Dimensional data calculation device, dimensional data calculation method, program, product manufacturing system, and terminal device

Abstract

To provide highly accurate dimension data for each part of an object.SOLUTION: A processing unit 2524 of a dimension data calculation device 2520 includes a first calculation section 2524D1, a second calculation section 2524D2, and a determination section 2524D3. The first calculation section 2524D1 calculates a first calculated value of dimension data of a plurality of parts including a specific part in an object by performing polynomial regression using a coefficient learned by machine learning, on the attribute data including at least the total length data of the object. The second calculation section 2524D2 calculates a second calculated value of dimension data of the specific part in the object from the image data of the object. The determination section 2524D3 determines the dimension data of each part in the object based on the first and second calculated values.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present disclosure relates to a dimension data calculation device, a dimension data calculation method, a program, a product manufacturing system, and a terminal device. [Background technology]

[0002] Conventionally, devices for calculating the size of an object have been studied. For example, Patent Document 1 (JP 2011-227692 A) discloses a technology for measuring the size of a specific part of a designated person based on a subject image in which the designated person's whole body image appears and the height input by an input operation. Summary of the Invention [Problem to be solved by the invention]

[0003] However, conventional techniques have not always been able to calculate each part of an object with high accuracy. [Means for solving the problem]

[0004] A dimensional data calculation device according to a first aspect determines dimensional data for each part of an object based on a first calculated value of dimensional data obtained by polynomial regression of attribute data and a second calculated value of dimensional data for a specific part obtained from image data. This makes it possible to obtain the dimensional data for each part of the object as a whole with high accuracy. Then, by using such dimensional data, it is possible to manufacture a product that fits the shape of the object. The dimension data calculation device of the second aspect is the dimension data calculation device of the first aspect, which can calculate with high accuracy dimension data of each part of a person (hereinafter also referred to as the subject), which is an object. Then, by using such dimension data, it is possible to manufacture a product that fits the shape of the subject. The dimension data calculation device of the third aspect is the dimension data calculation device of the second aspect, and receives image data including: a front image of an object in which the arms are spread apart so as not to overlap the torso in a front view and are positioned so as to fit within the width of the torso in a side view; a side image of the object in which the arms are spread apart so as not to overlap the torso in a front view and are positioned so as to fit within the width of the torso in a side view; and a side image of the object in which the arms are positioned higher than the waist in both the front and side views. Using such image data, dimension data of specific parts of the object can be determined with high accuracy. Furthermore, using such dimension data, products that are more closely matched to the shape of the object can be manufactured. A fourth aspect of the dimension data calculation device is the dimension data calculation device of any one of the first to third aspects, wherein if the difference between the first calculated value and the second calculated value is within a predetermined value, the dimension data of the specific part is set to the second calculated value. As a result, for the specific part where individual differences in shape are likely to occur, dimension data based on image data is used, making it possible to determine the dimension data of each part of the object as a whole with high accuracy. A fifth aspect of the dimension data calculation device is the dimension data calculation device of any one of the first to fourth aspects, wherein when the difference between the first calculated value and the second calculated value exceeds a predetermined value, the dimension data of the specific portion is set to the first calculated value, thereby maintaining the reliability of the dimension data as a whole even when irregular image data is acquired. The dimension data calculation device of the sixth aspect is the dimension data calculation device of the first to fifth aspects, in which the second calculated value of the dimension data of the specific part of the object is calculated based on the perimeter of the virtual figure, so that the second calculated value can be easily calculated. A seventh aspect of the dimension data calculation device is the dimension data calculation device of the sixth aspect, which calculates a second calculated value of dimension data of a specific part of an object using a learning device that has been trained to output the perimeter of a specific part of a predetermined object in response to an input of the perimeter of a predetermined virtual figure. This makes it possible to easily calculate the second calculated value of dimension data of a specific part of a complex shape using a virtual figure whose perimeter can be easily calculated. The dimension data calculation device of the eighth aspect uses image data including a depth map in the dimension data calculation devices of the first to seventh aspects, and therefore can calculate the second calculated value from a single image data. The dimension data calculation device of the ninth aspect is a dimension data calculation device of the first to eighth aspects, which generates a silhouette image of the object from image data and calculates the second calculated value from the silhouette image.Therefore, for specific parts where individual differences in shape are likely to occur, dimension data based on the silhouette image is used, making it possible to determine the dimension data of each part of the object as a whole with high accuracy. The dimension data calculation device of the tenth aspect is a dimension data calculation device of the first to ninth aspects, which accepts image data of an object transmitted from a terminal device, and if the difference between the first calculated value and the second calculated value exceeds a predetermined value, requests the terminal device to resend the image data, thereby enabling highly reliable dimension data to be calculated. The dimensional data calculation method of an eleventh aspect determines dimensional data for each part of an object based on a first calculated value of dimensional data obtained from attribute data and a second calculated value of dimensional data for a specific part obtained from image data. This makes it possible to calculate the dimensional data for each part of the object with high accuracy as a whole. Then, by using such dimensional data, it is possible to manufacture a product that fits the shape of the object. A program according to a twelfth aspect determines dimensional data for each part of an object based on a first calculated value of dimensional data obtained from attribute data and a second calculated value of dimensional data for a specific part obtained from image data. By having a computer execute such a program, the dimensional data for each part of the object can be calculated with high accuracy as a whole. As a result, a product that fits the shape of the object can be manufactured. A thirteenth aspect of the present invention provides a product manufacturing system that determines dimensional data for each part of an object based on first calculated values ​​of dimensional data obtained from attribute data and second calculated values ​​of dimensional data for a specific part obtained from image data, and a product manufacturing device that uses the determined dimensional data to manufacture a product related to the shape of the object. Thus, a product manufacturing system that manufactures a product that fits the shape of the object can be provided. In a fourteenth aspect, when the target object is a person, the terminal device outputs guide information prompting the capture of a front image of the target object with the arms spread apart so as not to overlap the torso in a front view and positioned so as to fit within the width of the torso in a side view, a side image of the target object with the arms spread apart so as not to overlap the torso in a front view and positioned so as to fit within the width of the torso in a side view, and a side image of the target object with the arms positioned higher than the waist in both a front view and a side view. By using such guide information, image data that enables highly accurate determination of dimensional data of specific parts of the target object can be transmitted to the dimension data calculation device. As a result, the dimension data calculation device can accurately determine dimensional data of each part of the target object. [Brief explanation of the drawings]

