3D Coordinate Estimation Method for Home Training Posture Calibration

Abstract

The present invention relates to a method for estimating 3D coordinates for home training posture correction, characterized by comprising: a shooting step of capturing a user's movement through a 2D camera; a reference coordinate setting step of setting a reference coordinate system from an image captured in the shooting step; a polar coordinate system generation step of calculating an angle difference between the reference coordinates after the reference coordinate setting step; a first angle transformation step of comparing the lengths connected to each coordinate around an axis including the reference coordinates; and a second angle transformation step of returning 3D coordinates according to a Roll Value so that 3D coordinates can be estimated. In addition, according to the present invention, 3D coordinates are estimated from the user's movements captured through a 2D camera, thereby having the effect of correcting the user's posture.

Description

Technology Field

[0001] The present invention relates to a method for estimating 3D coordinates for home training posture correction, and more specifically, to a method for estimating 3D coordinates for home training posture correction that estimates 3D coordinates for specific body parts by analyzing the movements of a user captured by a 2D camera. Background Technology

[0002] Over the past two years, the number of people exercising alone at home—known as "home trainers"—has been rapidly increasing due to COVID-19. However, an industry structure has formed where people simply follow movements from videos exposed through media without considering their own physical capabilities. Consequently, some companies are offering free fitness classes via social media platforms like YouTube and Instagram Live, while others are converting existing membership models into digital subscriptions to provide paid services. According to a survey by LMC, a UK leisure industry media company, the proportion of home trainers in the UK has nearly doubled since the lockdown to 53%, a figure significantly higher than that of gym members (15.6%). Additionally, 63% of respondents stated that exercise is more important than ever since the COVID-19 pandemic, and 67% said that exercise helps with mental health. Therefore, blindly following the various home workout exercises provided without considering individual differences in physical ability or structure, often with incorrect posture or excessive intensity, can lead to injuries ranging from muscle pain to shoulder dislocations and tendon damage.

[0003] As a prior art document for solving these problems, Korean Registered Patent No. 10-2320960 discloses a home training motion guidance and correction system tailored to the user's body.

[0004] However, according to conventional technology, there is a problem in that accurate motion analysis is difficult due to the limitations of 2D cameras. Prior art literature

[0005] (Prior Art 1) Republic of Korea Registered Patent No. 10-2320960 (October 28, 2021) The problem to be solved

[0006] The present invention was devised to solve the problems of the prior art as described above, and its purpose is to enable accurate 3D coordinates for each body part of the user to be estimated through images captured from a 2D camera. means of solving the problem

[0007] A method for estimating 3D coordinates for home training posture correction according to a preferred embodiment of the present invention comprises: a shooting step of capturing a user's movement through a 2D camera; a reference coordinate setting step of setting a reference coordinate system from an image captured in the shooting step; a polar coordinate system generation step of calculating an angle difference between the reference coordinates after the reference coordinate setting step; a first angle transformation step of comparing the lengths connected to each coordinate around an axis including the reference coordinates; and a second angle transformation step of returning 3D coordinates according to a Roll Value so that 3D coordinates can be estimated. Effects of the invention

[0008] By means of the solution to the above problem, the present invention has the effect of allowing 3D coordinates to be estimated from the user's movements captured through a 2D camera so that the user's posture can be corrected. Brief explanation of the drawing

[0009] FIG. 1 is a flowchart of a 3D coordinate estimation method for home training posture correction according to the present invention. FIG. 2 is a schematic diagram of a 3D coordinate estimation method for home training posture correction according to the present invention. FIG. 3 is a configuration diagram of a 3D coordinate estimation method for home training posture correction according to the present invention. Specific details for implementing the invention

[0010] The terms used in this specification will be briefly explained, and the invention will be described in detail.

[0011] The terms used in this invention have been selected based on currently widely used general terms while considering their functions within the invention; however, these terms may vary depending on the intent of those skilled in the art, case law, the emergence of new technologies, etc. Therefore, the terms used in this invention should be defined not merely by their names, but based on their meanings and the overall context of the invention.

[0012] When a part of a specification is described as “comprising” a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0013] Embodiments of the present invention are described below with reference to the attached drawings so that those skilled in the art can easily implement them. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein.

[0014] Specific details regarding the problem to be solved by the present invention, the means for solving the problem, and the effects of the invention are included in the embodiments and drawings described below. The advantages and features of the present invention, and the methods for achieving them, will become clear by referring to the embodiments described below in detail together with the accompanying drawings.

[0015] Hereinafter, the 3D coordinate estimation method for home training posture correction according to the present invention will be described in detail with reference to the attached drawings.

