A method, device, electronic device and storage medium for estimating cervical spine rotation

Abstract

The present invention provides a method, device, electronic device and storage medium for estimating the cervical spine rotation degree. This method mainly obtains an 18-joint point map from a photo of the cervical spine rotation posture, and calculates it from the coordinates of the nose, collarbone and left and right eyes on this joint point map, in combination with the cervical spine rotation degree calculation formula. The present invention obtains an 18-joint point map from a photo of the cervical spine rotation posture taken from the front of the tested person, obtains the coordinate relationship of the nose, collarbone, left eye and right eye from the 18 joint points, and designs a maximum rotation degree calculation formula according to the actual corresponding relationship of the body joints of the tested person. According to this formula, the cervical spine rotation angle of the tested person can be quickly obtained. According to the cervical spine rotation degree data of the tested person, it is convenient to monitor the cervical spine health status, which is also beneficial for doctors to diagnose cervical spine diseases, and for rehabilitation therapists to refer to the cervical spine status to massage the cervical spine.

Description

Technical Field

[0001] The present invention belongs to the technical field of visual recognition, and in particular relates to a method, device, electronic equipment and storage medium for estimating cervical vertebra rotation. Background Art

[0002] At present, more and more people are suffering from cervical spondylosis due to long hours of desk work and mobile phone use. The number of patients is increasing, and the age is getting younger. The public is also paying more and more attention to cervical health. The simplest and most effective way to detect cervical health is cervical rotation, but there is no simple and fast way to obtain this data. It often requires experienced people to roughly give the cervical rotation, which is very unfavorable for the promotion and self-examination of cervical health and is often easily overlooked by the public. Therefore, there is an urgent need for a method to quickly and accurately estimate the cervical rotation. Summary of the invention

[0003] The technical problem to be solved by the present invention is that in view of the deficiencies of the above-mentioned prior art, a method for estimating cervical rotation based on AI human posture assessment is provided. The method obtains 18 joint point diagrams from the cervical rotation posture photos of the test subject taken in a standing position from the front, and obtains the coordinate relationship of the nose, clavicle, left eye, and right eye from the 18 joint points. The coordinate relationship with the actual corresponding relationship of the body joints of the test subject is designed, and the maximum rotation degree calculation formula is designed. According to the formula, the rotation angle of the cervical spine of the test subject can be quickly obtained. According to the cervical rotation degree data of the test subject, the health status of the cervical spine can be conveniently monitored, which is also helpful for doctors to diagnose cervical spine conditions, and rehabilitation therapists can refer to their cervical spine status to massage the cervical spine.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for estimating cervical rotation based on AI human posture assessment, characterized in that it includes:

[0005] The angle of cervical rotation of the test subject in standing state is calculated by the formula of cervical rotation degree;

[0006] The calculation formula of the cervical rotation degree θ is:

[0007]

[0008] Where μ = 0.59, Δx represents the distance between the abscissas of the nose and clavicle in the 18 joint point diagram; D 14-15 The distance between the two eyes in the 18 joint points diagram.

[0009] Preferably, the 18 joint point graphs are obtained by the following method:

[0010] First, a posture photo of the test subject's cervical spine rotation in a standing state is obtained, and then 18 joint point diagrams of the test subject are obtained from the posture photo according to the human body posture evaluation algorithm model.

[0011] Preferably, the coordinates of the left eye and the right eye are obtained from the 18 joint point maps. The left eye is joint point 15, and the coordinates are (x 15 , y 15 ); the right eye is joint point 14, and the coordinates are (x 14 , y 14 ), then:

[0012] The present invention also discloses a cervical spine rotation degree estimation device based on AI human body posture evaluation, including: a calculation module for calculating the angle of cervical spine rotation of the tested person in the standing state according to the formula of the cervical spine rotation degree;

[0013] The calculation formula of the cervical spine rotation degree θ:

[0014]

[0015] where μ = 0.59, Δx represents the distance between the abscissas of the nose and the collarbone in the 18 joint point maps; D 14-15 The distance between the two eyes in the 18 joint point maps.

[0016] The present invention also discloses an electronic device, the electronic device includes:

[0017] At least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned method for estimating the cervical spine rotation degree based on AI human body posture evaluation.

[0018] The present invention also discloses a computer-readable storage medium, which stores computer instructions, and the computer instructions are operated to execute the above-mentioned method for estimating the cervical spine rotation degree based on AI human body posture evaluation. Preferably,

[0019] The present invention has the following advantages compared with the prior art:

[0020] 1. The present invention obtains 18 joint point maps from the cervical spine rotation posture photos taken of the tested person standing face-on, obtains the coordinate relationships of the nose, collarbone, left eye, and right eye from the 18 joint points, and the actual corresponding relationships of the body joints of the tested person, designs a maximum rotation degree calculation formula, and can quickly obtain the cervical spine rotation angle of the tested person according to this formula. According to the cervical spine rotation degree data of the tested person, it is convenient to monitor the cervical spine health status, which is also beneficial for doctors to diagnose cervical spine diseases, and for rehabilitation therapists to refer to the cervical spine status to massage the cervical spine.