[0005] [Figure 1] FIG. 2 is a schematic diagram showing the configuration of a dimension data calculation device 2520 according to one embodiment of the present invention. [Figure 2] FIG. 10 is a schematic diagram showing the configuration of a calculation unit 2524D according to the same embodiment. [Figure 3] FIG. 10 is a schematic diagram showing an example of shape data of the bust, waist, and hips on the side according to the embodiment. [Figure 4] FIG. 10 is a schematic diagram showing an example of shape data of the bust, waist, and hips on the front side according to the embodiment. [Figure 5] FIG. 10 is a schematic diagram showing an example of shape data of the bust, waist, and hips on the side according to the embodiment. [Figure 6] 10 is a flowchart for explaining the operation of the dimension data calculation device 2520 according to the same embodiment. [Figure 7] FIG. 25 is a schematic diagram showing the concept of a product manufacturing system 2501. [Figure 8] FIG. 25 is a schematic diagram showing the configuration of a terminal device 2510. [Figure 9] FIG. 10 is a schematic diagram showing the concept of guide information GI. [Figure 10] FIG. 2 is a sequence diagram for explaining the operation of the product manufacturing system 2501. DETAILED DESCRIPTION OF THE INVENTION

[0006] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings, the same components are designated by the same reference numerals, and redundant description will be omitted.

[0007] (1) Dimensional data calculation device 1 is a schematic diagram showing the configuration of a dimension data calculation device 2520 according to one embodiment of the present invention. The dimension data calculation device 2520 can be realized by an arbitrary computer, and includes a storage unit 2521, an input / output unit 2522, a communication unit 2523, and a processing unit 2524. Note that the dimension data calculation device 2520 may be realized as hardware using an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or the like.

[0008] The storage unit 2521 stores various types of information and is realized by any storage device such as a memory or a hard disk. For example, the storage unit 2521 stores weighting coefficients Wsi necessary for executing the information processing described below, in association with the length, weight, etc. of the object. The weighting coefficients are acquired in advance by executing machine learning from training data including attribute data and dimensional data described below. Note that when the object is a "person," the object is treated as different for each gender. In this case, different weighting coefficients are used for men and women. However, in the following description, when the object is a person, the description will be given without distinguishing between genders for convenience.

