Program for manufacturing pillow

Abstract

To provide a program for manufacturing a pillow that reduces a pressure to an ear when a subject lies on his / her side.SOLUTION: In a step 201, a head image of a subject in a standing position is input. In a step 202, the head image in the standing position of the subject is used as input teacher data, and a right ear image and a left ear image of the subject are estimated using the learned learning control model 18 and extracted together with the positions thereof. In a step 203, transverse cross-sectional shapes of a right through-hole 15 and a left through-hole 16 of the pillow are estimated and extracted using the learned learning control model 19 with the estimated and extracted right ear image and left ear image of the subject as input teacher data. In a step 204, a distance between the estimated and extracted right ear image and left ear image is calculated. In a step 205, the pillow is designed by using other data such as a depth of the through-holes, pillow size data, and a filling member, on the basis of the estimated and extracted cross-sectional shapes of the right side through-hole 15 and the left side through-hole 16 of the pillow and the calculated distance between the right ear image and the left ear image.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a program for manufacturing a pillow for relieving pressure on the ears when a subject is in a recumbent position. [Background technology]

[0002] The applicant of the present application has proposed a pillow shown in Figs. 4 to 8 as a pillow for reducing pressure on the ears when a subject is in a recumbent position (see Patent Document 1).

[0003] 4 is an upper perspective view of the pillow, FIG. 5 is a lower perspective view of FIG. 4, FIG. 6 is a top view of FIG. 4, FIG. 7 is a front view of FIG. 4, and FIG. 8 is a cross-sectional view taken along line VIII-VIII of FIG.

[0004] As shown in Fig. 4 to Fig. 8, the pillow is a two-layered pillow, and is composed of an upper pillow unit 101 and a lower pillow unit 102 provided on a part of the lower surface of the upper pillow unit 101. The upper pillow unit 101 is a sewn product with a hollow space formed by filling a pipe, polyester cotton, etc., and the lower pillow unit 102 is an integrally molded product made of urethane or the like.

[0005] The upper pillow unit 101 has an occipital support region 111, a neck support region 112 adjacent to the front side of the occipital support region 111, and a right side temporal support region 113 and a left side temporal support region 114 adjacent to the right and left sides of the occipital support region 111 and the neck support region 112.

[0006] As shown in Fig. 4, Fig. 6, and Fig. 8, in the upper pillow unit 101, a cylindrical right through-hole 115 is provided in the right side temporal support region 113, while a cylindrical left through-hole 116 is provided in the left side temporal support region 114. Also, as shown in Fig. 8, the bottom 115a of the right through-hole 115 and the bottom 116a of the left through-hole 116 are blocked by the lower pillow unit 102. The cross-sectional shapes, for example, diameters, of the right through-hole 115 and the left through-hole 116 are determined to fit the right and left ears of the subject, thereby making it easier for the subject to turn from the supine position to the lateral position. In addition, at this time, the right through-hole 115 and the left through-hole 116 are formed by sewing, and the depths of the through-holes 115 and 116, i.e., the thickness of the upper pillow unit 101, are determined so that the pressure on the ears is reduced and the reverberation of sound is reduced.

[0007] 5 and 7, the front region of the right support region 113 of the upper pillow unit 101 and the front region of the left support region 114 of the upper pillow unit 101 protrude forward from the lower pillow unit 2. In other words, the lower pillow unit 102 is smaller than the upper pillow unit 101. This makes it easier for the subject to turn over from the supine position to the lateral position.

[0008] The subject's movement from supine to lateral position will be described in detail. When the subject turns from supine to lateral position, the subject's ear fits into the right through-hole 115 or the left through-hole 116, and the subject's shoulder fits under the right side temporal support area 113 or the left side temporal support area 114 of the upper pillow unit 101 where the lower pillow unit 102 is not present. As a result, the subject can smoothly turn from supine to lateral position.

[0009] 4, 6, and 8, the occipital support region 111 of the upper pillow unit 101 is partitioned by partition cloths (gussets) 111a and 111b. In this case, the right through-hole 115 and the left through-hole 116 are formed almost vertically, so that the occipital support region 111 is not affected by the right through-hole 115 and the left through-hole 116. Therefore, the partition cloths 111a and 111b stably maintain the subject in a supine position due to the bowl-shaped cross-sectional shape of the filling member.

