Posture monitoring and training method and system

Abstract

This application relates to the field of physical training monitoring technology and provides a posture monitoring and training method. This posture monitoring and training method is used to identify and monitor the posture of a supine crunch exercise, including: acquiring real-time pressure distribution data and arm posture data of the trainee during the supine crunch exercise; based on the pressure distribution data and arm posture data, extracting the pressure center, arm angle, and back-to-ground angle of the trainee during the current movement, wherein the back-to-ground angle is the maximum ground-to-ground angle during the current movement; determining whether the current movement is erroneous based on the pressure center, arm angle, and back-to-ground angle, and if so, determining the type of violation and outputting a prompt message; and determining and outputting the valid number of repetitions performed by the trainee after completing the supine crunch exercise. This application can reduce the false judgment rate of supine crunch exercise compliance judgment and improve the accuracy of monitoring.

Description

Technical Field

[0001] This application relates to the field of physical training monitoring technology, and in particular to a posture monitoring and training method and system. Background Technology

[0002] The supine crunch is a core exercise that uses abdominal muscles to lift the upper back off the ground. Its effectiveness is highly dependent on proper form. In actual training, trainees often exhibit improper form due to fatigue or compensation, such as bending their elbows or extending their shoulders forward, leading to decreased training effectiveness or even sports injuries.

[0003] Many related technologies rely on pressure changes to determine the range of motion and identify non-standard movements. However, these monitoring methods cannot simultaneously identify spatial posture violations such as elbow flexion, making it difficult to effectively detect hidden violations such as malformed movements even when the range of motion meets the standard. This results in a high probability of missed detections in these technologies. Summary of the Invention

[0004] In view of this, this application aims to propose a posture monitoring and training method and system to reduce the probability of misjudgment and improve monitoring accuracy.

[0005] To achieve the above objectives, the technical solution of this application is implemented as follows:

[0006] The first aspect of this application provides a posture monitoring and training method for identifying and monitoring the posture of a supine crunch movement, wherein the posture monitoring and training method includes:

[0007] The real-time pressure distribution data and arm posture data of the trainee during the supine crunch exercise are obtained. The pressure distribution data is the pressure data between the trainee and the training mat. The supine crunch exercise includes the movement process of multiple supine crunch movements.

[0008] Based on the pressure distribution data and the arm posture data, the pressure center, arm angle and back-off-ground angle of the trainee during the current movement are extracted, wherein the back-off-ground angle is the maximum off-ground angle during the current movement.

[0009] Based on the pressure center, arm angle, and back-off-ground angle of the current action process, determine whether the current action process is in violation of the rules, and if it is in violation, determine the type of violation and output a prompt message;

[0010] Once the trainer completes the supine crunch exercise, determine and output the effective number of repetitions performed by the trainer.

[0011] In some embodiments, determining whether the current movement is in violation of regulations based on the center of pressure, arm angle, and back-off-ground angle of the current movement includes:

[0012] If the distance between the pressure center and the preset baseline exceeds the preset distance value, the current action process is determined to be in violation, and the violation type is shoulder movement violation;

[0013] If the arm angle is less than a preset arm angle threshold and the duration reaches a preset duration, the current action process is determined to be in violation, and the violation type is elbow flexion violation.

[0014] If the back angle off the ground is less than the preset ground angle threshold, the current action is determined to be in violation, and the violation type is upper body forward flexion failure.

[0015] In some embodiments, the attitude monitoring and training method further includes:

[0016] Determine the start time of the current action process;

[0017] Determine whether the current action process has ended;

[0018] When the current movement process ends, find the fingertip pad pressure on the starting line and the ending line of the current movement process from the pressure distribution data. The starting line and the ending line are the positions that the trainee's hands should touch in sequence during the supine crunch movement.

[0019] Based on the fingertip pressure on the starting line, determine whether the starting moment is a case where both sides of the starting line are effective.

[0020] Based on the fingertip pressure on the termination line, determine whether the termination line is effectively touched on both sides at the end time.

[0021] If the start time is not within the case where both sides of the start line are valid, and / or the end time is not within the case where both sides of the end line are valid, then it is determined that there is a violation in the current action process.

[0022] In some embodiments, the attitude monitoring and training method further includes:

[0023] Acquire the trainer's initial image and initial hand touch pressure sensing data at a preset starting line on the training mat;

[0024] The pressure center point of the fingertip pad of the starting line is determined based on the initial hand touch pressure sensing data.

[0025] Based on the trainer's initial image, the direction of the trainer's arm movement is determined, and the position of the pressure center point of the fingertip touch pad at the termination line is predicted according to the direction of the arm movement.

[0026] Based on the trainer's initial image, the trainer's body shape characteristics are analyzed, and the normal sensing width of the trainer's start and end lines is determined according to the body shape characteristics. The normal sensing width of the start and end lines is the reasonable fluctuation range of the hand position after considering the current differences in the trainer's body shape.

[0027] Based on the normal sensing width of the start and end lines, and the pressure center point position of the start line and the pressure center point position of the end line, the trainer's start and end line sensing range is determined.

[0028] The step of finding the fingertip pad pressure on the start and end lines of the current action process from the pressure distribution data includes:

[0029] From the pressure distribution data, find the fingertip pressure on the pad within the starting and ending line sensing range during the trainee's movement.

[0030] In some embodiments, the attitude monitoring and training method further includes:

[0031] When the trainer begins to perform supine crunches, during the preset correction time period after the start of the exercise, the trainer's exercise video data is acquired, and the trainer's hand position data at the starting line and the ending line is analyzed based on the exercise video data.

[0032] Based on the newly acquired hand position data, the normal sensing width of the trainer's start and end lines is corrected, and based on the corrected normal sensing width of the start and end lines, as well as the pressure center point position of the start line and the pressure center point position of the end line, the trainer's start and end line sensing range is redefined.

[0033] In some embodiments, the attitude monitoring and training method further includes:

[0034] If the current action is determined to be a violation, the trainer's motion image data is acquired;

[0035] Extract the torso baseline from the motion image data, wherein the torso baseline is the midline of the trainee's torso longitudinal axis;

[0036] Calculate the relative offset between the trunk baseline and the initial trunk baseline, which is the trunk baseline of the trainee during the initial movement phase;

[0037] Based on the relative offset, determine the position compensation values ​​corresponding to the starting line and the ending line respectively;

[0038] Based on the position compensation value, the start and end line sensing intervals used in subsequent movements are adjusted so that the relative relationship between the adjusted start and end line sensing intervals and the trainee's current torso position is consistent with the relative relationship between the start and end line sensing intervals and the initial torso baseline in the initial movement phase.

[0039] After adjusting the start and end line sensing range, the fingertip pressure of the adjusted start and end line sensing range is detected and the effectiveness of the action is judged for subsequent actions.