[0009] The input / output unit 2522 is realized by a keyboard, a mouse, a touch panel, etc., and is used to input various information to the computer and output various information from the computer.

[0010] The communication unit 2523 is realized by an arbitrary network card or the like, and enables communication with communication devices on a network via wired or wireless communication.

[0011] The processing unit 2524 executes various information processing operations and is realized by a processor such as a CPU or a GPU and a memory. Here, the CPU, GPU, or the like of the computer reads a program stored in the storage unit 2521, causing the processing unit 2524 to function as an acquisition unit 2524A, an extraction unit 2524B, a conversion unit 2524C, and a calculation unit 2524D.

[0012] The acquisition unit 2524A acquires attribute data including at least total length data of the object. The attribute data is data indicating the attributes of the object, such as the total length, weight, and time elapsed since creation (including age) of the object. If the object is a person, height, weight, age, etc. are defined as the attribute data.

[0013] Furthermore, the acquisition unit 2524A acquires image data of an object. When the image data is captured by a single camera, the acquisition unit 2524A acquires image data of the object captured from different directions. Specifically, the acquisition unit 2524A receives image data of the front and side of the object.

[0014] The extraction unit 2524B extracts shape data indicating the shape of an object from the image data. Specifically, the extraction unit 2524B extracts the shape data of the object by extracting an object region included in the image data using a semantic segmentation algorithm (Mask R-CNN, etc.) prepared for each type of object.

[0015] The conversion unit 2524C converts the shape data based on the full-length data to silhouette it. This normalizes the shape data. The conversion unit 2524C may convert the shape data into monochrome image data using not only the full-length data but also the depth data of the object region, generating a "gradation silhouette image" (new shape data). A gradient silhouette image is not simply black-and-white binary data, but a monochromatic, multi-gradation monochrome image represented by data with brightness values ​​ranging from 0 ("black") to 1 ("white") based on the depth data. In other words, the gradient silhouette image data is associated with the depth data and contains even more information regarding the shape of the object. In the following description, a gradient silhouette image may be referred to simply as a silhouette image without distinguishing between the two.

[0016] 2, the calculation unit 2524D has functions as a first calculation unit 2524D1, a second calculation unit 2524D2, and a determination unit 2524D3, and calculates the dimension data of each part of the object. In the following description, all calculated values ​​of the dimension data corresponding to each part of the object may be referred to as a "dimension data set."

[0017] The first calculation unit 2524D1 calculates a first calculated value of the dimensional data of each part of the object using the attribute data acquired by the acquisition unit 2524A. Specifically, the first calculation unit 2524D1 calculates the dimensional data of each part of the object by performing quadratic regression on the attribute data using weighting coefficients Wsi learned through machine learning. The weighting coefficients are optimized for each part of the object, and the weighting coefficient for the i-th part of the object is represented as Wsi. Note that the symbol i=1 to j, and the symbol j represents the total number of dimension locations for which dimensional data is to be calculated. The symbol s represents the number of elements used in the calculation obtained from the attribute data.

[0018] For example, assume that the attribute data of an object consists of total length data h, weight data w, and time-lapse data a. In other words, the attribute data is a set of elements (h, w, a). In this case, the first calculation unit 2524D1 calculates the value obtained by squaring each element (h, w, a) of the attribute data of the object (also called a quadratic term), the value obtained by multiplying each element (also called an interaction term), and the value of each element itself (also called a linear term). As a result, data having the following nine elements is obtained:

number

[0019] When the object is a "person", such first calculation unit 2524D1 calculates first calculated values ​​of dimensional data such as neck circumference, back shoulder width, back width, back height, total length, back length, sleeve length, elbow length, arm circumference, upper arm circumference, elbow circumference, wrist circumference, front length, breast drop, bust, underbust, chest width, bust point distance, waist, hips, mid-hip, waist length, armpit length, inseam length, knee length, thigh circumference, knee circumference, calf, ankle circumference, and rise.

[0020] The second calculation unit 2524D2 calculates a "second calculated value" of the dimensional data of the specific part of the object from the image data. Specifically, the second calculation unit 2524D2 regards the specific part as a virtual figure from the multiple shape data corresponding to the front and side shapes converted by the conversion unit 2524C, and calculates its perimeter. Then, the second calculation unit 2524D2 calculates the second calculated value of the dimensional data of the specific part of the object based on the perimeter.