[0021] The technical solution of the present invention will be further described in detail below with reference to the drawings and embodiments. Brief Description of the Drawings

[0022] Figure 1 This is a diagram showing 18 joint points of the human body of the tester disclosed in Embodiment 1 of the present invention.

[0023] Figure 2 This is a schematic diagram showing the rotation angles of the nose point and collarbone point of the tester in a top-down view disclosed in Embodiment 1 of the present invention.

[0024] Figure 3 This is a scatter plot of the training set data of sinθ and λ disclosed in Embodiment 1 of the present invention.

[0025] Figure 4 This is the code for linear regression training using sklearn in Embodiment 1 of the present invention.

[0026] Figure 5 This is a display of the test results of the cervical spine rotation degree in Embodiment 1 of the present invention. Detailed Description of the Invention

[0027] Embodiment 1

[0028] A method for estimating the cervical spine rotation degree based on AI human pose assessment in an embodiment of the present invention, the method comprising: taking a posture photo of the cervical spine rotation of a tested person standing in a front view, and then obtaining 18 joint point diagrams of the tested person from the posture photo according to an open-source OpenPose human pose assessment algorithm model, obtaining the coordinates of the nose, collarbone, left eye and right eye from the 18 joint point diagrams of the tested person, and calculating Δx and D 14-15 , and then substituting into the calculation formula to solve for the cervical spine rotation degree, as Figure 5 shown.

[0029] In this embodiment, the calculation formula of the cervical spine rotation degree θ is obtained according to the following steps, including:

[0030] S101. Obtain a posture photo of the cervical spine rotation of a sample person, and then obtain 18 joint point diagrams of the sample person from the posture photo according to an open-source OpenPose human pose assessment algorithm model, as Figure 1 shown, so as to obtain the coordinates of the joint points where the nose, collarbone, right eye and left eye are located from the 18 joint point diagrams, Figure 1 where joint point 0 represents the nose, joint point 1 represents the nose, joint point 14 represents the right eye, joint point 15 represents the left eye, the coordinates of joint point 0 are (x0, y0), the coordinates of joint point 1 are (x1, y1), the coordinates of joint point 14 are (x 14 , y 14 ), and the coordinates of joint point 15 are (x 15 , y15 )

[0031] S102. Calculate the distance D between the two eyes in the joint point diagram based on the coordinates of the joint points where the right eye (joint point 14) and the left eye (joint point 15) are located. 14-15 ;

[0032]

[0033] Thus, the distance ratio K of the actual distance of the sample person's body in the joint point diagram can be calculated.

[0034]

[0035] Among them, D actual represents the actual distance between the two eyes of the sample person's body.

[0036] S103. It is known that when the human body does not make a cervical spine rotation movement (i.e., stands normally), the nose joint point 0 and the collarbone joint point 1 are almost on the same vertical line. Therefore, it is determined that the abscissas of the nose joint point 0 and the collarbone joint point 1 are the same, that is, x0 = x1.

[0037] Therefore, the abscissa x1 of the nose joint point 1 is used as the initial point x initial , x initial = x1, as shown in Figure 2 the position of point A in it. When the human cervical spine rotates, the nose joint point 1 rotates from point A to point B. At this time, the abscissa of the nose joint point 1 is x1'. The center point of the human head in the top-down view is Figure 2 the point O in the letter in it. Then the rotation angle ∠BOP after cervical spine rotation is θ, which falls in the triangle △BOP. Then the distance Δx from point B to point P is Δx = |x1 - x1'| = |x0 - x1'|. Then Δx can also represent the distance between the abscissas of the nose and the collarbone in the joint diagram.

[0038]

[0039] Δx actual represents the actual gap between the human nose and the collarbone in the horizontal direction.

[0040] Δx actual = K·Δx;

[0041] Suppose: The actual distance from the center point of the head of the sample person to the nose is the radius r of the cervical spine rotation movement, and the cervical spine rotation angle is θ. Then:

[0042]

[0043] Suppose

[0044] Then:

[0045] S104. Train using a linear regression model Obtain the parameter μ;

[0046] where the loss function is:

[0047]

[0048] where θ' is the true rotation angle of the tester;

[0049] Let This value can be calculated from the coordinates identified by the joint diagram, then:

[0050] L loss = θ' - arcsin(μ·λ);

[0051] Obtain the human cervical spine rotation angle data of 50 sample personnel, where 40 groups are used as the training set and 10 groups are used as the test set:

[0052] λ = [1.49, 0.91, 0.37, 0.75, 1.01, 1.3, 1.32, 1.55, 0.35, 0.25, 1.52, 1.81, 0.54, 1.33, 0.35, 1.65, 1.66, 0.68, 0.3, 1.33, 0.26, 1.22, 0.25, 1.03, 0.9, 1.41, 0.08, 0.32, 1.11, 1.38, 1.34, 1.24, 0.62, 0.65, 1.76, 1.01, 1.24, 1.38, 1.23, 1.17];

[0053] sinθ = [0.74, 0.54, 0.07, 0.39, 0.54, 0.78, 0.64, 0.81, 0.09, 0.0, 0.91, 0.93, 0.29, 0.77, 0.12, 0.93, 0.85, 0.41, 0.12, 0.77, 0.16, 0.67, 0.12, 0.56, 0.42, 0.85, 0.02, 0.1, 0.52, 0.8, 0.68, 0.68, 0.22, 0.36, 0.93, 0.45, 0.66, 0.83, 0.59, 0.64];

[0054] Use sklearn for linear regression training to obtain the weight μ and the bias of the loss function as 0.59 and -0.06 respectively. The code is as Figure 4 shown.

[0055] The above method is used to predict the cervical spine rotation degree, where the tester calculates λ through the coordinates identified by the joint diagram, measures the true cervical spine rotation degree θ of the tester, obtains the true value sinθ, and obtains the predicted cervical spine rotation degree θ' and sinθ' through the above method.

[0056]

[0057]

[0058] As can be seen from the above table, the prediction accuracy of this application can reach 95%, which has met the usage requirements. It has reference value for doctors to diagnose cervical spine conditions and for rehabilitation therapists to perform massage and rehabilitation. If you want to further improve the prediction accuracy, you can supplement the training dataset of linear regression and further train to improve the prediction accuracy.

[0059] Example 2

[0060] This example also discloses a cervical spine rotation degree estimation device based on AI human posture assessment, including: a calculation module for calculating the angle of cervical spine rotation of the tested person in the standing state according to the formula of cervical spine rotation degree;

[0061] The calculation formula of the cervical spine rotation degree θ:

[0062]

[0063] where μ = 0.59, and Δx represents the distance between the abscissas of the nose and the collarbone in the 18-joint point diagram; D 14-15 The distance between the two eyes in the 18-joint point diagram.

[0064] Example 3

[0065] This example also discloses an electronic device, characterized in that the electronic device includes:

[0066] At least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a method for estimating the cervical spine rotation degree based on AI human posture assessment described in Example 1.

[0067] Example 4

[0068] This example also discloses a computer-readable storage medium storing computer instructions, and the computer instructions are operated to execute a method for estimating the cervical spine rotation degree based on AI human posture assessment described in any one of Example 1.

[0069] The above are only the preferred embodiments of the present invention and do not impose any limitations on the present invention. Any simple modifications, changes, and equivalent variations made to the above embodiments based on the technical essence of the invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for estimating cervical rotation based on AI human posture assessment, characterized in that, Including: Calculating the angle of cervical spine rotation of the tested person in the standing state according to the formula of cervical spine rotation degree; The cervical spine rotation degree Calculation formula: ； Among them , represents the distance between the abscissas of the nose and the collarbone in the 18-joint-point diagram; D 14-15 The distance between the two eyes in the 18-joint-point diagram; The 18 joint point maps are obtained by the following method: First, obtaining the posture photo of the cervical spine rotation of the tested person in the standing state, and then obtaining the 18 joint point maps of the tested person from the posture photo according to the human posture evaluation algorithm model.

2. The method for estimating cervical rotation based on AI human posture assessment according to claim 1, characterized in that, Obtain the coordinates of the left-eye joint points and the right-eye joint points from 18 joint-point diagrams. The coordinates of the left-eye joint points are ( x 15 ， y 15 ); the coordinates of the right-eye joint points are ( x 14 ，y 14 ), then: .

3. A device for estimating cervical rotation based on AI human posture assessment, characterized in that, Including: A calculation module, configured to calculate the angle of cervical spine rotation of the tested person in the standing state according to the formula of cervical spine rotation degree; The cervical spine rotation degree The calculation formula is as follows: ； Among them , represents the distance between the abscissas of the nose and the collarbone in the 18-joint-point diagram; D 14-15 The distance between the two eyes in the 18-joint-point diagram; The 18 joint point maps are obtained by the following method: First, obtaining the posture photo of the cervical spine rotation of the tested person in the standing state, and then obtaining the 18 joint point maps of the tested person from the posture photo according to the human posture evaluation algorithm model.

4. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a method for estimating the cervical spine rotation degree based on AI human posture evaluation according to any one of claims 1-2.

5. A computer-readable storage medium storing computer instructions, wherein the computer instructions are operated to execute the method for estimating cervical rotation based on AI human posture assessment according to any one of claims 1-2.

Images