[0040] In some embodiments, the attitude monitoring and training method further includes:

[0041] If a fault is detected in the starting line sensor and / or the ending line sensor, the auxiliary sensor, which is set parallel to the surface of the training mat, is controlled to detect the contact state between the trainee's hand and the ground.

[0042] Based on the acquired video data of the trainer's movements, the relative positional relationship between the trainer's hands and the starting and ending lines is analyzed.

[0043] Based on the contact state between the hand and the ground detected by the auxiliary sensor, and the relative positional relationship, it is determined whether the current action violates the start and end line rules.

[0044] In some embodiments, the attitude monitoring and training method further includes:

[0045] Identify the missing time periods where the completion of actions was not accurately recorded due to sensor malfunction;

[0046] Extract the video data within the missing time period from the trainer's motion video data;

[0047] Based on the video data within the missing time period, the trainer's fingertip movement trajectory is extracted from the video data. Based on the fingertip movement trajectory, it is analyzed whether the trainer's actions within the missing time period violate the rules by touching the start and end lines, so as to determine the number of valid actions within the missing time period.

[0048] A second aspect of this application provides a posture monitoring and training system for identifying and monitoring the posture of a supine crunch movement. The posture monitoring and training system includes:

[0049] The training mat body is equipped with a pressure sensor array, which can detect the pressure data between the trainee and the training mat body and obtain pressure distribution data.

[0050] The smart armband is used to monitor the bending angle of the trainee's arm and obtain arm posture data;

[0051] The posture monitoring device is electrically connected to the pressure sensor inside the training pad and to the smart arm ring.

[0052] The posture monitoring device is used to receive real-time pressure distribution data and arm posture data of the trainee during the supine crunch exercise, wherein the supine crunch exercise includes the movement process of multiple supine crunch movements.

[0053] Based on the pressure distribution data and the arm posture data, the pressure center, arm angle and back-off-ground angle of the trainee during the current movement are extracted, wherein the back-off-ground angle is the maximum off-ground angle during the current movement.

[0054] Based on the pressure center, arm angle, and back-off-ground angle of the current action process, determine whether the current action process is in violation of the rules, and if it is in violation, determine the type of violation and output a prompt message;

[0055] Once the trainer completes the supine crunch exercise, determine and output the effective number of repetitions performed by the trainer.

[0056] A third aspect of this application provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the above-described attitude monitoring and training method.

[0057] Compared with related technologies, this application has at least the following technical effects:

[0058] This application employs a multimodal fusion measurement method combining pressure sensing and arm posture measurement, which can simultaneously capture pressure changes and spatial posture. This allows for the identification of violations in spatial posture, thereby improving the accuracy of action validity recognition and reducing the probability of misjudging hidden violations such as "amplitude meets the standard but the action is deformed".

[0059] Furthermore, this application can also replace manual monitoring, achieve fully automatic judgment, improve monitoring efficiency, and unify judgment standards, making it easier to meet the needs of large-scale training and assessment. Attached Figure Description

[0060] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0061] Figure 1 This is a schematic diagram of the method flow for the attitude monitoring and training method described in the embodiments of this application;

[0062] Figure 2 This is a schematic diagram of the process for detecting whether the touch line is compliant in the posture monitoring and training method described in the embodiments of this application;

[0063] Figure 3 This is a flowchart illustrating the process of determining the sensing intervals of the starting and ending lines in the attitude monitoring and training method described in this application embodiment.

[0064] Figure 4 This is a schematic diagram of the process for adjusting the sensing range in the attitude monitoring and training method described in the embodiments of this application;

[0065] Figure 5 This is a schematic diagram illustrating the process of determining the effectiveness of an action in the posture monitoring and training method described in this application when the start line sensor and the end line sensor fail to monitor pressure normally. Detailed Implementation

[0066] To make the technical solution and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0067] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0068] Furthermore, it should be noted that in the description of this application, if terms such as "upper," "lower," "inner," or "outer" appear, indicating orientation or positional relationship, these are based on the orientation or positional relationship shown in the accompanying drawings and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on this application. In addition, if terms such as "first" or "second" appear, they are also used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0069] Furthermore, in the description of this application, unless otherwise expressly defined, the terms "installation," "connection," "joining," and "connector" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application in light of the specific circumstances.

[0070] In this application, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0071] The present application will now be described in detail through exemplary embodiments. However, it should be understood that, without further description, elements, structures, and features in one embodiment may be advantageously incorporated into other embodiments.

[0072] An embodiment of the first aspect of this application provides a posture monitoring and training method. This method is used to identify the effective number of movements of a trainer during a supine crunch exercise. Furthermore, by identifying spatial posture violations such as elbow flexion, the effective number of movements is jointly identified and determined, thereby improving the accuracy of effective movement number identification and monitoring.

[0073] As a fundamental exercise for core strength training, the supine crunch has specific requirements for standard form. Specifically, the trainee should maintain a supine position with legs bent and feet flat on the ground. Initially, the back and shoulders should be fully in contact with the training mat, and the arms should be extended at the sides. By contracting the abdominal muscles, the upper back should be lifted off the ground to a preset angle (usually no more than 30°). At the same time, the hands should touch the preset starting and ending lines on the training mat in sequence. During the movement, it is important to avoid problems such as excessive forward movement of the shoulders, bending of the arms, and insufficient forward bending of the upper body.

[0074] However, the relevant technologies have significant limitations in monitoring supine crunch movements. Specific monitoring methods include the following three:

[0075] (1) Technologies that mainly rely on a single monitoring method, such as judging the range of motion by detecting pressure changes through a pressure pad. This method can only roughly sense whether the back is off the ground, but cannot identify spatial posture violations such as elbow flexion and shoulder movement. When the trainee flexes their elbows ("arm angle less than 180° and duration exceeding the threshold") or moves their shoulders ("pressure center deviates from the preset baseline by more than the limit"), the single pressure sensing solution cannot capture these hidden violations, resulting in invalid movements that are "at the required range but deformed" being misjudged as valid, leading to a high rate of missed or misjudged movements.

[0076] (2) Visual recognition scheme, which is a scheme that uses video data for recognition and analysis. Although this scheme can make up for the shortcomings of spatial posture monitoring to a certain extent, it is significantly affected by environmental factors such as lighting conditions and clothing obstruction. The accuracy rate is low when there is strong light, low light or when the trainee is wearing heavy training clothes.

[0077] (3) Manual monitoring scheme. Manual monitoring is not only inefficient (a single examiner can only monitor one person per minute), but also subject to subjective judgment bias, making it difficult to meet the standardization requirements of large-scale training and assessment.

[0078] Therefore, to address the issue of low accuracy in recognizing the number of effective movements in related technologies, this application provides a posture monitoring and training method for identifying and monitoring the posture of a supine crunch. This method is applied to a posture monitoring and training system, which includes a training mat, a smart armband, and a posture monitoring device.

