Hunched posture determination device, hunched posture determination system, hunched posture determination method, and program

Abstract

To provide a hunched posture determination device, a hunched posture determination system, a hunched posture determination method, and a hunched posture determination program capable of easily and quickly determining whether or not a person has a hunched posture.SOLUTION: The curved-back determination device 13 includes a length calculation part 21 and a determination part 22. The length calculation unit 21 obtains a first length L1 of a line segment connecting the seventh cervical vertebra and the fourth lumbar vertebra identified based on a captured image obtained by imaging, from the side, a subject to be determined for the round-shouldered state, and a maximum second length L2 that is the longest among second lengths L2max from the line segment to a contour line on the back side in a subject image of the captured image. The determination unit 22 determines that the posture is bow-legged when the bow-leggedness index obtained by the calculation formula of (L2max / L1) * 100 is equal to or greater than a preset reference value, and determines that the posture is not bow-legged when the bow-leggedness index is less than the reference value.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a hunchback posture determination device, a hunchback posture determination system, a hunchback posture determination method, and a hunchback posture determination program. [Background technology]

[0002] Posture is considered important as it affects the state of the human body. To ensure proper posture, for example, Patent Document 1 discloses a skeletal identification system that identifies the femoral line connecting the greater trochanter of the femur and the kneecap, and the tibial line connecting the kneecap and the ankle from a deformed skeletal model, and calculates the angle at which these femoral and tibial lines intersect. This skeletal identification system generates an angle identification image including the calculated angle in an image of the human body acquired in conjunction with three-dimensional measurement, thereby allowing the condition of the target skeletal location to be confirmed in terms of an angle value along with the image of the human body.

[0003] Furthermore, hunched posture is considered to be a posture that should be improved, and a method has been proposed to determine whether or not a person has a hunched posture using a kyphosis index. The kyphosis index is calculated by drawing a line at a specific position on a photograph of a person, measuring the length of the line, and measuring the distance from the line to the back of the person, and then using the length of the line and the distance. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] International Publication No. 2017 / 170264 Summary of the Invention [Problem to be solved by the invention]

[0005] Although using the kyphosis index to determine whether or not a person has a hunched back is effective, drawing a line on a photograph and manually measuring the line and the distance as described above is inefficient. Furthermore, the skeleton identification system described in Patent Document 1 identifies specific lines on the body, but generates an angle-specific image including the calculated angle in the captured image of the human body acquired in conjunction with three-dimensional measurement. Therefore, even if it is used to determine posture, the process of obtaining the determination result is complicated and time-consuming.

[0006] Therefore, an object of the present invention is to provide a hunchback posture determination device, a hunchback posture determination system, a hunchback posture determination method, and a hunchback posture determination program that can easily and quickly determine whether or not a person has a hunchback posture. [Means for solving the problem]

[0007] The hunchback assessment device of the present invention includes a length calculation unit and a assessment unit. The length calculation unit calculates a maximum second length L2, which is the longest of a first length L1 of a line segment connecting the seventh cervical vertebra and the fourth lumbar vertebra identified based on an image captured from the side of a subject to be assessed for hunchback, and a second length L2 from the line segment to a dorsal contour line in the subject image in the image. max The determination part calculates (L2 max If the kyphosis index calculated using the formula (L1 / L1) × 100 is equal to or greater than a preset reference value, the subject is determined to have a hunched back, and if it is less than the reference value, the subject is determined not to have a hunched back.

[0008] Preferably, the image is a side image of the subject's entire body. Preferably, the image is a side image of the subject in a standing position.

[0009] It is preferable that the apparatus further comprises an acquisition unit and an estimation unit. The acquisition unit acquires the image. The estimation unit estimates pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra in the subject image and contour information indicating at least the dorsal contour line of the subject image. In this case, the length calculation unit calculates a first length L1 based on the pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra, calculates a second length L2 based on the pixel coordinates and the contour information, and selects the maximum second length L2 from the calculated multiple second lengths. max is determined by identifying

[0010] It is preferable that the estimation unit learns a dataset including a side image showing the side of a human body and the pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra in the subject image in the side image, and, when a side image is input, uses a trained model that outputs the pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra, to identify the pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra in the subject image in the image when the image is input as a side image and output them to the length calculation unit.

[0011] It is preferable that the image processing device further comprises a display control unit that generates a display image showing the determination result determined by the determination unit.

