 in the user's side view image, 80 pixels vertically upward (towards the top of the head), and 100 pixels vertically downward (towards the heel). In Case 11, for the user's side view image, the range of "p11" in the direction of the user's height, including the specific position 22, is defined as "S11" and is considered a specific region. Then, the number of pixels in the S11 region is extracted, and the object quantity in the S11 region is calculated based on the user's height equivalent length m calculated from the user's coordinates (the coordinates of the user's head 10 and heel 13) and the user's actual height, which is calculated from the number of pixels per predetermined unit on the captured image. Figure 16 shows a graph and regression equation illustrating the correlation between the object volume in the S11 region of 31 subjects, similar to Case 1, and the BMI calculated from the measured height and weight of the 31 subjects. The regression equation shown in Figure 16 uses BMI as the dependent variable and the object volume in the S11 region and the subject's actual height as independent variables. The coefficient of determination (R) in this case is 2 The value is 0.93, indicating a high correlation between the number of objects in the S11 region and BMI.

[0045] From Cases 1 through 11, the following can be concluded. 1) It is possible to estimate BMI with high accuracy as an indicator related to body size from numerical values ​​indicating specific regions in the user's lateral image. 2) It is possible to estimate weight with high accuracy as an indicator related to body size from numerical values ​​indicating specific areas in the user's lateral image. 3) When using numerical values ​​indicating a specific range and numerical values ​​related to height as explanatory variables in a regression equation with body mass index (BMI) as the dependent variable, it is possible to estimate with higher accuracy by using the head top 10, heel 13, and the height equivalent length m on the coordinate system obtained from their respective coordinates. 4) The specific region can be estimated with high accuracy by setting it to a position 50 to 80 pixels vertically upward (towards the top of the head) and 50 to 140 pixels vertically downward (towards the heel) relative to the specific position 22 in the user's side view image. In particular, estimation with high accuracy is possible by setting it to a position 100 to 120 pixels vertically downward (towards the heel). This is thought to be because body fat and visceral fat often accumulate at positions 50 to 80 pixels vertically upward (towards the top of the head) and 50 to 140 pixels vertically downward (towards the heel) relative to the specific position 22, and these influence indicators related to body size. 5) Indicators related to body size cannot be estimated with high accuracy from numerical values ​​indicating specific areas in a user's frontal image. This is thought to be because body fat and visceral fat often accumulate in the anterior-posterior direction (front / back) of specific areas, and these influence indicators related to body size, but this point is difficult to grasp from a user's frontal image.

[0046] Furthermore, the "predetermined range" from which a specific region is extracted from Case 1 to Case 11 is preferably as follows. 1) The predetermined range is a range of different lengths in the upward direction (towards the top of the head) and the downward direction (towards the heel) relative to the user's height with respect to a specific position 22. 2) Set a predetermined range such that the length in the downward direction is longer than the length in the upward direction. In this method, since the specific location is identified as being near the abdomen (near the navel), and the areas where visceral fat and body fat tend to accumulate are near the specific location and below it, setting the range described in 1) and 2) above allows for the estimation of body size indicators with high accuracy. Furthermore, areas above the specific location, where bones such as ribs are present, tend to accumulate less visceral fat and body fat compared to the abdomen, thus having less impact on body size indicators. Thus, the inventors of this application have studied the ranges that particularly affect body size indicators from various perspectives and have found that setting the range described in 1) and 2) above allows for the estimation of body size indicators with high accuracy. 3) While it is preferable for the downward length to include the entire buttocks, if the downward length exceeds a predetermined value, the estimation accuracy tends to decrease, albeit slightly. In item 2) above, it was stated that "a predetermined range is set so that the length in the downward direction is longer than the length in the upward direction," but for example, extending the range all the way to the heel is undesirable. If the length in the downward direction is made longer than necessary, it will include a large range that does not have much effect on the indicators related to body size, and as a result, the estimation accuracy is thought to be low. 4) For a specific position, the ratio of the upper length to the lower length is preferably 10:10 or 10:30, and more preferably 10:20 to 10:25. As mentioned above, this range is derived from research results in a way that reliably includes the parts that affect the indicators related to physique, and does not include a large portion of the range that has little effect on the indicators related to physique, thus enabling highly accurate estimation.