[0010] Further, the through hole described above is a cylinder having a circular cross section, but it may be a cylinder having any other cross section, for example, a polygonal cross section such as a square or triangular cross section. [Prior art documents] [Patent documents]

[0011] [Patent Document 1] Utility model registration application No. 2023-001877 Summary of the Invention [Problem to be solved by the invention]

[0012] However, when manufacturing the pillow proposed above, basically, posture information of the subject in a recumbent position is required, so the subject may be located remotely, which poses the problem of high manufacturing costs for the pillow. [Means for solving the problem]

[0013] In order to solve the above-mentioned problems, the pillow manufacturing program of the present invention is a pillow including an upper pillow unit and a lower pillow unit provided on the underside of the upper pillow unit, the upper pillow unit having an occipital support region, a neck support region adjacent to the front side of the occipital support region, and a right side temporal support region and a left side temporal support region adjacent to the right side and left side of the occipital support region and the neck support region, a right side through hole and a left side through hole are provided in the right side temporal support region and the left side temporal support region of the upper pillow unit, respectively, and the bottom of the right side through hole and the bottom of the left side through hole are blocked by the lower pillow unit, and the pillow manufacturing program uses an image of a human head in a standing posture as input teacher data and an image of a human right ear and a left ear as output teacher data. The present invention is for causing a computer to execute a first extraction processing procedure in which a first learning model that has been trained using a human right ear image and left ear image is used as input data to estimate and extract the right ear image and left ear image of the subject, a second extraction processing procedure in which a second learning model that has been trained using a human right ear image and left ear image as input teacher data and the cross-sectional shapes of the pillow's right through hole and left through hole as output teacher data is used to estimate and extract the cross-sectional shapes of the pillow's right through hole and left through hole as input data to estimate and extract the cross-sectional shapes of the pillow's right through hole and left through hole, and a design procedure in which a pillow is designed using the cross-sectional shapes of the right through hole and the cross-sectional shapes of the left through hole that have been estimated and extracted.

[0014] Further, a pillow manufacturing program according to the present invention is a manufacturing program for a pillow which is a sewn product having an occipital support region, a neck support region adjacent to the front side of the occipital support region, and right and left temporal support regions adjacent to the right and left sides of the occipital support region and the neck support region, a right recess and a left recess respectively provided in the right and left temporal support regions, the bottoms of the right recess and the left recess are closed, and which has a built-in filling member, the manufacturing program for the pillow using a first learning model which has been trained using an image of a human head in a standing position as input teacher data and images of a human right ear and left ear as output teacher data, and inputting an image of the head of a subject in a standing position. The method causes a computer to execute a first extraction processing procedure of estimating and extracting the right ear image and the left ear image of the subject as force data, a second extraction processing procedure of estimating and extracting the cross-sectional shapes of the right recess and the left recess of the pillow using a second learning model that has been trained using a human right ear image and a left ear image as input teacher data and the cross-sectional shapes of the right recess and the left recess of the pillow as output teacher data, and the estimated extracted right ear image and the estimated extracted left ear image of the subject as input data, and a design procedure of designing a pillow using the estimated extracted cross-sectional shape of the right recess and the estimated extracted cross-sectional shape of the left recess. Effect of the Invention

[0015] According to the present invention, the through holes or recesses for the ears of the pillow are estimated and extracted by learning control using an image of the back of the head of the subject in a standing position, thereby reducing manufacturing costs. [Brief description of the drawings]

[0016] [Figure 1] FIG. 2 is a block circuit diagram showing a computer for executing an embodiment of a pillow manufacturing program according to the present invention. [Diagram 2] 2 is a flowchart showing a pillow manufacturing program (routine) executed by the computer of FIG. 1. [Diagram 3] 13 is a flowchart showing another example of a pillow manufacturing program (routine) executed by the computer of FIG. [Figure 4]FIG. 1 is a top perspective view showing the pillow proposed earlier. [Diagram 5] FIG. 5 is a bottom perspective view of FIG. [Figure 6] FIG. 5 is a top view of FIG. [Figure 7] FIG. 5 is a front view of FIG. [Figure 8] FIG. 8 is a cross-sectional view taken along line VIII-VIII in FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] FIG. 1 is a block circuit diagram showing a computer for executing an embodiment of a pillow manufacturing program according to the present invention.