[0012] The hunchback detection system of the present invention includes a terminal device and the hunchback detection device. The terminal device includes a camera and a transmitter that transmits an image captured by the camera.

[0013] The hunchback assessment method of the present invention includes a length calculation step and a assessment step. The length calculation step calculates a maximum second length L2, which is the longest of a first length L1 of a line segment connecting the seventh cervical vertebra and the fourth lumbar vertebra identified based on an image of a subject to be assessed for hunchback, and a second length L2 from the line segment to a dorsal contour line in the subject image. max The decision step is (L2 max If the kyphosis index calculated by the formula (L1 / L1)×100 is equal to or greater than a predetermined reference value, the subject is determined to have a hunched back, and if it is less than the reference value, the subject is determined not to have a hunched back.

[0014] The hunchback assessment program of the present invention causes a computer to execute the above steps. [Effects of the Invention]

[0015] According to the present invention, it is possible to easily and quickly determine whether or not a person has a hunched back. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is an explanatory diagram of a hunchback determination system according to an embodiment; [Figure 2] 10A and 10B are explanatory diagrams of a method for capturing an image of a subject to be determined as having a hunched posture. [Figure 3] FIG. 2 is a functional block diagram of the hunchback determination device. [Figure 4A] FIG. 4 is an explanatory diagram of a storage mode in a user storage unit. [Figure 4B] FIG. 4 is an explanatory diagram of a storage mode of a reference storage unit. [Figure 5] FIG. 10 is an explanatory diagram of pixel coordinates. [Figure 6A] FIG. 1 is an explanatory diagram of a method for imaging a subject. [Figure 6B] FIG. 1 is an explanatory diagram of a method for imaging a subject. [Figure 6C] FIG. 1 is an explanatory diagram of a method for imaging a subject. [Figure 7] FIG. 1 is an explanatory diagram illustrating how to determine the kyphosis index. DETAILED DESCRIPTION OF THE INVENTION

[0017] The hunchback assessment system 10 shown in Fig. 1 is for determining whether an image of a person has a hunchback based on a captured image of the person as a subject, and for notifying the result of the determination (hereinafter referred to as the determination result). The hunchback assessment system 10 comprises a user terminal device (hereinafter referred to as the user terminal) 11, a hunchback assessment device 13, and a management terminal device (hereinafter referred to as the management terminal) 15. The user terminal 11, the management terminal 15, and the hunchback assessment device 13 communicate with each other via a communication network CN, i.e., send and receive various information.

[0018] The user terminal 11 is a terminal device used by a user who uses the service provided by the hunchback detection device 13, specifically, a user who uses a service that determines whether or not the user has a hunchback and notifies the user of the determination result. Examples of users include, but are not limited to, a person to be determined to be hunchbacked (hereinafter simply referred to as the "subject"), a physical therapist or chiropractor involved in analyzing or improving the subject's posture and / or body movement, a trainer who provides physical training to the subject, etc. In this embodiment, the following description will be given by way of example of a case in which the user is the subject, but the same applies to other users.

[0019] The user terminal 11 includes a camera 11a (see FIG. 2) that captures an image of the subject. In this example, the user terminal 11 further includes a transmitter 11b that transmits the captured image to the hunchback assessment device 13, and a display 11c (see FIGS. 6A and 6B) that displays the assessment results. The camera 11a may be any device capable of capturing a still image of the subject. In this example, the display 11c is a touch panel display that also functions as an input unit for input operations. However, the input unit may be a known input device capable of various input operations, such as a mouse or keyboard. The user terminal 11 is not particularly limited and may be any device, such as a mobile terminal, smartphone, or personal computer, as long as it is equipped with a camera and a transmitter. Although only one user terminal 11 is depicted in FIG. 1, the number of user terminals 11 may be multiple, so that multiple users can use each terminal. In this embodiment, an example will be described in which an image is captured by the camera 11a of one user terminal 11 and the judgment result from the hunchback judgment device 13 is received by the one user terminal 11, but it may also be configured so that an image is captured by the camera 11a of one user terminal 11 and the judgment result from the hunchback judgment device 13 is received by another user terminal 11. Note that in this example, the user terminal 11 is also used for input operations to register a user.