[0079] The training mat contains an array of pressure sensors that detect the pressure between the trainee and the mat, providing pressure distribution data. This array (with a spacing of 3-5 cm) collects pressure distribution data between the trainee and the mat surface at a frequency of 100 Hz (or other values, not limited here). The array covers the shoulders, hips, and the start and end lines, capturing information such as shoulder pressure center shift and fingertip pressure, thus obtaining pressure distribution data.

[0080] The smart armband is used to monitor the trainee's arm flexion angle and obtain arm posture data. The smart armband adopts a detachable design and is equipped with a nine-axis sensor (three-axis accelerometer + gyroscope + magnetometer). It can collect arm motion data, including inertial parameters such as acceleration and angular velocity, at a frequency of 200Hz (or other values, which are not limited here), to obtain arm posture data.

[0081] The pressure sensor array and smart armband synchronously transmit the detected pressure distribution data and arm posture data to the posture monitoring device, which then monitors the effectiveness of the movement. The pressure sensor array and smart armband can transmit data synchronously via Bluetooth 5.0 (latency <50ms), ensuring consistency in the time dimension.

[0082] The attitude monitoring device then executes this attitude monitoring and training method, referring to... Figure 1 The posture monitoring and training method specifically includes the following steps S110-S140.

[0083] Step S110: Obtain real-time pressure distribution data and arm posture data of the trainee during the supine crunch exercise.

[0084] The pressure distribution data refers to the pressure data between the trainee and the training mat. The supine crunch exercise process includes the movement process of multiple supine crunch movements.

[0085] For example, the pressure distribution data can be a two-dimensional pressure matrix (e.g., establishing a coordinate system with a preset reference point as the origin on the training mat, and recording the pressure value corresponding to each coordinate (x, y) in the coordinate system to obtain the pressure distribution data), specifically including the shoulder pressure center coordinates, hip pressure distribution, and fingertip pressure on the mat in the start and end line areas (unit: N). Arm posture data includes inertial parameters such as real-time arm angle and angular velocity.

[0086] For example, when a trainer completes a crunch, the pressure sensor records the change in the center of pressure in the shoulder area from (50,30) mm to (65,30) mm, while the armband records the process of the left arm bending angle decreasing from 180° to 160° and then returning to 180°.

[0087] Step S120: Based on pressure distribution data and arm posture data, extract the trainee's pressure center, arm angle, and back-to-ground angle during the current movement.

[0088] Specifically, during each supine crunch training session, the trainer continuously performs supine crunches, for example, completing 50 supine crunches in one go. After the trainer begins training, once the current supine crunch movement is detected, various parameters of that movement are measured, i.e., step S120 is executed. After the current supine crunch movement is detected and completed, the effectiveness is assessed based on the parameters measured in step S120.

[0089] The back lift-off angle is the maximum lift-off angle during the current movement. Specifically, a six-axis IMU (gyroscope + accelerometer) is integrated in the center of the training mat. The real-time back lift-off angle is measured using the IMU and an angle fusion algorithm. During the supine crunch movement, the back lift-off angle initially increases and then decreases; therefore, the maximum back lift-off angle is selected to determine if the trainee's upper body flexion range is within the standard range. For example, the maximum back lift-off angle is 25°.

[0090] More specifically, the pressure center can be the shoulder pressure center, and the process of determining the shoulder pressure center can include: based on the training pad's preset coordinate system (with the midpoint of the long side of the training pad as the origin, the length direction as the X-axis, and the width direction as the Y-axis), delineating the shoulder monitoring area (e.g., X-axis range 50-150cm, Y-axis range 20-60cm, which can be adaptively adjusted according to the average shoulder width of an adult), and extracting only the detection data of the pressure sensor array within this area.

[0091] Then, a weighted average is calculated of the coordinates (xᵢ, yᵢ) of each contact point within the area and the corresponding pressure value Fᵢ to obtain the x-coordinate and y-coordinate of the shoulder pressure center. For example, for the x-coordinate, the product of the x-coordinate of each contact point and the pressure value F is calculated. Then, the sum of x*F for each contact point and the sum of the corresponding pressure values ​​F are calculated. The x-coordinate of the shoulder pressure center is obtained by dividing the sum of x*F by the sum of the pressure values ​​F. Similarly, for the y-coordinate of the shoulder pressure center, the sum of y*F is calculated and then divided by the sum of the pressure values ​​F to obtain the y-coordinate of the shoulder pressure center, thus obtaining the shoulder pressure center, which is also the coordinate of the pressure center of the current movement. For example, the coordinates of the pressure center of the current movement are (55, 30).

[0092] More specifically, the process of determining the arm angle can include: using arm posture data measured by a nine-axis sensor on the armband, and calculating the angle between the two arms using a posture calculation algorithm (such as the quaternion method), the arm angle can be obtained. For example, the arm angle is 175°.

[0093] Furthermore, once the current supine crunch movement meets the preset termination condition, the current supine crunch movement is considered to have ended. After the current supine crunch movement ends, the next supine crunch movement begins, and the process proceeds to step S130 to verify the validity of the supine crunch movement.

[0094] Specifically, the preset termination condition may include:

[0095] If the angle of the trainer's back off the ground decreases from the maximum peak (the "maximum angle off the ground" extracted in step S120) to ≤5° (within the range of ±1.5° error) and this state lasts for ≥100ms (to avoid misjudgment due to brief touch), the preset termination condition can be considered met.

[0096] In addition, if the back angle from the ground decreases from the maximum peak to within 5° and remains so for 100ms, and the pressure value in the buttock area remains stable (pressure fluctuation ≤5% over 10 consecutive sampling points (100Hz frequency, i.e., 100ms), and the trainee has fully returned to a supine position, then the preset termination condition is considered met. No further restrictions are imposed here.

[0097] Step S130: Based on the pressure center, arm angle, and back-off-ground angle of the current movement process, determine whether the current movement process is in violation of the rules, and if it is in violation, determine the type of violation and output a prompt message.

[0098] In some embodiments, step S130 above, determining whether the current movement is in violation of regulations based on the pressure center, arm angle, and back-off-ground angle of the current movement process, may specifically include:

[0099] If the distance between the center of pressure and the preset baseline exceeds a preset distance value, the current movement is determined to be a violation, and the violation type is shoulder movement violation. If the arm angle is less than a preset arm angle threshold and the duration reaches a preset duration, the current movement is determined to be a violation, and the violation type is elbow flexion violation. If the back lift angle is less than a preset lift angle threshold, the current movement is determined to be a violation, and the violation type is upper body forward flexion failure.