[0021] When the object is a person, specific features include the bust, hips, and waist. These specific features are extracted based on predetermined rules. For example, in the example shown in FIG. 3, the second calculation unit 2524D2 calculates the widths hB1, hW1, and hH1 of the bust, waist, and hips on the side from the shape data representing the silhouette shape of the side view. Here, a predetermined region L of the shape data is extracted, and the shape data is divided into three parts in the height direction at a predetermined ratio. In the shape data LW, which is the second-highest division from the top, the most concave part in the width direction is determined as the waist position. In the shape data LB, which is the first-highest division from the top, the most convex part is determined as the bust position. In the shape data LH, which is the third-highest division from the top, the most convex part is determined as the hip position. Next, as shown in FIG. 4, the second calculation unit 2524D2 calculates the bust position, waist position, and hip position corresponding to the side shape data in the shape data representing the silhouette shape of the front view. The second calculation unit 2524D2 then calculates the widths hB2, hW2, and hH2 of the bust, waist, and hips on the front side, respectively. From the calculated front and side widths, the second calculation unit 2524D2 calculates the perimeter of the specific part when it is considered as a virtual figure. The virtual figure can have any shape as long as the perimeter can be calculated from the front and side widths. Here, the virtual figure is assumed to be an ellipse. The second calculation unit 2524D2 then outputs dimensional data of the specific part from the perimeter of this ellipse using a predetermined inference model. The above-mentioned inference model is constructed by machine learning a model such as linear regression or a neural network using a dataset including the perimeter of the predetermined virtual figure and the perimeter of the specific part of the predetermined object.

[0022] If the object is a person, the side shape data shown in FIG. 5 may be used in addition to the side shape data shown in FIG. 3. The side shape data shown in FIG. 5 shows the arms positioned higher than the waist. This prevents the waist and arms from overlapping in a side view, allowing the position and width of the waist to be determined with high accuracy. In addition to the waist, the position and width of the hips may also be determined from the side shape data shown in FIG. 5.

[0023] The determination unit 2524D3 determines a dimension data set, which is a collection of dimension data for each portion of the object, based on the calculation results by the first calculation unit 2524D1 and the second calculation unit 2524D2. Here, if the difference between the first calculated value and the second calculated value of the dimension data for a specific portion of the object is within a predetermined value, the determination unit 2524D3 adopts the dimension data of the second calculated value as the dimension data for the specific portion. Furthermore, if the difference between the first calculated value and the second calculated value of the dimension data exceeds the predetermined value, the determination unit 2524D3 outputs an error. Note that if the difference between the first calculated value and the second calculated value of the dimension data exceeds the predetermined value, the determination unit 2524D3 may adopt the dimension data of the first calculated value as the dimension data for the specific portion instead of outputting an error, or in addition to outputting an error.

[0024] FIG. 6 is a flowchart for explaining the operation of the dimension data calculation device 2520 according to the modified example of this embodiment.

[0025] First, the dimension data calculation device 2520 receives input of attribute data related to an object via a terminal device used by a user. As a result, the dimension data calculation device 2520 acquires this attribute data (X1). In particular, if the object is a "person," the dimension data calculation device 2520 acquires height, weight, age, and the like as attribute data.

[0026] The dimension data calculation device 2520 also acquires image data of an object via a terminal device used by the user (X2). The dimension data calculation device 2520 extracts shape data indicating the shape of each part of the object from each image data. The dimension data calculation device 2520 then performs a rescaling process to convert each piece of shape data to a predetermined size based on the total length data.

[0027] The terminal device may output guide information to a display screen or the like so that image data of an object having a predetermined shape can be acquired. For example, if the object is a person, the terminal device prompts the person to be photographed to assume a predetermined pose. Here, the terminal device outputs guide information so that a front image of the object (FIG. 4(a)) is captured in which the arms are spread so as not to overlap the torso when viewed from the front and are positioned so as to fit within the width of the torso when viewed from the side, a side image of the object (FIG. 3(a)) in which the arms are spread so as not to overlap the torso when viewed from the front and are positioned so as to fit within the width of the torso when viewed from the side, and a side image of the object (FIG. 5(a)) in which the arms are positioned higher than the waist when viewed from the front and from the side. For convenience, silhouetted images are shown as examples.