[0020] The hunchback assessment device 13 performs predetermined processing in response to input of information from each of the user terminal 11 and the management terminal 15. For example, when a assessment request for assessment of hunchback is sent from the user terminal 11, the hunchback assessment device 13 responds to the input and, based on the captured image included in the assessment request, assesses whether the subject image has a hunchback or not, and transmits the assessment result to the user terminal 11 to display it on the display 11c. Details of the hunchback assessment device 13 will be described later using another drawing.

[0021] The management terminal 15 is a terminal device that manages various settings and / or data of the hunchback assessment device 13. Examples of various settings include registering the reference values ​​described below when determining whether or not a person has a hunchback. Examples of data management include managing (storing) user information about registered users and managing (storing) the reference values ​​described below. The management terminal 15 in this example is configured as a computer equipped with an input unit and a display unit (display), but from the perspective of managing (storing) data, it does not necessarily have to include an input unit and / or a display unit. Any of a mobile terminal, a smartphone, a personal computer, etc. may be used, and there is no particular limitation.

[0022] The user terminal 11 and the management terminal 15 may operate on a browser by logging in to the hunchback assessment device 13, or may be installed with predetermined application software and operate by executing a program of this application software. The hunchback assessment device 13 has a predetermined program installed, and by executing this program, it functions as each unit described below and performs predetermined processing.

[0023] The program incorporated in the hunchback assessment device 13 causes a computer to execute a length calculation step and a assessment step. The length calculation step calculates the maximum second length L2, which is the longest of the first length L1 of the line segment connecting the seventh cervical vertebra and the fourth lumbar vertebra identified based on an image of the subject to be assessed for hunchback, captured from the side, and the second length L2 from the line segment to the dorsal contour line of the subject image in the image. max The decision step is (L2max If the hunchback index calculated by the formula (L1 / L1) × 100 is equal to or greater than a preset reference value, the user is determined to have a hunchback, and if it is less than the reference value, the user is determined not to have a hunchback. When this program is used in the application software program and incorporated into the user terminal 11, the user terminal 11 can be operated and used as the hunchback determination device 13.

[0024] In FIG. 2, the user terminal 11 is arranged so that the camera 11a mounted thereon faces the side of the subject P. This allows for a captured image of the subject from the side. The camera 11a is arranged so that it can capture an image of at least a partial region BP of the subject P's entire body BW, from the seventh cervical vertebra Pa1 to the fourth lumbar vertebra Pb1. This allows for a captured image of the partial region BP of the subject P captured from the side. However, it is preferable to arrange the camera 11a so that the entire body BW of the subject P, from the top of the head to the soles of the feet, is captured in order to obtain more accurate determination results. In this example, an image of the subject's entire body captured from the side is obtained in this manner. It is preferable to arrange the camera 11a so that the imaging optical axis OA intersects with the sagittal plane SP of the subject P's body in order to obtain more accurate determination results, and more preferably, it is arranged so that it is perpendicular. In this example, the camera 11a is also arranged so that the imaging optical axis OA is as perpendicular as possible to the sagittal plane SP.

[0025] The subject P may be imaged in a standing position, i.e., standing up, or in a sitting position, i.e., sitting in a chair or the like. In this example, the subject P is imaged in a state in which the standing position or the sitting position is selected on the user terminal 11, and the hunchback determination device 13 makes a determination according to the selected position. The method of selecting the standing position or the sitting position on the user terminal 11 will be described later using another drawing.

[0026] 3, hunchback assessment device 13 includes length calculation unit 21 and assessment unit 22. It is preferable that hunchback assessment device 13 further includes acquisition unit 25, estimation unit 26, user storage unit 31, reference storage unit 32, and controller 33, as is the case in this example. User storage unit 31 and reference storage unit 32 form a database DB.

[0027] The controller 33 comprehensively controls each component of the hunchback determination device 13. The controller 33 also functions as a transmitter / receiver that transmits and receives data to and from the user terminal 11 and the management terminal 15, and as a display controller that generates images to be displayed on the display 11c of the user terminal 11 and the display of the management terminal 15.