[0100] For example, the preset baseline is a baseline along the length of the shoulder pressure center (initial coordinates (X0, Y0) determined by the "weighted average method" in step S120) when the trainee is initially lying flat. This preset distance value can be specifically set based on the adult's shoulder range of motion and training standards (e.g., ±15cm). The straight-line distance between the shoulder pressure center and the preset baseline is calculated in real time during the current movement. If the straight-line distance is greater than the preset distance value and lasts for ≥50ms (reducing the probability of false positives), it can be judged as a "shoulder movement violation". Example: The trainee's initial shoulder pressure center is (50, 30). During a crunch movement, the pressure center shifts to (70, 30) (a distance of 20cm), and this state lasts for 60ms, which is judged as a shoulder movement violation.

[0101] For example, the preset arm angle threshold can be 180°; the preset duration can be, for example, 100ms (to avoid instantaneous angle changes caused by arm swinging during the movement). If the real-time arm angle is <180° and lasts for ≥100ms (i.e., 20 consecutive sampling points meet the angle threshold), it is determined to be a "violation of elbow flexion".

[0102] For example, if the maximum angle off the ground in the current movement is less than 28.5° (preset angle off the ground threshold), it is determined that "the upper body forward flexion does not meet the standard".

[0103] Step S140: After the trainee completes the supine crunch exercise, determine and output the effective number of repetitions of the trainee.

[0104] Specifically, the validity of each supine crunch exercise is detected through steps S120 and S130. When a violation is determined, it indicates that the supine crunch exercise is invalid and is not counted as a valid exercise. Only when the exercise has no violation type is it counted as a valid exercise.

[0105] After training is completed (e.g., when the trainer presses the "End" button on the touch sensor area), the posture monitoring device can display the total number of valid attempts on the touch screen of the internally pre-installed intelligent interaction module. It can also upload data to the server via the communication module (Ethernet or Bluetooth) to generate an electronic report containing details of each action (such as the time and type of violation).

[0106] For example, if a trainer completes 15 repetitions, and 3 of them are deemed invalid due to bent elbows or failure to touch the line, the system will ultimately output 12 valid repetitions and synchronize them to the preset physical fitness assessment system.

[0107] Through steps S110-S140, the multimodal fusion measurement method of pressure sensing and arm posture measurement can simultaneously capture pressure changes and spatial posture. This allows for the identification of violations in spatial posture, thereby improving the accuracy of action validity recognition and reducing the probability of misjudging hidden violations such as "amplitude meets the standard but the action is deformed".

[0108] Furthermore, steps S110-S140 can replace manual monitoring, achieve fully automatic judgment, improve monitoring efficiency, and unify judgment standards, making it easier to meet the needs of large-scale training and assessment.

[0109] In some embodiments, the posture monitoring device also includes voice broadcast and / or light prompts. During the supine crunch exercise, for each violation of the supine crunch movement, a flashing light and voice broadcast can be activated to provide a prompt (e.g., "Second attempt: fingertips did not touch the line!"). Furthermore, if the trainee violates the rules three or more times consecutively, the training mat can vibrate as a reminder, and the exercise can be forcibly paused with a corrective animation. This allows for real-time correction of movements through lights and voice, and automatically generates an electronic record after training, solving the problems of lost and difficult-to-trace manual records.

[0110] In some embodiments, refer to Figure 2 The posture monitoring and training method also includes steps S210-S260.

[0111] Step S210: Determine the start time of the current action process.

[0112] Step S220: Determine whether the current action process has ended.

[0113] For example, in determining whether the current action violates regulations, the compliance of touching the start and end lines can also be assessed to determine if there are any violations due to failure to touch the start and end lines. More specifically, the fingertip pressure on the start and end lines during the current action can be monitored to determine violations.

[0114] More specifically, the methods for determining the start and end times of the current movement can include: the start time can be the instant when the back lift angle increases from ≤5° to >5°. The end time can be determined according to the method for determining the end time of the movement in step S120. For example, it can be the instant when the back lift angle decreases to ≤5° and lasts for ≥100ms.

[0115] Step S230: When the current action process ends, find the fingertip pad pressure on the start line and end line of the current action process from the pressure distribution data.

[0116] The starting line and the ending line are the positions that the trainee's hands should touch in sequence during the supine crunch exercise.

[0117] Step S240: Determine whether the fingertip pressure on the starting line is effective based on the pressure applied to the starting line.

[0118] Step S250: Determine whether the fingertip touch pad pressure at the termination line is effective based on the situation.

[0119] Specifically, pressure sensors are built into the preset positions of the start and end lines of the training mat to collect fingertip touch pressure (unit: N). By analyzing the fingertip touch pressure data in the pressure distribution data, the fingertip touch pressure on the start line and the fingertip touch pressure on the end line can be determined.

[0120] The criterion for determining whether the double-sided touch line is effective is that the pressure of both left and right fingertips is greater than or equal to the preset pressure value. The preset pressure value can be set based on the average touch pressure of an adult's fingertips, for example, 5N.

[0121] For example, if there is fingertip pad pressure on both sides of the starting line, and the fingertip pad pressure is greater than 5N, then the current action is considered to be effective if both sides touch the starting line. Similarly, if there is fingertip pad pressure on both sides of the ending line, and the fingertip pad pressure is greater than 5N, then the current action is considered to be effective if both sides touch the ending line.

[0122] Step S260: If it is not a case where both sides of the starting line are valid, and / or it is not a case where both sides of the ending line are valid, then it is determined that there is a violation in the current action process.

[0123] For example, if the fingertip pressure on one or both sides of the starting line is <5N (the starting line is not touched), it indicates that the case of both sides touching the starting line is not effective. If the fingertip pressure on one or both sides of the ending line is <5N (the starting line is not touched), it indicates that the case of both sides touching the ending line is not effective.

[0124] If the current action does not fall under the condition that both sides of the starting line are valid, and / or does not fall under the condition that both sides of the ending line are valid, then the current action process is determined to have a violation, that is, it is judged as "starting and ending line touching violation".

[0125] For example: During a certain exercise, the trainer's left hand fingertip pressure at the starting line was 3N (not up to standard) and the right hand pressure was 5N (up to standard). The system judged that the starting line did not meet the requirement of valid double-sided contact, and the exercise was a violation.

[0126] By adding contact pressure detection, the shortcomings of relying solely on angle or posture monitoring are compensated for, and the hidden violation problem of failure to touch the line even when the amplitude meets the standard is solved.

[0127] In some embodiments, refer to Figure 3 The posture monitoring and training method may also include steps S310-S350.

[0128] Step S310: Acquire the initial image of the trainer and the initial hand touch pressure sensing data at the preset starting line on the training mat.

[0129] Specifically, the initial head image is captured by a high-definition camera positioned directly above the training mat, showing the trainee in an initial lying position. The image includes the trainee's entire upper body (from shoulders to hips) and the position of both hands at the starting line, which is used for subsequent analysis of body shape characteristics and arm movement direction.