[0028] Next, the dimension data calculation device 2520 calculates the dimension data of each part of the object using the attribute data acquired by the acquisition unit 2524A (X3 to X6). Specifically, the dimension data calculation device 2520 calculates a first calculated value of the dimension data of each part of the object by performing quadratic regression on the attribute data using the machine-learned weighting coefficients Wsi.

[0029] Next, the dimension data calculation device 2520 calculates a second calculated value of the dimension data of the specific part of the object based on the image data (X7). Specifically, the second calculation unit 2524D2 of the dimension data calculation device 2520 calculates the perimeter of a virtual shape of the specific part of the object from the front and side images of the part. If the object is a "person" and the specific part is the waist, the dimension data calculation device 2520 assumes that the cross section along the imaging direction is an ellipse and calculates the perimeter of the ellipse from the image data. Next, the dimension data calculation device 2520 (second calculation unit 2524D2) outputs the dimension data of the waist using a model trained using the calculated perimeter of the ellipse and the actual dimension data of the waist as training data.

[0030] Next, the dimension data calculation device 2520 compares the first calculated value with the second calculated value for the specific part of the object, and if the difference between them is equal to or less than a predetermined value, adopts the second calculated value as the dimension data for the specific part (X8-Yes, X9). That is, among the dimension data for each part calculated from the first calculated value, the dimension data for the specific part is replaced with the second calculated value to determine the entire dimension data set.

[0031] On the other hand, the dimension data calculation device 2520 compares the first calculated value with the second calculated value, and if the difference between them exceeds a predetermined value, outputs an error (X8-No, X10). If the dimension data calculation device 2520 outputs an error, it sends a request to resend the image data to the user's terminal device or the like. However, instead of sending a resend request, the dimension data calculation device 2520 may adopt the first calculated value of the dimension data of a specific part. In this case, the entire dimension data set is determined from the first calculated value.

[0032] As described above, the dimension data calculation device 2520 according to this embodiment uses both attribute data and image data to calculate overall dimension data for multiple locations on an object with high accuracy. This effect is particularly pronounced when calculating dimension data for a human body. Additionally, the inventors discovered that while the dimension data for each part of the human body can generally grasp a certain trend based on height, weight, age, and gender, individual differences are relatively large for specific parts. Specific parts with large individual differences include the bust, hips, and waist. Therefore, when calculating human body dimension data, the dimension data calculation device 2520 according to this embodiment calculates the dimension data for parts with little individual difference at high speed with a certain degree of accuracy using polynomial regression using attribute data, and calculates the dimension data for specific parts with large individual differences with high accuracy using image data. This allows the dimension data for each part of the object to be calculated with high accuracy overall. Furthermore, using such dimension data makes it possible to manufacture products that fit the shape of the object.

[0033] Furthermore, in the dimension data calculation device 2520 according to this embodiment, by using an inference model that has learned the perimeter of a virtual figure that can be calculated from a small number of images and the actual length measured in advance as training data, the second calculated value of the dimension data of a specific part of a complex shape can be easily calculated from a small number of images.

[0034] Furthermore, in the dimension data calculation device 2520 according to this embodiment, the determination unit 2524D3 sets the dimension data of the specific part to the second calculated value if the difference between the first calculated value and the second calculated value is within a predetermined value, thereby improving the accuracy of the dimension data of the specific part, which is prone to individual differences in shape.

[0035] Furthermore, in the dimension data calculation device 2520 according to this embodiment, the determination unit 2224D3 outputs an error if the difference between the first calculated value and the second calculated value exceeds a predetermined value. Here, the determination unit 2224D3 may set the dimension data of a specific portion to the first calculated value in response to the error output. This configuration maintains the reliability of the dimension data set even when irregular image data is acquired. Furthermore, image data may contain noise due to factors such as the shooting environment, which may cause the second calculated value to deviate significantly from the actual dimensions. Therefore, when image data containing noise is acquired and an irregular second calculated value is calculated, adopting the first calculated value prevents abnormal values ​​from being included in part of the dimension data set. This allows a dimension data set with a certain level of reliability as a whole to be output without requiring reacquisition of image data, thereby improving user convenience.