[0028] The user storage unit 31 stores user information that associates a user's identification information (ID) with a determination result for the user. The user information may also include, for example, the user's name. In addition to a newly obtained determination result, past determination results (hereinafter referred to as historical determination results) may be stored as the determination result. The determination result may include, in addition to a determination category indicating whether the user has a hunched back, the date of the determination, a posture category indicating whether the determination was made in a standing or sitting position, and the like. In this example, the determination category defines a hunched back as hunched and a non-hunched back as normal. The user storage unit 31 may store user information for each of multiple users, and this example does so as well (see FIG. 4A). The user information, excluding the determination result, such as the user's name and ID, is stored in the user storage unit 31 by the controller 33 when a user registration request requesting user registration is sent from the user terminal 11. The determination result is stored by the determination unit 22. In the case of storing the historical judgment results, the latest judgment result newly obtained by the judgment unit 22 is stored in addition to the historical judgment results. When storing user information in the management terminal 15, a user memory unit 31 may be provided in the management terminal 15 instead of the hunchback judgment device 13, and the above user information may be read from the user memory unit 31 provided in the management terminal 15 via the controller 33 at an appropriate timing and stored in the user memory unit 31.

[0029] The reference memory unit 32 stores a reference value of the kyphosis index, which is set as a criterion for determining whether or not a subject has hunched posture. The kyphosis index will be described later with reference to another drawing. The reference value of the kyphosis index is preset, for example, by an administrator using the management terminal 15. When a reference value registration request for registering the reference value is sent from the management terminal 15, the controller 33 stores the reference value in the reference memory unit 32. Note that instead of storing the reference value in the reference memory unit 32 in response to the reference value registration request, the reference value may be preset in a program and stored in the reference memory unit 32. The reference value of the kyphosis index may be either a standing reference value for determining whether a subject is standing or a sitting reference value for determining whether a subject is sitting, or both. In this example, the reference memory unit 32 stores both a standing reference value and a sitting reference value (see FIG. 4B). The standing reference value and the sitting reference value are not particularly limited and can be set as appropriate, and may be set by referring to values ​​disclosed in papers, etc. In this example, the standing reference value is set to "12.70" and the sitting reference value is set to "17.90" and stored in the reference storage unit, and these values ​​can be updated when a new reference value registration request is sent from the management terminal 15. When the reference values ​​are stored in the management terminal 15, a reference storage unit 32 may be provided in the management terminal 15 instead of the hunchback assessment device 13. In this case, the above-mentioned reference values ​​may be read from the reference storage unit 32 provided in the management terminal 15 via the controller 33 at an appropriate timing and stored in the reference storage unit 32.

[0030] The acquisition unit 25 acquires the above-mentioned determination request from the user terminal 11 via the controller 33 and sends it to the estimation unit 26. The determination request includes the user ID, the captured image captured by the camera 11a, and a posture classification indicating whether the posture of the subject P was selected as standing or sitting at the time of capturing the image. The acquisition unit 25 may identify the user ID indicated in the acquired determination request in the user storage unit 31, and store the user information including the identified user ID together with the captured image included in the acquired determination request.

[0031] The estimation unit 26 is a posture estimation AI (Artificial Intelligence) having a trained model 26a, and uses this trained model 26a to estimate pixel coordinates of the seventh cervical vertebra Pa2 (see FIG. 7) and the fourth lumbar vertebra Pb2 (see FIG. 7) in the subject image of the captured image input from the acquisition unit 25, as well as contour information indicating at least the dorsal contour line of the subject image. The estimation unit 26 sends the estimated pixel coordinates and contour information of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 to the length calculation unit 21, together with the user ID and posture classification input from the acquisition unit 25. The pixel coordinates will be described later using another drawing.

[0032] The trained model 26a has trained a dataset in which a side image showing the side of a human, the human region in the side image, and the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 identified in the side image are input parameters, and a pth (PyTorch) file serving as an estimation model file is output parameters. When a side image is input, the trained model 26a estimates a subject image of the subject P, estimates the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2, generates a masked image in which the region of the subject image is masked, and estimates contour information of the subject image. Thus, when a captured image showing the side of the subject P is input, the estimation unit 26 estimates the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 in the subject image in the captured image, generates a masked image, estimates contour information, and outputs the masked image to the length calculation unit 21. The contour information is composed of the pixel coordinates of multiple points constituting the contour of the subject image. The estimation unit 26 can use any publicly available library (open source library) and is not particularly limited. The estimation unit 26 may be configured with one AI or two AIs. When configured with two AIs, one of them estimates the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2, while the other generates a masked image in which the area of ​​the subject image is masked to estimate contour information of the subject image. The other AI can use MediaPipe, a known open-source library. When configured with two AIs, one of them has a trained model that, when a side image is input, uses a dataset including a side image showing the side of a human, the human area in the side image, and the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 identified in the side image, and is trained to estimate the human subject image and output the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2.When configured with two AIs, one of the AIs has a trained model that uses a dataset including a side image showing the side of a human and the human area in the side image, and when configured with two AIs, uses a side image showing the side of a human and the human area in the side image, and when a side image is input, it estimates a human subject image and outputs a masked image in which the area of ​​the subject image is masked and contour information of the subject image.