[0130] Specifically, the acquisition of this initial hand touch pressure sensing data may include:

[0131] Sensor arrays are sequentially arranged along the length of the starting and ending lines. One or more sensors can be controlled to operate as needed to detect the pressure of objects placed on them. Initially, all sensors on the starting and ending lines are active. The trainee lightly touches the starting line with their fingertips using a standard technique, and the initial hand pressure data is detected. For example, after the trainee touches the starting line, the detected pressure values ​​at multiple coordinate points within the starting line area are: 3N at (80,200), 5N at (85,202), 4N at (115,202), and 2N at (120,200).

[0132] Step S320: Determine the pressure center point of the fingertip pad of the starting line based on the initial hand touch pressure sensing data.

[0133] Specifically, in step S320, effective contact points with pressure values ​​≥1N within the initial line area are selected from the initial hand touch pressure sensing data (interference signals are eliminated). For the coordinates (xᵢ, yᵢ) and corresponding pressure value Fᵢ of each effective contact point, a weighted average method is used to calculate the center point, and the position of the calculated center point is the position of the pressure center point.

[0134] For example, the x-coordinate of the center point is the ratio of the sum of x*F of all effective contact points to the sum of F of all effective contact points. The y-coordinate of the center point is the ratio of the sum of y*F of all effective contact points to the sum of F of all effective contact points.

[0135] Step S330: Based on the trainer's initial image, determine the trainer's arm movement direction, and predict the position of the pressure center point of the fingertip touching the pad at the termination line according to the arm movement direction.

[0136] Specifically, firstly, the length direction of the trainer's arm is identified from the initial image obtained in step S310, and the length direction of the arm is the direction of the trainer's arm movement.

[0137] Specifically, for this initial image, a skeletal keypoint detection algorithm (such as MediaPipe) can be used to extract the two acromion points (prominent points of the shoulder bones) and fingertip points (the fingertips touching the starting line) of the arm. For example, the coordinates of the left acromion point are identified as image pixels (300, 450), and the coordinates of the left fingertip point are (320, 600); the coordinates of the right acromion point are (700, 450), and the coordinates of the right fingertip point are (680, 600). The straight line formed by connecting the acromion point and the fingertip point on the same side represents the length direction of the arm, which is also the direction of the trainer's arm movement.

[0138] Then, using the starting line fingertip pad pressure center point obtained in step S320 as a reference, extend it along the arm movement direction to the termination line. The intersection of the extended line and the termination line is the predicted coordinate of the fingertip pad pressure center point of the termination line.

[0139] Step S340: Based on the initial image of the trainer, analyze the trainer's body shape characteristics and determine the normal sensing width of the trainer's start and end lines according to the body shape characteristics.

[0140] Among them, the normal sensing width of the start and end lines is the reasonable fluctuation range of the hand position after taking into account the differences in the current trainer's body shape.

[0141] Specifically, in step S340, the trainer's body shape characteristics are analyzed. This may include: extracting features from the initial image in step S310, and using an image segmentation algorithm to obtain shoulder width (the horizontal distance between the two acromions) and arm length (the straight-line distance from the acromion to the wrist) to obtain body shape characteristics. For example, the analysis may show that the trainer's shoulder width is 48cm and arm length is 60cm.

[0142] Then, when determining the normal sensing width of the trainer's start and end lines based on body shape characteristics, specifically, the trainer's shoulder width can be used to determine the range to which the trainer's shoulder width belongs, and the trainer's normal sensing width of the start and end lines can be determined based on the preset sensing width corresponding to the range to which the trainer's shoulder width belongs.

[0143] For example, when the shoulder width is less than 40cm (slender build), the corresponding sensing width is 4cm (2cm to the left and right of the center point). It is worth noting that the wider the shoulder width, the larger the corresponding preset sensing width value.

[0144] Step S350: Based on the normal sensing width of the start and end lines, and the pressure center point positions of the start and end lines, determine the trainer's start and end line sensing range.

[0145] The starting and ending line sensing intervals specifically include the starting line sensing interval and the ending line sensing interval.

[0146] For example, taking the center point of the starting line in step S320 as a reference and combining it with the sensing width determined in step S340, the sensing range of the starting line is obtained. The sensing range of the starting line is the range with the center of the starting line as the midpoint and along the length direction of the starting line, and the length is the normal sensing width of the starting and ending lines.

[0147] Similarly, the sensing range of the termination line is the sensing range defined by taking the center point of the termination line in step S330 as the reference and combining it with the sensing width in step S340.

[0148] In step S230 above, finding the fingertip pad pressure on the start and end lines during the current movement from the pressure distribution data can specifically include: finding the fingertip pad pressure within the sensing range of the start and end lines during the trainee's movement from the pressure distribution data.

[0149] For example, after determining the trainer's start and end line sensing range, when judging the compliance of the trainer's start and end line touch, the pressure measured by sensors in other ranges on the start and end line can be shielded, and only the pressure measured in the start and end line sensing range can be used as the judgment condition. This can reduce the chance of misjudging the effectiveness of the action due to accidental touch of the start and end line.

[0150] In some embodiments, continue to refer to Figure 3 In this attitude monitoring and training method, after step S350, the following steps S360-S370 are also included.

[0151] Step S360: When the trainee starts performing supine crunches, acquire the trainee's exercise video data within a preset correction time period after the start of the exercise, and analyze the trainee's hand position data at the starting line and the ending line based on the exercise video data.

[0152] For example, the preset correction time period can be set to the first 3 complete movements (about 10 seconds) after the start of training. At this time, the trainee's movements have not yet been significantly deformed due to fatigue, and can reflect the real movement habits.

[0153] Specifically, the motion video data can be obtained by recording the complete process of these three movements using a high-definition camera (30fps) above the training mat.

[0154] Specifically, each frame in the motion video data is identified to capture the instant when the fingertips touch the starting and ending lines (one frame is marked for each touch). For example, the recorded video contains: the first action where both hands touch the starting line (frame 10), and the second and third actions where both hands touch the ending line (frame 30). The same applies to the second and third actions, resulting in a total of 6 frames marked for the touches.

[0155] Then, each marked video frame is analyzed to analyze the hand position data. Specifically, the fingertip coordinates can be extracted from the marked frames using a skeletal keypoint algorithm and converted into the actual position coordinates in the training pad coordinate system.

[0156] Step S370: Based on the newly acquired hand position data, the normal sensing width of the trainer's start and end lines is corrected, and based on the corrected normal sensing width of the start and end lines, as well as the pressure center point position of the start line and the pressure center point position of the end line, the trainer's start and end line sensing range is redefined.

[0157] Specifically, in step S370, the position of the sensing zone is adjusted based on the average offset of the actual position to ensure that the zone is aligned with the trainer's movement habits. The specific implementation method is as follows:

[0158] Using the initial center point determined in steps S320 and S330 as a reference, the distance between each actual fingertip position and the corresponding center point is determined, and the starting and ending line sensing width is corrected based on the distance.