[0036] The dimension data calculation device 2520 may include a depth data measurement device. By using the depth data measurement device, the image data acquired by the acquisition unit 2524A can be converted into RGB-D (Red, Green, Blue, Depth) data. This allows the perimeter of the virtual figure to be calculated from either the front or side image data, without necessarily acquiring both the front and side image data. In other words, since the image data includes a depth map, the second calculated value can be calculated from a single image data. An example of a depth data measurement device is a stereo camera. In this specification, the term "stereo camera" refers to any type of imaging device capable of simultaneously capturing images of an object from multiple different directions, reproducing binocular parallax, and constructing a depth map. In addition to a stereo camera, depth data may be obtained using a LiDAR (Light Detection and Ranging) device to construct a depth map.

[0037] (2) Application to product manufacturing systems An example in which the above-described dimension data calculation device 2520 is applied to a product manufacturing system 2501 will be described below.

[0038] FIG. 7 is a schematic diagram showing the concept of a product manufacturing system 2501 according to this embodiment. Product manufacturing system 2501 is a system for manufacturing a desired product 2506, and includes a dimension data calculation device 2520 capable of communicating with a terminal device 2510 held by a user 2505, and a product manufacturing device 2530. Fig. 7 shows, as an example, a concept of the overall configuration when object 2507 is a person and product 2506 is a chair. However, object 2507 and product 2506 of the product manufacturing system according to this embodiment are not limited to these.

[0039] The terminal device 2510 is realized by an arbitrary computer. Here, as an example, it is assumed to be realized by a smart device. Specifically, when a user program is installed in the terminal device 2510, the terminal device 2510 functions as an acquisition unit 2511, a communication unit 2512, a processing unit 2513, and an input / output unit 2514, as shown in FIG. 8. The terminal device 2510 may be carried by a user or may be installed in a store.

[0040] The acquisition unit 2511 receives input of attribute data indicating attributes of the object 2507. Examples of "attributes" include the total length, weight, and time elapsed since creation (including age) of the object 2507.

[0041] Furthermore, the acquisition unit 2511 acquires image data of the object 2507. For example, the acquisition unit 2511 is configured with an arbitrary monocular camera. The data acquired by the acquisition unit 2511 is processed by the processing unit 2513, and image data of the object is generated. Note that the acquisition unit 2511 may have a stereo camera function that simultaneously captures images of the object from multiple different directions to reproduce binocular parallax. Also, here, the acquisition unit 2511 is configured to have a camera function, but the camera function may be provided separately from the acquisition unit 2511. In this case, the image data is not limited to that captured by the terminal device 2510, and may be, for example, that captured using a stereo camera separately installed in a store.

[0042] The communication unit 2512 is realized by a network interface such as a network card, and enables wired or wireless communication with communication devices on a network. The function of the communication unit 2512 enables the terminal device 2510 to send and receive various information to and from the dimension data calculation device 2520 and the product manufacturing device 2530.

[0043] The processing unit 2513 is realized by a processor such as a CPU (Central Processing Unit) and / or a GPU (Graphical Processing Unit) and a memory in order to execute various information processes, and executes various processes by loading programs.

[0044] The input / output unit 2514 receives input of various information to the terminal device 2510 and outputs various information from the terminal device 2510. Specifically, the input / output unit 2514 is realized by an arbitrary touch panel. Furthermore, the input / output unit 2514 displays guide information on the screen as indicated by the dotted line in FIG. 9 so that an object 2507 of a desired shape can be photographed. This allows the user to photograph the object 2507 (a person in this case) so that it fits within the dotted line. If the image data is photographed so that the object 2507 fits within the dotted line, there is a high probability that the second calculated value in the dimension data calculation device 2520 will be a normal value. Here, when the object is a person, the input / output unit 2514 outputs guide information that prompts the user to capture a front image of the object with the arms spread so as not to overlap the torso when viewed from the front and positioned so as to fit within the width of the torso when viewed from the side, a side image of the object with the arms spread so as not to overlap the torso when viewed from the front and positioned so as to fit within the width of the torso when viewed from the side, and a side image of the object with the arms positioned higher than the waist when viewed from the front and from the side.