[0033] The side image used for training may be a captured image of a person taken from the side, or an image generated using image generation AI; in this example, the latter is used. As is well known, unlike AI that uses a structured trained model that clearly separates explanatory variables from target variables, generation AI uses the unstructured trained model to output (generate) a certain degree of creative processing results. When a generation request is made to such a trained model (a prompt indicating the request is sent), the image generation AI generates an image as a creative processing result. A publicly available image generation AI can be used, such as Stable Diffusion, which generates images based on input text and / or image prompts.

[0034] The side images used for learning are preferably side images showing the side of the whole human body when capturing images including the whole body BW of the subject P. In this example, too, captured images including the whole body BW are captured, and the side images are side images showing the side of the whole human body.

[0035] It is preferable to use both standing side images showing the side view of the entire body of a person in a standing position and sitting side images showing the side view of the entire body of a person in a sitting position as the side images used for learning, as this allows for more accurate assessment of hunched posture in both the standing and sitting positions.

[0036] The seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 in the lateral image may be identified in advance, for example, in the management terminal 15, or in a terminal device separate from the management terminal 15 and the user terminal 11; in this example, the latter is used. While AI is used for the estimation unit 26 in this example, it is not limited to AI. For example, markers may be attached to the body surface above the seventh cervical vertebra Pa1 and the body surface above the fourth lumbar vertebra Pb1 of the subject P, and the pixel coordinates of the marker images in the captured image may be identified. Examples of markers include markers used in motion capture, such as reflective markers that reflect light and emission markers that emit light.

[0037] When the masking image, the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2, and the contour information are input from the estimation unit 26, the length calculation unit 21 calculates, in response to this input, the longest of the first length L1 of the line segment LP (see FIG. 7) connecting the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 in the masking image and the second length L2 from the line segment LP to the dorsal contour line, which is the maximum second length L2. max Since the masking image is obtained by masking the subject image based on the captured image, the dorsal contour line in the masking image can be considered to match the dorsal contour line in the subject image in the captured image. In this example, when capturing an image of the subject P, the subject P is captured from the right lateral side, and a captured image capturing the right side of the subject P is obtained. Therefore, in the captured image and the masking image, the contour line present to the left of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 is specified as the dorsal contour line. The length calculation unit 21 calculates the calculated first length L1 and maximum second length L2 max The length calculation unit 21 sends the first length L1, the second length L2, the maximum second length L2, together with the user ID and the posture classification input to the determination unit 22. max The method for determining this will be described in detail later with reference to another drawing.

[0038] The determination unit 22 receives the first length L1 and the maximum second length L2 from the length calculation unit 21. maxWhen the input is given, the kyphosis index is calculated in response to this input. The kyphosis index is (L2 max The determination unit 22 reads out the reference value of the kyphosis index for the posture category input from the length calculation unit 21 from the reference storage unit 32. The determination unit 22 compares the read-out reference value with the obtained kyphosis index, and determines that the user has a hunched back if the obtained kyphosis index is equal to or greater than the reference value, and determines that the user does not have a hunched back (in this example, normal) if the obtained kyphosis index is less than the reference value. The determination unit 22 sends the determination result to the user terminal 11 via the controller 33. At this time, the controller 33 may generate a determination result display image showing the determination result, for example, a determination result display image including a determination result display such as "Determination result: hunched back" or "Determination result: normal," and display it on the display 11c of the user terminal 11. The determination unit 22 further identifies user information including the user ID input from the length calculation unit 21 in the user storage unit 31, and stores the obtained determination result in the identified user information as the latest determination result.

[0039] The pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 identified in the lateral image, and the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 estimated by the estimation unit in the captured image, and the pixel coordinates constituting the contour information, are identified and estimated using the pixel coordinates of the upper left corner of the lateral image and the captured image as the origin D0 of (0,0), as shown in Fig. 5. Note that in the pixel coordinate system, the X coordinate increases toward the right in the horizontal direction (lateral direction), and the Y coordinate increases toward the bottom in the vertical direction (longitudinal direction).