[0159] Specifically, if in step S360, the maximum positional deviation between the starting line and the initial center is 6 in multiple actions of a certain hand within a preset time period (assuming the original interval width is 10), then the sensing width of the starting line is corrected from 10 to 12.

[0160] Then, using the initial center point as a reference, the corrected start and end line sensing range is determined according to the corrected start and end line sensing width. For example, if the starting line sensing width is corrected to 12, then the distance from the boundary to the initial center point is 6. In this way, the range is dynamically adjusted based on actual motion data, avoiding misjudgments caused by individual motion habits where the effective contact exceeds the original range, thus improving the adaptability of the judgment.

[0161] Furthermore, after a trainer has committed a foul, such as a foul that causes a significant deviation from the original position, subsequent trainers may touch the start / stop line but fail to detect it, leading to misjudgments in the validity of the trainer's actions.

[0162] Therefore, refer to Figure 4 The posture monitoring and training method also includes the following steps S410-S460.

[0163] Step S410: If the current action process is judged to be a violation, obtain the trainer's motion image data.

[0164] For example, if the trainer violates the shoulder movement rule, meaning the current movement is deemed a violation, the trainer's actual position may change after the violation has occurred. Therefore, it is necessary to acquire motion image data of the trainer.

[0165] Step S420: Extract the torso baseline from the motion image data.

[0166] The trunk baseline is the midline of the trainer's longitudinal axis.

[0167] Specifically, in step S420, the extraction of the trunk baseline can be achieved by using a skeletal detection algorithm (such as MediaPipe Pose) to extract the midpoint of the acromion (the midpoint of the line connecting the left and right acromions) and the midpoint of the hip (the midpoint of the line connecting the left and right hip bones) from the motion image data in step S410.

[0168] The straight line formed by connecting the midpoint of the acromion and the midpoint of the hip is the midline of the longitudinal axis of the trunk (trunk baseline).

[0169] Step S430: Calculate the relative offset between the torso baseline and the initial torso baseline.

[0170] The initial trunk baseline is the trunk baseline of the trainee during the initial movement phase.

[0171] Specifically, in step S430, while the trainer is in the initial preparation state, the torso baseline recorded at the start of training (initial movement phase) is extracted in the manner described in step S420. For example, the linear equation of the initial torso baseline is X=97cm (horizontal direction of the X-axis).

[0172] Next, the horizontal offset between the current torso baseline and the initial torso baseline is compared. It's worth noting that, when the current torso baseline and the initial torso baseline are not parallel, the offset between the current torso baseline and the intersection of the initial torso baseline and the starting line is calculated, as are the offsets between the current torso baseline and the ending line, and the initial torso baseline and the ending line. The offset between these two intersection points is then used to obtain the relative offset between the two baselines.

[0173] Step S440: Determine the position compensation values ​​corresponding to the starting line and the ending line based on the relative offset.

[0174] Specifically, the position compensation value must be consistent with the torso offset to ensure that the starting and ending line sensing ranges shift synchronously with the torso. For example, if the relative offset of the starting line is -2, then the position compensation value is -2, where positive and negative indicate direction.

[0175] Step S450: Based on the position compensation value, adjust the start and end line sensing intervals used in subsequent movements so that the relative relationship between the adjusted start and end line sensing intervals and the trainee's current torso position is consistent with the relative relationship between the start and end line sensing intervals and the initial torso baseline in the initial movement phase.

[0176] Step S460: After adjusting the start and end line sensing range, for subsequent actions, detect the fingertip touch pad pressure in the adjusted start and end line sensing range and judge the effectiveness of the action.

[0177] Specifically, the relative relationship between the adjusted start and end line sensing range and the trainee's current torso position is consistent with the relative relationship in the initial movement phase. For example, in the initial phase, the range of the start line sensing range relative to the initial torso baseline is 50cm to the left and 50cm to the right; then the range of the currently adjusted start and end line sensing range relative to the current torso baseline remains 50cm to the left and 50cm to the right.

[0178] For subsequent movements (such as the 6th time and beyond), the adjusted start and end line sensing range can be invoked. A pressure sensor array collects the fingertip pressure within this range to determine if the touch was valid. This way, even if the trainer has already violated the rules, by synchronously adjusting the range position with torso movement, the problem of valid touches being misjudged as violations after torso shifts is avoided, thus improving the accuracy of subsequent movement determinations.

[0179] Furthermore, if the starting or ending line sensor malfunctions during the trainee's supine crunch exercise, it may result in a false judgment that "the starting and ending lines have not been touched," thus reducing the accuracy of the trainee's supine crunch exercise validity recognition.

[0180] Based on this, refer to Figure 5 The posture monitoring and training method also includes the following steps S510-S530.

[0181] Step S510: If a fault is detected in the start line sensor and / or the end line sensor, control the auxiliary sensor set parallel to the surface of the training mat to detect the contact state between the trainee's hand and the ground.

[0182] Specifically, the communication status and data validity of the start and end line sensors can be monitored in real time. For example, if the sensor provides no data feedback, or the pressure value remains constant at 0N (no contact), or the pressure value exceeds 100N (outside the physically reasonable range) during three consecutive actions, the sensor is considered faulty. For instance, if the end line sensor continuously outputs 0N during the 5th to 8th actions, it is considered faulty.

[0183] The training mat has a built-in array of infrared photocells (spaced, for example, 1cm apart) parallel to the surface, which automatically activates upon fault triggering. The distance between the light emitted by the infrared sensors and the surface of the training mat is less than a preset value, for example, 0.5cm. The infrared sensors determine whether the hand is in contact with the training mat by the reception time of the returned light. When the hand touches the mat, it blocks the light, and the light returns within a preset time. When the hand is not in contact with the mat, the light does not return within a preset time, or the return time is longer. Based on the duration of the returned light, the system determines whether the hand is in contact with the training mat.

[0184] Step S520: Based on the acquired motion video data of the trainer, analyze the relative positional relationship between the trainer's hand and the starting line and the ending line.

[0185] Specifically, in step S520, based on the motion video data in real time, a target detection algorithm (such as YOLOv8) is used to extract the hand contour (fingertip key points) from each video frame, and the start and end lines (preset as black solid lines) on the training mat are identified through image segmentation. The pixel distance between the fingertip key points and the lines is calculated, converted into actual distance, and the relative position is determined. For example, if the distance between the fingertip projection on the training mat and the line is ≤1cm, it is determined that "the hand is located within the line area"; if the distance is >1cm, it is determined that "the hand is located outside the line area".

[0186] Step S530: Based on the contact state between the hand and the ground detected by the auxiliary sensor and the relative position relationship, determine whether the current action violates the start and end line contact rule.

[0187] Specifically, in step S530, the "contact state" determined in step S510 and the "relative position" determined in step S520 are combined to jointly determine whether the current action violates the start and end line contact rule. For example:

[0188] Compliance: The auxiliary sensor detects "contact," and the video shows "the hand is within the line area." Violation: The auxiliary sensor detects "no contact," or the video shows "the hand is outside the line area" (regardless of whether there was contact).