[0045] As described above, the dimension data calculation device 2520 can be realized by an arbitrary computer. Here, the storage unit 2521 of the dimension data calculation device 2520 stores information transmitted from the terminal device 2510 in association with identification information identifying the user 2505 of the terminal device 2510 and / or identification information identifying the object 2507. The storage unit 2521 also stores parameters and the like necessary for executing information processing, which will be described later. For example, the storage unit 2521 stores the weighting coefficients Wsi described above in association with attribute items of the object 2507 and the like.

[0046] Product manufacturing device 2530 is a device that manufactures a desired product related to the shape of object 2507 using the dimension data set calculated using dimension data calculation device 2520. Note that product manufacturing device 2530 can be any device that can automatically manufacture and process a product, and can be realized by, for example, a three-dimensional printer.

[0047] FIG. 10 is a sequence diagram for explaining the operation of the product manufacturing system 2501 according to the bookbinding embodiment.

[0048] First, the user 2505 inputs (Y1) attribute data indicating the attributes of the object 2507 to the terminal device 2510. Here, the attribute data inputted includes the total length data, weight data, and time-lapse data (including age, etc.) of the object 2507.

[0049] Next, the object 2507 is imaged multiple times via the terminal device 2510 so that the entire object 2507 is captured from different directions, and multiple pieces of image data of the object 2507 are generated (Y2). Here, the object 2507 is a person, and as described with reference to Figures 3 to 5, a front image of the object with the arms spread so as not to overlap with the torso when viewed from the front and positioned so as to fit within the width of the torso when viewed from the side, a side image of the object with the arms spread so as not to overlap with the torso when viewed from the front and positioned so as to fit within the width of the torso when viewed from the side, and a side image of the object with the arms positioned higher than the waist when viewed from the front and from the side are captured.

[0050] Then, these multiple pieces of image data and attribute data are transmitted from the terminal device 2510 to the dimension data calculation device 2520. At this time, the terminal device 2510 may be configured to determine whether or not an object included in the image data (i.e., reflected in the image data) is a pre-registered object. Specifically, the processing unit 2513 of the terminal device 2510 determines whether or not the object included in the image data is a pre-registered object using an "object identification model" that identifies whether or not each pixel is a predetermined object (Y3). For example, if the object is a "person," the processing unit 2513 determines whether or not a person is reflected in the image data, and if a person is not reflected, outputs this to the input / output unit 2514. This allows the user to confirm whether or not the image data is appropriate before transmitting the image data to the dimension data calculation device 2520 (Y3-Yes, Y4).

[0051] When the dimension data calculation device 2520 receives a plurality of image data and attribute data from the terminal device 2510, it uses these data to calculate dimension data for each part of the object 2507 and outputs a dimension data set (Y5). Note that the terminal device 2510 displays predetermined dimension data on the screen according to the settings.

[0052] Then, the product manufacturing device 2530 manufactures the desired product 2506 based on the dimension data set calculated by the dimension data calculation device 2520 (Y6).

[0053] As described above, the dimension data calculation device 2520 calculates the dimension data set of the object 2507 with high accuracy, and therefore it is possible to provide a desired product 2506 related to the shape of the object 2507. For example, a chair that fits a person can be manufactured based on the shape of the person as the desired product 2506. Also, for example, clothing that is customized to fit the body shape of the person can be manufactured.

[0054] In addition, when the object is a person, not only can front and side images of the object be acquired in which the arms are spread apart so as not to overlap the torso in the front view and positioned so as to fit within the width of the torso in the side view, but also side images of the object in which the arms are positioned higher than the waist in the front and side views can be acquired to obtain side images in which the arms are not captured in the waist and hip areas. This allows for highly accurate extraction of the waist and hip areas. As a result, waist and hip dimensional data, which vary significantly among individuals, can be calculated with high accuracy while reducing the computational load. Additionally, when capturing a side image of a person, the arms and torso may overlap in the side view, making it difficult to accurately capture the silhouette of the torso. This is due, for example, to the fact that many people arch their backs, or that many people actually lower their arms at an angle forward or backward even when they think they are lowering their arms straight down. Therefore, the inventors have made it possible to improve the accuracy of the dimension dataset by outputting guide information that encourages the capture of side images of different body types and acquiring images prompted by this guide information.

[0055] Additionally, the product manufacturing system 1001 can manufacture organ models from measurements of the shapes of various organs such as the heart. Also, for example, various healthcare products can be manufactured from measurements of a person's waist shape. Also, for example, a figure product of a person can be manufactured from the shape of the person.