[0018] 1, computer 1 is made up of a central processing unit (CPU) 11, a read-on memory (ROM) 12 for storing programs and the like, a random access memory (RAM) 13 for storing temporary data and the like, a flash memory 14 for storing images and the like, an input / output (I / O) interface 14 connected to a USB (registered trademark) memory or the like, an image interface 16 connected to a camera 2, a communication interface 17 connected to the Internet 2, learning control models 18, 19, etc. An operator can access computer 1 using an information terminal 4, for example a smartphone, connected via the Internet 3.

[0019] The learning control model 18 uses multiple images of the right ear and multiple images of the left ear of a person in a standing posture, for example, an image seen from the front, an image seen from the rear, an image seen from the right lateral direction, and an image seen from the left lateral direction, as input teacher data, and is trained by deep learning (machine learning) using left ear information and right ear information as output teacher data. This learning can be performed in advance outside the computer 1, and the operator can input the learning results to the learning control model 18 using the information terminal 4. However, learning may also be performed within the computer 1. The estimation and extraction of the right ear image and the left ear image using this learning control model 18 acts as a preprocessing for the learning control model 19, and is useful for improving the estimation accuracy.

[0020] The learning control model 19 learns by deep learning (machine learning) using the right ear image and the left ear image of a person as input teacher data, and the cross-sectional shape of the right through-hole and the left through-hole of the actual pillow recognized as a pillow shown in Figs. 4 to 8 used by the person as having less pressure on the ears and less sound echo as output teacher data. This learning is also performed in advance outside the computer 1, and the learning result may be stalled in the learning control model 19 by the operator using the information terminal 4. However, learning may also be performed within the computer 1. In this case, the cross-sectional shape of the right through-hole and the left through-hole is, for example, a circle when the right through-hole and the left through-hole are cylindrical, but all cylindrical shapes of cross sections are targeted. In this case, there is a correlation between the cross-sectional shape of the right ear image (left ear image) and the right through-hole (left through-hole). However, the shape of the ear image does not necessarily match the cross-sectional shape of the through-hole. From this point of view, it is meaningful to use the cross-sectional shape of the through-hole actually recognized by a person as the output teacher data.

[0021] FIG. 2 is a flow chart showing a manufacturing program (manufacturing routine) executed by the computer 1 of FIG.

[0022] First, in step 201, head images of the subject in a standing posture, such as an image of the back of the head, an image of the right face, and an image of the left face, are input. For example, images stored in advance in a USB memory or the like by the subject, images taken by a camera 2, and images taken by the subject or the like using a smartphone and sent via the Internet 3 are input. Therefore, images can be input even if the subject is located remotely.

[0023] Next, in step 202, the subject's right ear image and left ear image are estimated and extracted together with their positions using the trained learning control model 18 with the subject's head image in a standing posture as input teacher data.

[0024] Next, in step 203, the estimated extracted right ear image and left ear image of the subject are used as input teacher data to estimate and extract the cross-sectional shapes of the right side through hole 15 and the left side through hole 16 of the pillow shown in Figures 4 to 8 using a trained learning control model 19.

[0025] Next, in step 204, the distance between the right ear image and the left ear image estimated and extracted in step 202 is calculated. For example, the distance between the right ear image and the left ear image is calculated by multiplying the sum of the distance between the right ear image and the left ear image in the occipital image, the distance between the right ear image and the occipital edge in the right face image, and the distance between the left ear image and the occipital edge in the left face image by a predetermined coefficient.

[0026] Next, in step 205, the pillow is designed based on the estimated cross-sectional shapes of the right side through-hole 15 and the left side through-hole 16 of the pillow and the calculated distance between the right ear image and the left ear image, as well as other data such as the depth of the through-holes, pillow size data, and filling materials.

[0027] The manufacturing routine of FIG.

[0028] In this way, the pillow is manufactured by learning control using an image of the subject's head in an upright position, so that information on the recumbent position is not required, and manufacturing costs can be reduced.

[0029] The pillow described above is a two-layer pillow, and is composed of an upper pillow unit 101 made of a sewn material and a lower pillow unit 102 made of a molded product, but it may be a one-layer pillow made of a sewn material with beads, pipes, polyester cotton, etc. as filling materials. In this case, the through hole becomes a recess.

[0030] For a one-layer pillow, the learning control model 19 learns by deep learning (machine learning) using the right ear image and the left ear image of a person as input teacher data, and the cross-sectional shapes of the right and left recesses of an actual pillow recognized as a one-layer pillow used by the person with less pressure on the ears and less sound reverberation as output teacher data. In this case, the cross-sectional shapes of the right and left recesses are, for example, circles when the right and left recesses are cylindrical.