[0189] Further, continue to refer to Figure 5 The posture monitoring and training method also includes the following steps S540-S560.

[0190] Step S540: Determine the missing time period in which the completion status of the action was not accurately recorded due to sensor failure.

[0191] For example, if the sensor data feedback is abnormal during three consecutive actions, it is considered that the sensor is faulty. Therefore, the validity identification result of these three consecutive actions may be misjudged due to the sensor malfunction. Thus, the complete time period of these three consecutive actions is regarded as the missing time period.

[0192] Specifically, based on the end time of the current action, the start times of the two actions preceding the current action can be found, and the time interval between the start time of the earliest action and the end time of the current action can be used as the missing time interval.

[0193] For example, after the 10th action, the system detected that the sensor data for the termination line of the 8th, 9th, and 10th actions was consistently 0N (contradicting the video footage of the hand touching the line), thus determining these three actions as abnormal records during a sensor malfunction. Therefore, using the earliest abnormal action start time (10:06:20 for the 8th action) as the starting point and the latest abnormal action end time (10:06:35 for the 10th action) as the ending point, the missing time period is determined to be 10:06:20-10:06:35.

[0194] Step S550: Extract video data within the missing time period from the trainer's motion video data.

[0195] Specifically, from the motion video of the complete training process stored in the camera, a video segment (30fps, 450 frames in total) corresponding to the missing time period (10:06:20-10:06:35) determined in step S540 is extracted. This segment contains the complete process of the 8th, 9th, and 10th movements, and the image clearly shows the relationship between the trainer's hand movements and the starting and ending lines.

[0196] Step S560: Based on the video data within the missing time period, extract the trainer's fingertip movement trajectory from the video data. Based on the fingertip movement trajectory, analyze whether the trainer's actions within the missing time period involve touching the start and end lines in violation of the rules, so as to determine the number of valid actions within the missing time period.

[0197] Specifically, in step S560, for each video frame, a skeletal tracking algorithm (such as MediaPipe) is used to analyze the video segment of the missing time period frame by frame, extract the coordinate changes of the fingertip of each finger, and obtain the fingertip movement trajectory of each finger.

[0198] When recording the number of effective movements within the missing time period, the effectiveness of each movement is analyzed separately. Specifically, in the supine crunch exercise, the trainer usually uses the tip of the middle finger as the primary contact point (because the middle finger is the easiest to exert force and is the most forward when the hand is extended). From the movement trajectory of each fingertip, the tip of the middle finger is selected as the core judgment object, ignoring the interference of secondary fingers such as the ring finger and little finger (to avoid misjudgment caused by the natural spread of the fingers).

[0199] The coordinates of the middle fingertip (training mat coordinate system) are extracted frame by frame to generate the motion trajectory. The straight-line distance between the core fingertip coordinates and the starting and ending lines are calculated respectively. The effective touch threshold is set to ≤2mm (allowing for slight deviations between the fingertip and the line, considering video pixel accuracy and motion tolerance). That is, when the distance between the core fingertip and the line is ≤2mm, it is considered a "valid touch".

[0200] Then, based on the validity of the starting line touch and the validity of the ending line touch on both sides, it is determined whether there is a mistake in the contact between the starting line and the ending line. If there is no mistake and no other fouls, the action is considered valid, and the number of valid actions is recorded.

[0201] Thus, through steps S510-S560 above, the motion record can still be accurately completed even when the sensor fails, by using video backtracking, thereby reducing the statistical error of the number of valid motions in the missing time period.

[0202] In other embodiments, this posture monitoring and training method can also acquire video data of the trainee's movements during supine crunches. The video data, the center of pressure during the current movement, the arm angle, and the back-off-ground angle are all input into a preset movement compliance model for analysis to determine the valid and invalid movements during the supine crunches, and to identify the type of violation for each invalid movement. The results of the valid movements counted by the preset movement compliance model are compared with the results of the valid movements counted based on a threshold, and if discrepancies exist, the video is manually reviewed for final confirmation. This further improves the accuracy of the test.

[0203] Specifically, this preset movement compliance model is an intelligent judgment model based on multimodal data fusion. By integrating video visual information and sensor quantitative data, it achieves high-precision analysis of the compliance of supine crunch movements. The inputs to the preset movement compliance model include movement video data and sensor quantitative data (coordinates of the center of pressure, arm angle, and back-off-ground angle). This data is synchronized through timestamps to ensure the consistency of the model's analysis of the movement over time.

[0204] The preset action compliance model is a deep learning model that is trained in advance using a large amount of sample data. The sample data can include a large number of labeled cases, covering the action data of trainers with different body types and movement habits. Each sample is labeled with a valid / invalid label and a specific violation type (such as shoulder movement, elbow flexion, and failure to meet the upper body forward flexion standards).

[0205] Finally, the video data and parameters are input into the preset action compliance model, which will output the action validity results and violation types, as well as the number of valid records.

[0206] An embodiment of the second aspect of this application provides a posture monitoring and training system for identifying and monitoring the posture of a supine crunch movement. The posture monitoring and training system includes a training mat body, a smart armband, and a posture monitoring device.

[0207] The training mat itself contains an array of pressure sensors that detect the pressure data between the trainee and the mat, obtaining pressure distribution data. The smart armband monitors the trainee's arm flexion angle, obtaining arm posture data.

[0208] The posture monitoring device is electrically connected to the pressure sensor inside the training mat and to the smart armband. The posture monitoring device receives real-time pressure distribution data and arm posture data from the trainee during the supine crunch exercise, which includes multiple crunch movements. Based on the pressure distribution data and arm posture data, the device extracts the trainee's center of pressure, arm angle, and back-to-ground angle during the current movement, with the back-to-ground angle being the maximum angle of departure. Based on the center of pressure, arm angle, and back-to-ground angle during the current movement, the device determines whether the movement is in violation of the rules. If a violation occurs, the type of violation is determined and a warning message is output. After the trainee completes the supine crunch exercise, the device determines and outputs the number of valid repetitions.

[0209] An embodiment of the second aspect of this application provides a computer-readable medium having a computer program stored thereon, characterized in that the program, when executed by a processor, implements the content of the above-described method embodiments.

[0210] The above are merely some embodiments of this application and are not intended to limit this application. The technical features or structures in the foregoing different embodiments can be arbitrarily combined to form other specific technical solutions as needed. For those skilled in the art, this application can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the protection scope of the claims of this application.