[0040] A method for selecting a standing posture or a sitting posture on the user terminal 11 and capturing an image will be described with reference to FIGS. 6A and 6B. In this example, when a user logs in to the user terminal 11, the controller 33 (see FIG. 3) cooperates with the camera 11a to display the captured area captured by the camera and various displays as image G1 on the display 11c of the user terminal 11. Image G1 shows a state in which the standing posture has been selected, and includes a posture selection indicator D1 reading "Captured in a standing posture," indicating that the selected posture is the standing posture. Image G1 includes an upper limit mark M1 and a lower limit mark M2, which are horizontally extending lines and are preset to obtain a subject image suitable for assessment. Thus, a captured image suitable for assessment can be obtained by capturing an image of the subject P (see FIG. 2) so that at least the aforementioned partial region BP (see FIG. 2), preferably the entire body BW (see FIG. 2), fits within the region between the upper limit mark M1 and the lower limit mark M2 as the body region to be captured. Image G1 has a shutter button B1, and is captured when this shutter button B1 is touched.

[0041] After capturing the image, the controller 33 causes an image G2 for confirming the captured image to be displayed on the display 11c as shown in Fig. 6B. Image G2 has an image portion G2a showing the captured image, and image G1 has a button B2 for requesting a judgment; when this button B2 is touched, a judgment request including the captured image and the user ID is transmitted from the user terminal 11 to the hunchback detection device 13. Image G2 further has a button B3 for redisplaying image G1 to recapture the image; when this button B3 is touched, the controller 33 causes image G2 on the display 11c to transition to image G1, allowing the user (the subject in this example) to recapture the image.

[0042] Image G1 (see FIG. 6A) has a switch button B4 for switching the posture during image capture. In image G1 where the standing posture is selected, the switch button B4 displays "shoot in sitting position." When this switch button B4 is touched, controller 33 displays image G3 on display 11c, as shown in FIG. 6C, with a posture selection indicator D1 for "shoot in sitting position," indicating that the selected posture is the sitting position. Like image G1, image G3 has an upper limit mark M1, a lower limit mark M2, a shutter button B1, and a button B4, and the same touch operations as those on image G1 are performed. Note that an image (not shown) for selecting the standing posture or the sitting posture may be displayed before image G1 or image G2 is displayed. In this case, images G1 and G2 do not need to have the switch button B3.

[0043] First length L1, second length L2, maximum second length L2 max The method for calculating the kyphosis index will be described with reference to FIG. 7. The captured image and the masked image G3 generated from this captured image have the same size in the vertical and horizontal directions, and pixel coordinates in the captured image and pixel coordinates in the masked image G3 are the same. The length calculation unit 21 calculates a line segment LP with the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 as its endpoints based on the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 in the masked image G3, and calculates a first length L1. The coordinates of points on the line segment LP are defined in an XY coordinate system in which the upper left corner of the masked image G3 is set as the origin (0,0) and orthogonal X and Y axes are set, with the X coordinate increasing toward the right and the Y coordinate increasing toward the bottom, as shown in FIG. 7.

[0044] The length calculation unit 21 then virtually scans the line segment LP, sequentially determining the distance from each of the plurality of points on the line segment LP to the dorsal contour line OL as a second length L2. The number of points selected on the line segment LP to determine the second length L2 is not particularly limited, and in this example, a plurality of points are selected with a difference in Y coordinate of 1. The length calculation unit 21 then selects the maximum value from the determined second lengths L2 as the maximum second length L2. maxThe determination unit 22 determines the max The kyphosis index is calculated using the formula: / L1) × 100. Note that in FIG. 7, for convenience of explanation, outlines of bones are shown by two-dot chain lines (virtual lines) within the masking region, but bones are not depicted in the masking image G3. In the example shown in FIG. 7, the masking region is the entire area within the contour line of the subject image, but the masking region is not limited to this example. For example, a masking region of a predetermined constant width along the contour line may be formed inside the contour line of the subject image.