[0031] For a one-layer structure pillow, the pillow is manufactured using the flow chart shown in Fig. 3 instead of the flow chart in Fig. 2. That is, in Fig. 3, steps 203' and 205' are provided instead of steps 203 and 205 in Fig. 2.

[0032] In step 203', the cross-sectional shapes of the right and left recesses of the single-layer pillow are estimated and extracted using the trained learning control model 19 with the estimated and extracted right and left ear images of the subject as input teacher data.

[0033] In step 205', the pillow is designed using other data based on the cross-sectional shapes of the right and left recesses of the one-layer pillow estimated and extracted and the calculated distance between the right and left ear images.

[0034] In this way, the routine of FIG. 3 also manufactures a pillow through learning control using an image of the subject's head in an upright position, eliminating the need for supine position information and reducing manufacturing costs.

[0035] In the above embodiment, the learning control model 18 is provided as a pre-processing for the learning control model 19, but the learning control model 19 alone can be used. That is, in this case, the learning control model 19 learns by deep learning (machine learning) using a large number of images of the right ear and a large number of images of the left ear in a standing posture of a person, for example, an image seen from the front, an image seen from the back, an image seen from the right lateral direction, and an image seen from the left lateral direction, as input teacher data, and the cross-sectional shapes of the right through-hole (or recess) and the left through-hole (or recess) of an actual pillow recognized as a pillow used by the person with less pressure on the ear and less sound echo as output teacher data. As a result, the cross-sectional shapes of the right through-hole (recess) and the left through-hole (recess) of the pillow can be directly estimated and extracted using the only learning control model 19 that has been learned using the image of the subject's head in a standing posture as input teacher data. However, the estimation accuracy is lower than when two learning control models are provided.

[0036] The present invention can also be expanded to the following pillow manufacturing program. A pillow manufacturing program that causes a computer to execute an extraction processing procedure that uses a trained learning model in which an image of a human head in a standing position is used as input training data and the shape of a pillow is used as output training data, and an image of the subject's head in a standing position is used as input data to estimate and extract the shape of the subject's pillow.

[0037] Furthermore, the present invention can be applied to any modifications within the obvious scope of the above-described embodiments. [Explanation of symbols]

[0038] 1: Computer 2: Camera 3: Internet 4: Information terminal 11: CPU 12:ROM 13: RAM 14: Flash memory 15: Input / Output Interface 16: Image interface 17: Communication interface 18, 19: Learning control model 101: Upper pillow unit 102: Lower pillow unit 111: Occipital support area 112: Neck support area 113: Right side support area 114:Left side support area 115: Right side through hole 115a: Bottom 116: Left side through hole 116a: Bottom

Claims

1. A pillow comprising an upper pillow unit and a lower pillow unit provided on the lower surface of the upper pillow unit, The upper pillow unit has an occipital support region, a neck support region adjacent to the front side of the occipital support region, and a right side temporal support region and a left side temporal support region adjacent to the right side and the left side of the occipital support region and the neck support region, A manufacturing program for a pillow in which a right through-hole and a left through-hole are provided in the right side temporal support region and the left side temporal support region of the upper pillow unit, respectively, and a bottom of the right through-hole and a bottom of the left through-hole are blocked by the lower pillow unit, a first extraction processing procedure for estimating and extracting the right ear image and the left ear image of a subject using a first learning model that has been trained using a head image of a human in a standing posture as input teacher data and right ear images and left ear images of the human as output teacher data, and using an image of the head of the subject in a standing posture as input data; a second extraction processing procedure for estimating and extracting the cross-sectional shapes of the right-side through-hole and the left-side through-hole of the pillow using a second learning model that has been trained using the right-side ear image and the left-side ear image of the person as input teacher data and the cross-sectional shapes of the right-side through-hole and the left-side through-hole of the pillow as output teacher data, and using the estimated extracted right-side ear image and the estimated extracted left-side ear image of the subject as input data; a design procedure for designing the pillow using the cross-sectional shape of the estimated and extracted right-side through hole and the cross-sectional shape of the estimated and extracted left-side through hole; A pillow manufacturing program for causing a computer to execute the above.

2. 2. A pillow manufacturing program according to claim 1, wherein the head images are a right side face image, a left side face image and an occipital head image of the subject.