Claims

1. A method of posture monitoring and training, characterized by, The posture monitoring and training method is used to identify and monitor the posture of the supine crunch movement. The posture monitoring and training method includes: The real-time pressure distribution data and arm posture data of the trainee during the supine crunch exercise are obtained. The pressure distribution data is the pressure data between the trainee and the training mat. The supine crunch exercise includes the movement process of multiple supine crunch movements. Based on the pressure distribution data and the arm posture data, the shoulder pressure center, arm angle and back lift-off angle of the trainee during the current movement are extracted, wherein the back lift-off angle is the maximum lift-off angle during the current movement. Based on the shoulder pressure center, arm angle, and back off-ground angle of the current movement, determine whether the current movement is in violation of the rules, and if so, determine the type of violation and output a prompt message. Once the trainer completes the supine crunch exercise, determine and output the effective number of repetitions performed by the trainer.

2. The method of claim 1, wherein, The determination of whether the current movement is in violation of regulations based on the shoulder pressure center, arm angle, and back lift-off angle during the current movement includes: If the distance between the shoulder pressure center and the preset baseline exceeds a preset distance value, the current action process is determined to be in violation, and the violation type is shoulder movement violation; If the arm angle is less than a preset arm angle threshold and the duration reaches a preset duration, the current action process is determined to be in violation, and the violation type is elbow flexion violation. If the back angle off the ground is less than the preset ground angle threshold, the current action is determined to be in violation, and the violation type is upper body forward flexion failure.

3. The attitude monitoring and training method according to claim 2, characterized in that, The attitude monitoring and training method also includes: Determine the start time of the current action process; Determine whether the current action process has ended; When the current movement process ends, find the fingertip pad pressure on the starting line and the ending line of the current movement process from the pressure distribution data. The starting line and the ending line are the positions that the trainee's hands should touch in sequence during the supine crunch movement. Based on the fingertip pressure on the starting line, determine whether the starting moment is a case where both sides of the starting line are effective. Based on the fingertip pressure on the termination line, determine whether the termination line is effectively touched on both sides at the end time. If the start time is not within the case where both sides of the start line are valid, and / or the end time is not within the case where both sides of the end line are valid, then it is determined that there is a violation in the current action process.

4. The attitude monitoring and training method according to claim 3, characterized in that, The attitude monitoring and training method also includes: Acquire the trainer's initial image and initial hand touch pressure sensing data at a preset starting line on the training mat; The pressure center point of the fingertip pad of the starting line is determined based on the initial hand touch pressure sensing data. Based on the trainer's initial image, the direction of the trainer's arm movement is determined, and the position of the pressure center point of the fingertip touch pad at the termination line is predicted according to the direction of the arm movement. Based on the trainer's initial image, the trainer's body shape characteristics are analyzed, and the normal sensing width of the trainer's start and end lines is determined according to the body shape characteristics. The normal sensing width of the start and end lines is the reasonable fluctuation range of the hand position after considering the current differences in the trainer's body shape. Based on the normal sensing width of the start and end lines, and the pressure center point position of the start line and the pressure center point position of the end line, the trainer's start and end line sensing range is determined. The step of finding the fingertip pad pressure on the start and end lines of the current action process from the pressure distribution data includes: From the pressure distribution data, find the fingertip pressure on the pad within the starting and ending line sensing range during the trainee's movement.

5. The attitude monitoring and training method according to claim 4, characterized in that, The attitude monitoring and training method also includes: When the trainer begins to perform supine crunches, during a preset correction time period after the start of the exercise, the trainer's exercise video data is acquired, and the trainer's hand position data at the starting line and the ending line is analyzed based on the exercise video data. Based on the newly acquired hand position data, the normal sensing width of the trainer's start and end lines is corrected, and based on the corrected normal sensing width of the start and end lines, as well as the pressure center point position of the start line and the pressure center point position of the end line, the trainer's start and end line sensing range is redefined.

6. The attitude monitoring and training method according to claim 4 or 5, characterized in that, The attitude monitoring and training method also includes: If the current action is determined to be a violation, the trainer's motion image data is acquired; Extract the torso baseline from the motion image data, wherein the torso baseline is the midline of the trainee's torso longitudinal axis; Calculate the relative offset between the trunk baseline and the initial trunk baseline, which is the trunk baseline of the trainee during the initial movement phase; Based on the relative offset, determine the position compensation values ​​corresponding to the starting line and the ending line respectively; Based on the position compensation value, the start and end line sensing intervals used in subsequent movements are adjusted so that the relative relationship between the adjusted start and end line sensing intervals and the trainee's current torso position is consistent with the relative relationship between the start and end line sensing intervals and the initial torso baseline in the initial movement phase. After adjusting the start and end line sensing range, the fingertip pressure of the adjusted start and end line sensing range is detected and the effectiveness of the action is judged for subsequent actions.

7. The attitude monitoring and training method according to claim 4, characterized in that, The attitude monitoring and training method also includes: In the event of a malfunction detected in the start line sensor and / or the end line sensor, an auxiliary sensor arranged parallel to the surface of the training mat is controlled to detect the contact state between the trainee's hand and the ground, wherein the start line sensor and the end line sensor are used to detect the pressure of an object placed on the sensor. Based on the acquired video data of the trainer's movements, the relative positional relationship between the trainer's hands and the starting and ending lines is analyzed. Based on the contact state between the hand and the ground detected by the auxiliary sensor, and the relative positional relationship, it is determined whether the current action violates the start and end line rules.

8. The attitude monitoring and training method according to claim 7, characterized in that, The attitude monitoring and training method also includes: Identify the missing time periods where the completion of actions was not accurately recorded due to sensor malfunction; Extract the video data within the missing time period from the trainer's motion video data; Based on the video data within the missing time period, the trainer's fingertip movement trajectory is extracted from the video data. Based on the fingertip movement trajectory, it is analyzed whether the trainer's actions within the missing time period violate the rules by touching the start and end lines, so as to determine the number of valid actions within the missing time period.

9. An attitude monitoring and training system, characterized in that, The posture monitoring and training system is used to identify and monitor the posture of the supine crunch movement. The posture monitoring and training system includes: The training mat body is equipped with a pressure sensor array, which can detect the pressure data between the trainee and the training mat body and obtain pressure distribution data. The smart armband is used to monitor the bending angle of the trainee's arm and obtain arm posture data; The posture monitoring device is electrically connected to the pressure sensor inside the training pad and to the smart arm ring. The posture monitoring device is used to receive real-time pressure distribution data and arm posture data of the trainee during the supine crunch exercise, wherein the supine crunch exercise includes the movement process of multiple supine crunch movements. Based on the pressure distribution data and the arm posture data, the shoulder pressure center, arm angle and back lift-off angle of the trainee during the current movement are extracted, wherein the back lift-off angle is the maximum lift-off angle during the current movement. Based on the shoulder pressure center, arm angle, and back off-ground angle of the current movement, determine whether the current movement is in violation of the rules, and if so, determine the type of violation and output a prompt message. Once the trainer completes the supine crunch exercise, determine and output the effective number of repetitions performed by the trainer.

10. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-8.

Images