[0056] In the above description, the dimension data calculation device 2520 and the product manufacturing device 2530 are described as separate devices, but they may also be configured as an integrated device.

[0057] The present disclosure is not limited to the above-described embodiments as they are. The present disclosure can be embodied by modifying the components within the scope of the gist of the disclosure in the implementation stage. Furthermore, the present disclosure can be formed into various disclosures by appropriately combining multiple components disclosed in the above-described embodiments. For example, some components may be deleted from all the components shown in the embodiments. Furthermore, the components may be appropriately combined in different embodiments. [Explanation of symbols]

[0058] 2501 Product Manufacturing Systems 2505 users 2506 products 2507 Object 2510 Terminal Equipment 2511 Acquisition Department 2512 Communications Department 2513 Processing section 2514 Input / output section 2520 Dimensional data calculation device 2521 Storage section 2522 Input / output section 2523 Communications Department 2524 Processing section 2524A Acquisition Department 2524B Extraction part 2524C conversion unit 2524D Calculation Unit 2524D1 First Calculation Section 2524D2 Second calculation unit 2524D3 Decision Section 2530 Product manufacturing equipment GI Guide Information [Prior art documents] [Patent documents]

[0059] [Patent Document 1] Japanese Patent Application Laid-Open No. 2011-227692

Claims

1. an acquisition unit that acquires attribute data including at least total length data of the object; a first calculation unit that calculates first calculated values ​​of dimensional data of a plurality of portions of the object including a specific portion by performing polynomial regression on the attribute data using coefficients learned by machine learning; a receiving unit that receives image data of the object; a second calculation unit that calculates a second calculated value of dimensional data of the specific portion of the object from the image data; a determination unit that determines dimensional data of each portion of the object based on the first calculated value and the second calculated value; Equipped with the determination unit sets the dimensional data of the specific portion to the second calculated value when a difference between the first calculated value and the second calculated value is within a predetermined value. Dimensional data calculation device.

2. an acquisition unit that acquires attribute data including at least total length data of the object; a first calculation unit that calculates first calculated values ​​of dimensional data of a plurality of portions of the object including a specific portion by performing polynomial regression on the attribute data using coefficients learned by machine learning; a receiving unit that receives image data of the object; a second calculation unit that calculates a second calculated value of dimensional data of the specific portion of the object from the image data; a determination unit that determines dimensional data of each portion of the object based on the first calculated value and the second calculated value; Equipped with the determination unit sets the dimensional data of the specific portion to the first calculated value when a difference between the first calculated value and the second calculated value exceeds a predetermined value. Dimensional data calculation device.

3. The object is a person.

3. The dimension data calculation device according to claim 1 or 2.

4. The reception unit, a front image of the object in which the arms are spread so as not to overlap the torso when viewed from the front and the arms are positioned so as to fit within the width of the torso when viewed from the side; a side image of the object in which the arms are spread so as not to overlap the torso when viewed from the front and the arms are positioned so as to fit within the width of the torso when viewed from the side; Side images of the object with the arms positioned higher than the waist in front and side views; Accepting image data including:

4. The dimension data calculation device according to claim 3.

5. The second calculation unit calculating a perimeter of a virtual figure from the image data, and calculating a second calculated value of dimensional data of the specific portion of the object based on the perimeter; The dimension data calculation device according to any one of claims 1 to 4.

6. the second calculation unit calculates a second calculated value of dimensional data of the specific part of the object using a learning device that has been trained to output the perimeter of the specific part of a predetermined object in response to an input of the perimeter of a predetermined virtual figure.

6. The dimension data calculation device according to claim 5.

7. The image data is image data including a depth map. The dimension data calculation device according to any one of claims 1 to 6.

8. a generation unit that generates a silhouette image of the object from the image data, wherein the second calculation unit calculates the second calculated value from the silhouette image. The dimension data calculation device according to any one of claims 1 to 7.

9. a communication unit that communicates with a terminal device that generates image data of the object, the reception unit receiving the image data of the object transmitted from the terminal device; the determination unit requests the terminal device to retransmit the image data when a difference between the first calculated value and the second calculated value exceeds a predetermined value; The dimension data calculation device according to any one of claims 1 to 8.

Images