3. Further, the method includes a procedure of calculating a distance between the estimated right ear image and the estimated left ear image of the subject, The method for manufacturing a pillow according to claim 1 , wherein the design procedure designs the pillow using the cross-sectional shape of the estimated extracted right-side through hole and the cross-sectional shape of the estimated extracted left-side through hole, and the calculated distance.

4. The head support device has an occipital support region, a neck support region adjacent to the front side of the occipital support region, and a right side temporal support region and a left side temporal support region adjacent to the right side and the left side of the occipital support region and the neck support region, a right recess and a left recess are provided in the right side temporal support region and the left side temporal support region, respectively, and a bottom of the right recess and a bottom of the left recess are closed; A manufacturing program for a pillow, which is a sewn product incorporating a filling member, a first extraction processing procedure for estimating and extracting the right ear image and the left ear image of a subject using a first learning model that has been trained using a head image of a human in a standing posture as input teacher data and right ear images and left ear images of the human as output teacher data, and using an image of the head of the subject in a standing posture as input data; a second extraction processing procedure for estimating and extracting the cross-sectional shapes of the right recess and the left recess of the pillow using a second learning model that has been trained using the right ear image and the left ear image of the person as input teacher data and the cross-sectional shapes of the right recess and the left recess of the pillow as output teacher data, and using the estimated extracted right ear image and the estimated extracted left ear image of the subject as input data; a design procedure for designing the pillow using the estimated and extracted cross-sectional shape of the right recess and the estimated and extracted cross-sectional shape of the left recess; A pillow manufacturing program for causing a computer to execute the above.

5. 5. A pillow manufacturing program according to claim 4, wherein the head images are a right side face image, a left side face image and an occipital head image of the subject.

6. Further, the method includes a procedure of calculating a distance between the estimated right ear image and the estimated left ear image of the subject, The method for manufacturing a pillow according to claim 4, wherein the design procedure includes designing the pillow using the estimated cross-sectional shape of the right recess and the estimated cross-sectional shape of the left recess, and the calculated distance.

7. A pillow comprising an upper pillow unit and a lower pillow unit provided on the lower surface of the upper pillow unit, The upper pillow unit has an occipital support region, a neck support region adjacent to the front side of the occipital support region, and a right side temporal support region and a left side temporal support region adjacent to the right side and the left side of the occipital support region and the neck support region, A manufacturing program for a pillow in which a right through-hole and a left through-hole are provided in the right side temporal support region and the left side temporal support region of the upper pillow unit, respectively, and a bottom of the right through-hole and a bottom of the left through-hole are blocked by the lower pillow unit, an extraction processing procedure for estimating and extracting the cross-sectional shapes of the right side through-hole and the left side through-hole of the pillow using a trained learning model with a head image of a human in a standing posture as input teacher data and a cross-sectional shape of the right side through-hole and the left side through-hole of the pillow as output teacher data, and using an image of the head of a subject in a standing posture as input data; a design procedure for designing the pillow using the cross-sectional shape of the estimated and extracted right-side through hole and the cross-sectional shape of the estimated and extracted left-side through hole; A pillow manufacturing program for causing a computer to execute the above.

8. The head support device has an occipital support region, a neck support region adjacent to the front side of the occipital support region, and a right side temporal support region and a left side temporal support region adjacent to the right side and the left side of the occipital support region and the neck support region, a right recess and a left recess are provided in the right side temporal support region and the left side temporal support region, respectively, and a bottom of the right recess and a bottom of the left recess are closed; A manufacturing program for a pillow, which is a sewn product incorporating a filling member, an extraction processing procedure for estimating and extracting the cross-sectional shapes of the right recess and the left recess of the pillow using a trained learning model with a head image of a human in a standing posture as input teacher data and a cross-sectional shape of the right recess and the left recess of the pillow as output teacher data, and using an image of the head of the subject in a standing posture as input data; a design procedure for designing the pillow using the estimated and extracted cross-sectional shape of the right recess and the estimated and extracted cross-sectional shape of the left recess; A pillow manufacturing program for causing a computer to execute the above.

9. An extraction process procedure for estimating and extracting the shape of a pillow of a subject using a trained learning model with a head image of a human in a standing posture as input training data and a pillow shape as output training data, and using a head image of the subject in a standing posture as input data. A pillow manufacturing program for causing a computer to execute the above.

Images