[0016] A method for estimating 3D coordinates for home training posture correction according to a preferred first embodiment of the present invention, with reference to FIGS. 1 to 3, comprises a shooting step (S100) for capturing a user's movement through a 2D camera (11), a reference coordinate setting step (S200) for setting a reference coordinate system from an image captured in the shooting step (S100), a polar coordinate system generation step (S300) for calculating an angle difference between the reference coordinates after the reference coordinate setting step (S100), a first angle transformation step (S400) for comparing the lengths connected to each coordinate around an axis including the reference coordinates, a second angle transformation step (S500) for returning 3D coordinates according to a Roll Value so that 3D coordinates can be estimated, and an estimation learning step (S600) for learning the estimated 3D coordinates through a deep learning model provided with ResNet-152.

[0017] More specifically, the above shooting step (S100) ensures that an image of the user to be the object is captured before estimating 3D coordinates through a 2D video image. At this time, in the above shooting step (S100), a preprocessing process is performed to first recognize the user using a top-down method and to crop the image of the user within the video to estimate the pose within a bounding box. Accordingly, the above shooting step (S100) ensures that the video image is processed in real time so that only the image of the user within the video can be collected.

[0018] Next, the reference coordinate setting step (S200) allows reference coordinates and angles for 3D coordinate estimation to be set. First, 2D coordinates are set for each joint of the user from the 2D image, and bounding boxes are drawn.

[0019] Next, the polar coordinate system generation step (S300) sets two angles corresponding to the coordinates to be extracted and a reference coordinate to generate a polar coordinate system so that 2D coordinates can be converted into 3D coordinates. At this time, an angle is calculated from the reference coordinate and the difference of the angle is calculated (S300). In addition, the polar coordinate system generation step (S300) calculates the difference of the angle calculated from the reference coordinate using the polar coordinate system and the following mathematical formula 1.

[0020] (Mathematical Formula 1)

[0021]

[0023] Next, the first angle transformation step (S400) sets the range of the polar coordinate system using the following mathematical formula 2, based on the user's shoulder being rotated one full turn forward and backward.

[0024] (Mathematical Formula 2)

[0025]

[0026] More specifically, the lengths connected to each coordinate can be compared around the axis having the above reference coordinate (S400). And, when the posture is completely bent or there is no damage to the body part, the length between each coordinate does not change significantly, so the maximum value is applied as π (S400). In addition, the result of calculating the angle difference from the reference coordinate using the above polar coordinate system is applied to the above (Equation 2) to set the range of the polar coordinate system.

[0027] Next, the second angle transformation step (S500) returns (x,y,z) coordinates according to the Roll Value based on the user's shoulder range of motion so that 3D coordinates can be estimated.

[0028] Next, the estimation learning step (S600) allows data for coordinate estimation to be learned through a deep learning model (12) provided with ResNet-152, thereby improving the accuracy of coordinate estimation. That is, data for 3D coordinates estimated from the second angle transformation step (S500) is processed by the estimation learning step (S600) to generate highly reliable 3D coordinate estimation values.

[0029] Accordingly, according to the present invention, 3D coordinates are estimated from the user's movements captured through a 2D camera, thereby having the effect of correcting the user's posture.

[0030] The embodiments described above should be understood as exemplary in all respects and not limiting, and the scope of the invention is defined by the claims set forth below rather than by the detailed description above, and all modifications or variations derived from the meaning and scope of the claims and equivalent concepts thereof should be interpreted as being included within the scope of the invention. Explanation of the symbols

[0031] 10: 3D Coordinate Estimation System 11 : 2D camera 12 : Deep learning model

Claims

Claim 1 A shooting step that captures a user's movement through a 2D camera, first recognizes the user using a top-down method from the captured image, and performs a preprocessing step of cropping the user's image within the video to estimate the pose within a bounding box; a reference coordinate setting step that establishes a reference coordinate system for 3D coordinate estimation from the image captured in the shooting step, establishes 2D coordinates for each joint of the user from the image, and allows a bounding box to be drawn; a polar coordinate system generation step that, after the reference coordinate setting step, establishes two angles corresponding to the coordinates to be extracted and a reference coordinate to generate a polar coordinate system so that 2D coordinates can be converted into 3D coordinates, thereby calculating the angle difference between the reference coordinates; and a first angle transformation step that compares the lengths connected to each coordinate around an axis including the reference coordinates. A method for estimating 3D coordinates for home training posture correction, characterized by including: a second angle transformation step that returns 3D coordinates based on the Roll Value so that 3D coordinates can be estimated. Claim 2 A 3D coordinate estimation method for home training posture correction according to claim 1, wherein the polar coordinate system generation step is characterized by calculating the difference in angle calculated from the reference coordinate using the polar coordinate system and the following mathematical formula 1. (Mathematical Formula 1) Claim 3 A 3D coordinate estimation method for home training posture correction according to claim 1, wherein the first angle transformation step is characterized by setting the range of the polar coordinate system through the following mathematical formula 2 based on the user's shoulder being rotated one full turn forward and backward. (Mathematical Formula 2) Claim 4 A method for estimating 3D coordinates for home training posture correction, characterized in that, in claim 1, it further includes an estimation learning step that enables the 3D coordinates estimated through a deep learning model provided by ResNet-152 to be learned.

Images