[0047] <Body Mass Indicator Estimation System> Figure 17 shows a conceptual diagram of the body size index estimation system 400 (hereinafter sometimes referred to as "this system"). This system is a control device system and consists of an acquisition unit 310, a region determination unit 320, and an estimation unit 330. It also includes an information processing terminal 300 capable of performing various processes, and the acquisition unit 310, region determination unit 320, and estimation unit 330 are provided on this information processing terminal 300. The information processing terminal 300 includes input devices such as a keyboard and pointing device, an arithmetic processing unit, a storage unit 350, etc. The storage unit 350 stores data and programs necessary for controlling this system, as well as data and programs necessary for performing the body size index estimation method described above on this system. It also stores various libraries used in the region determination process (region determination processing) and body size index information used in the estimation process (estimation processing). Note that not all of this data needs to be stored in the storage unit 350; for example, a server (not shown) may be provided outside the information processing terminal 300, and the data may be stored on the server and accessed via a network. Furthermore, while it is preferable that the information processing terminal 300 is equipped with a display device (display unit 360), the display device may be provided outside the information processing terminal and connected via a network. In addition, the body size index estimation system 400 may be equipped with a built-in or external imaging device 20.

[0048] The acquisition unit 310 is a means for acquiring a user side image, which is an image of the user's side, and corresponds to an acquisition means. The region determination unit 320 is a means for determining a specific region of the user's body in a user side view image based on the user's specific position in the user side view image, and corresponds to a region determination means. The estimation unit 330 is a means for estimating an index related to the user's physique, and corresponds to an estimation means. The estimation unit 330 estimates an index related to the user's physique based on a numerical value indicating a specific region in the user's side view image and physique index information showing the correlation between the numerical value indicating the specific region and the index related to physique. This system is capable of executing the processing steps of the body mass index estimation method described above.

[0049] <Physical index estimation device> Figure 18 shows a conceptual diagram of the body size index estimation device 500 (hereinafter sometimes referred to as "this device"). This device is equipped with a control processing unit (control processing unit), which includes an acquisition unit 510, a region determination unit 520, and an estimation unit 530. The body size index estimation device 500 includes input devices such as a keyboard and a pointing device, a control processing unit, a storage unit 550, etc. The storage unit 550 stores data and programs necessary for controlling this system, as well as data and programs necessary for executing the body size index estimation method described above in this system. It also stores various libraries used in the region determination process (region determination processing) and body size index information used in the estimation process (estimation processing). Note that not all of this data needs to be stored in the storage unit 550; for example, a server (not shown) may be provided outside the body size index estimation device 500, and the data may be stored on the server and accessed via a network. Furthermore, it is preferable that the body size index estimation device 500 is equipped with a display device (display unit 560), but the display device may be provided outside the body size index estimation device 500 and connected via a network. In addition, the body size index estimation device 500 may be equipped with a built-in or external imaging device 20. The acquisition unit 510 is a means for acquiring a user side image, which is an image of the user's side, and corresponds to an acquisition means. The region determination unit 520 is a means for determining a specific region of the user's body in a user side view image based on the user's specific position in the user side view image, and corresponds to a region determination means. The estimation unit 530 is a means for estimating an index related to the user's physique, and corresponds to an estimation means. The estimation unit 530 estimates an index related to the user's physique based on a numerical value indicating a specific region in the user's side view image and physique index information showing the correlation between the numerical value indicating the specific region and the index related to physique. This system is capable of executing the processing steps of the body mass index estimation method described above.