[0045] According to the above configuration, the length calculation unit 21 and the determination unit 22 are provided, and the length calculation unit 21 calculates the maximum second length L2, which is the longest of the first length L1 of the line segment LP connecting the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 identified based on the captured image, and the second length L2 from the line segment LP to the dorsal contour line OL. max The determination unit 22 calculates a kyphosis index based on these and outputs a determination result using the reference value as a determination criterion. This makes it easy and quick to determine whether or not the subject has a hunched back. In addition, since the reference value is set and stored in advance, the determination by the determination unit 22 is extremely quick. Furthermore, since the line segment LP is calculated based on the pixel coordinates of the seventh cervical vertebra Pa2 and the fourth lumbar vertebra Pb2 in the image, the time required for calculation is extremely short.

[0046] In the above example, the hunchback judgment device 13 sends the judgment results to the user terminal 11 and displays them, but instead of this example, the judgment results may be used in a posture judgment device that classifies human postures in more detail and judges the posture of the subject P. [Explanation of symbols]

[0047] 10 Hunchback detection system 11 User terminal 11a camera, 11b transmitter 13 Hunchback detection device 21 Length calculation unit 22 Judgment section 25 Acquisition Department 26 Estimation part P Target Audience Pa2 7th cervical vertebra Pb2 4th lumbar vertebra BW whole body LP line segment

Claims

1. The longest of the first length L1 of the line segment connecting the seventh cervical vertebra and the fourth lumbar vertebra identified based on an image of the subject to be determined to have a hunched posture taken from the side and the second length L2 from the line segment to the dorsal contour line of the subject image in the image is the maximum second length L2. ｍａｘ a length calculation unit for calculating (L2 ｍａｘ a determination unit that determines that the subject has a hunched back when the kyphosis index calculated by the calculation formula (L1 / L2) × 100 is equal to or greater than a predetermined reference value, and determines that the subject does not have a hunched back when the kyphosis index is less than the reference value; A posture detection device comprising:

2. The hunchback detection device according to claim 1 , wherein the image is an image of the subject's entire body taken from the side.

3. 3. The hunchback detection device according to claim 1, wherein the image is an image of the subject in a standing position taken from the side.

4. an acquisition unit that acquires the image; an estimation unit that estimates pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra in the subject image and contour information that indicates at least a dorsal contour line of the subject image; Equipped with The length calculation unit The first length L1 is calculated based on pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra, and the second length L2 is calculated based on the pixel coordinates and the contour information, and the maximum second length L2 is calculated from the calculated second lengths. ｍａｘ The hunchback detection device according to claim 1 or 2, wherein the hunchback detection device is configured to determine the posture by

5. The estimation unit The hunchback judgment device described in claim 4 learns a dataset including a side image showing the side of a human and the pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra in the subject image of the side image, and uses a trained model that outputs the pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra when the side image is input, to identify the pixel coordinates of the seventh cervical vertebra and the fourth lumbar vertebra in the subject image in the image when the image is input as the side image and output them to the length calculation unit.

6. The hunchback detection device according to claim 1 or 2, further comprising a display control unit that generates a display image showing the determination result determined by the determination unit.

7. a terminal device including a camera and a transmission unit that transmits the image captured by the camera; The hunchback detection device according to claim 1 or 2, A posture detection system equipped with:

8. The longest of the first length L1 of the line segment connecting the seventh cervical vertebra and the fourth lumbar vertebra identified based on an image of the subject to be determined to have a hunched posture taken from the side and the second length L2 from the line segment to the dorsal contour line of the subject image in the image is the maximum second length L2. ｍａｘ a length calculation step for calculating (L2 ｍａｘ a determining step of determining that the subject has a hunched back when the kyphosis index calculated by the calculation formula (L1 / L2) × 100 is equal to or greater than a predetermined reference value, and determining that the subject does not have a hunched back when the kyphosis index is less than the reference value; A method for determining hunchback posture.

9. The longest of the first length L1 of the line segment connecting the seventh cervical vertebra and the fourth lumbar vertebra identified based on an image of the subject to be determined to have a hunched posture taken from the side and the second length L2 from the line segment to the dorsal contour line of the subject image in the image is the maximum second length L2. ｍａｘ a length calculation step for calculating (L2 ｍａｘ a determining step of determining that the subject has a hunched back when the kyphosis index calculated by the calculation formula (L1 / L2) × 100 is equal to or greater than a predetermined reference value, and determining that the subject does not have a hunched back when the kyphosis index is less than the reference value; A hunchback detection program that runs on a computer.

Images