[0050] <Body Mass Indicator Estimation Program> The information processing terminal 300 has a program (hereinafter referred to as "this program") installed on it that causes the information processing terminal 300 to execute the above-described body size index estimation method. The control processing unit of the body size index estimation device 500 also has this program installed on it that causes the body size index estimation method described above to be executed by the body size index estimation device 500. In this embodiment, the program is application software that estimates an index related to the user's body size based on a numerical value indicating a specific region in the user's side view image and body size index information showing the correlation between the numerical value indicating the specific region and the index related to body size. This program includes an acquisition step, a region determination step, and an estimation step. The acquisition step is the step (process) of acquiring a user side image, which is an image taken of the user's side. The region determination step is a step (process) in which a specific region of the user's body in a user's side view image is determined based on the user's specific position in the acquired user's side view image. The estimation step is a step (process) of estimating an index related to the user's physique based on a numerical value indicating a specific area in the user's side view image and physique index information showing the correlation between the numerical value indicating the specific area and the aforementioned physique-related index. Furthermore, it is preferable that this program further includes a feature point identification step and a identification step. The feature point identification step is a step (process) that identifies the locations of multiple feature points on the user's body in a frontal image of the user. The identification step is a step (process) of identifying a specific location in the user's side view image based on the positional relationship between a feature point and other such feature points.

[0051] As described above, the present invention has been explained with reference to specific embodiments, but the present invention is not limited to the embodiments described above, and includes various modifications, improvements, and other forms as long as the objectives of the present invention are achieved. <Variation> As described above, in this embodiment, MediaPipe Pose was used as a known skeletal point information acquisition technique when extracting human skeletal points from an image, but the invention is not limited to this, and other techniques such as VISION POSE® and OPEN POSE are also used.

[0052] In this embodiment, the user side image or user front image is a multi-frame moving image, but it may also be a multi-frame still image. In this case, it is more preferable that the frames include various orientations between the user and the imaging device 20.

[0053] In this embodiment, the number of pixels per predetermined unit was calculated using the user's actual height. Thus, the estimation process may further include a physical feature acquisition step to acquire physical features other than indicators related to the user's physique, and the estimation process may use physical features to estimate indicators related to the user's physique. Note that the physical feature acquisition step may be performed before the estimation process (step S300) shown in Figure 3. Furthermore, it is not necessary to use the actual height when calculating the number of pixels per predetermined unit. For example, the user's standing position during imaging may be set at a predetermined distance from the imaging device 20, and the user may stand at that position while imaging is performed. In this way, the number of pixels per predetermined unit can be calculated. Alternatively, for example, the user may be imaged by the imaging device 20 along with a predetermined scale. In this way, the number of pixels per predetermined unit can be calculated. Furthermore, considering the degree of freedom in the imaging environment, it is preferable to acquire physical characteristics.

[0054] In this embodiment, height was obtained as a "physical characteristic," but weight, age, and gender may also be obtained. Since there are differences in the distribution of visceral fat and body fat between men and women, this may be taken into consideration when determining a specific region. In this case, the physical characteristic acquisition process can be performed before the region determination process (step S200) shown in Figure 3.

[0055] In this embodiment, the body size index information used is a relational expression (regression equation) obtained from numerical values ​​indicating a specific region acquired from multiple subjects and body size-related indices acquired from multiple subjects, but it is not limited to this. Any information that shows a statistically determined correlation between numerical values ​​indicating a specific region determined from a specific position of the subject in a lateral image taken from the sides of multiple subjects and body size-related indices of multiple subjects is acceptable. For example, it may be information that shows the correlation between user lateral images, analyzed using machine learning, and body size-related indices statistically determined based on the analysis results.

[0056] Although this embodiment does not mention the user's clothing during imaging, thin T-shirts, sweatshirts, etc., are preferred. For example, it is preferable to avoid wearing bulky clothing such as a down jacket during imaging, as this may reduce the estimation accuracy.

[0057] In this embodiment, the height equivalent length (m) on the captured image was calculated from the skeletal point coordinates of the top of the head 10 and the heel 13, but this is not limited to this. For example, the top of the head may be calculated using the coordinates of the top of the head portion of the segmented data, and the skeletal point coordinates of the heel 13.

[0058] In this embodiment, multiple feature points of the user's body identified using a frontal image of the user were used to determine a specific location, but this is not limited to this. For example, when segmenting a side view of the user, the user's height equivalent length on the coordinate system may be calculated from the coordinates of the boundary between the user and the background. Based on this height equivalent length, the height of the user's center of gravity in the direction of their height may be determined, and this height of the center of gravity may be used as the specific location. In this way, a specific location can be determined using only a side view of the user, and a body size index can be estimated.

[0059] In this embodiment, the same body mass index information was used for estimation regardless of the user, but this is not limited to this. For example, there are differences in how visceral fat and body fat are distributed between elementary and junior high school students and adults. Therefore, body mass index information may be created from data from multiple subjects consisting of elementary and junior high school students, and the body mass index information to be used for estimation may be selected according to the user. Doing so makes it possible to improve the estimation accuracy. In addition to separating body mass index information into elementary and junior high school students and adults, it may also be separated into groups where characteristics of visceral fat and body fat distribution that affect body mass index are likely to be apparent, such as females and males.

[0060] In this embodiment, the specific position 22 is set to a position 2 / 3 of the way from the shoulder 11 to the hip 12 in the vertical distance (on the imaginary line 21) from the user's shoulder 11 to the hip 12, but it is not limited to this. The specific position 22 can be appropriately changed between 3 / 5 and 4 / 5 of the way from the shoulder 11 with respect to the imaginary line 21 connecting the shoulder 11 and the hip 12, and it is preferable to set it to 2 / 3 and 3 / 4 of the way from the shoulder 11. Furthermore, in this embodiment, the shoulder 11 and hip 12 were used as feature points when identifying the specific position 22, but this is not limited to this. For example, one feature point may be the shoulder 11, and another feature point may be the knee 15. Alternatively, the position may be between 2 / 5 and 9 / 25 of the distance from the shoulder 11 relative to a hypothetical line (not shown) connecting the shoulder 11 and the knee 15. Since the position identified in this way is the same as the position identified from the shoulder 11 and the hip 12, either can be used. Alternatively, the specific location may be identified by the average value of both the positional relationship between shoulder 11 (right shoulder) and hip 12 (right hip), and the positional relationship between shoulder 11 (left shoulder) and hip 12 (left hip). In other words, the specific location may be identified by the average value of the positional relationship between feature points present on the right half of the body and other feature points, and the positional relationship between feature points present on the left half of the body and other feature points. Due to lifestyle habits and other factors, many users have different skeletal positions (heights) between their right and left halves. Therefore, the estimation accuracy can be improved by identifying the specific location using the average value of the results for the right and left halves of the body. Alternatively, a specific location may be identified, for example, by the positional relationship between shoulder 11 (right shoulder) and hip 12 (left hip), and the positional relationship between shoulder 11 (left shoulder), hip 12 (right hip), and hip 12 (left hip). A specific location may also be identified by the intersection of the positional relationship between shoulder 11 (right shoulder) and hip 12 (left hip) (an imaginary line connecting shoulder 11 (right shoulder) and hip 12 (left hip)) and the positional relationship between shoulder 11 (left shoulder), hip 12 (right hip), and hip 12 (left hip) (an imaginary line connecting shoulder 11 (left shoulder), hip 12 (right hip), and hip 12 (left hip)). In other words, a specific location may be identified based on the positional relationship between a feature point on the right half of the body and other feature points on the left half of the body.

[0061] Furthermore, the present invention also applies to physique index estimation devices and physique index estimation programs, with preferred examples being the same as described above. The present invention is also applicable when an information processing program realizing the functions of the embodiments is supplied to a system or device and executed by a built-in processor. Therefore, the technical scope of the present invention includes programs installed on a computer, media containing such programs, WWW (World Wide Web) servers that allow the program to be downloaded, and processors that execute such programs, all necessary for realizing the functions of the present invention on a computer. In particular, at least a non-transitory computer-readable medium containing a program that causes a computer to execute the processing steps included in the embodiments described above is included within the technical scope of the present invention. [Explanation of symbols]

[0062] 10 top of head 11 Shoulder 12 Butt 13 Heel 14 Elbow 15 Knee 20 Imaging device 21 virtual lines 22 Specific location 300 Information Processing Terminals 310 Acquisition Department 320 Area determination section 330 Estimation Department 350 Storage section 360 display 400 Body Size Indicator Estimation System 500 Physique index estimation device 510 Acquisition Department 520 Area determination section 530 Estimation part 550 Storage section 560 Display section

Claims

1. The acquisition process involves obtaining a user side image, which is an image taken of the user's side. A region determination step in which a specific region of the user's body in the user's side view image is determined based on the specific position of the user in the user's side view image, This includes an estimation step of estimating an indicator related to the user's physique, The estimation process described above is: A numerical value indicating the specific region in the user side view image, A method for estimating a body size index, characterized by estimating the body size index of a user based on body size index information that shows the correlation between a numerical value indicating the specific region and the body size index.

2. The acquisition step further acquires a user front image captured from the front of the user, The process further includes a feature point identification step of identifying the positions of multiple feature points of the user's body in the user's frontal image, The method for estimating a body size index according to claim 1, further comprising a step of identifying the specific position in the user side image based on the positional relationship between the feature point and other feature points.

3. The method for estimating body size indices according to claim 2, wherein the aforementioned feature point corresponds to the user's shoulder joint, and the other aforementioned feature point corresponds to the user's hip joint.

4. The method for estimating body size indices according to any one of claims 1 to 3, wherein the specified region is a region obtained by cutting out a predetermined range in the direction of the user's height that includes the specified position from the user's side view image.

5. The method for estimating a body size index according to claim 4, wherein the predetermined range is a range of different lengths in the upward and downward directions of the user's height relative to the specific position.

6. The method for estimating a body size index according to claim 5, wherein the downward length is longer than the upward length.

7. The method for estimating a body mass index according to any one of claims 1 to 3, wherein the index related to body mass is BMI (Body Mass Index).

8. A body size index estimation system that estimates an index related to the user's body size using a control device, A means for acquiring a user side image, which is an image of the user's side, A region determination means for determining a specific region of the user's body in the user's side view image based on the user's specific position in the user's side view image, Includes an estimation means for estimating an indicator related to the user's physique, The estimation means is, A numerical value indicating the specific region in the user side view image, A body size index estimation system characterized by estimating the body size index of a user based on body size index information that shows the correlation between a numerical value indicating the specific region and the body size index.

9. A means for acquiring a user side image, which is an image of the user's side, A region determination means for determining a specific region of the user's body in the user's side view image based on the user's specific position in the user's side view image, The system includes estimation means for estimating indicators related to the user's physique, The estimation means is, A numerical value indicating the specific region in the user side view image, A body size index estimation device characterized by estimating the user's body size index based on body size index information showing the correlation between a numerical value indicating the specific region and the body size index.

10. A body size index estimation program for estimating indicators related to a user's body size, The acquisition step involves obtaining a user side image, which is an image taken of the user's side. A region determination step in which a specific region of the user's body in the user's side view image is determined based on the specific position of the user in the user's side view image, An estimation step for estimating indicators related to the user's physique, The estimation step described above is: A numerical value indicating the specific region in the user side view image, A body size index estimation program that estimates the body size index of the user based on body size index information showing the correlation between a numerical value indicating the specific region and the body size index, and performs the above steps.

Images