Sit-up quality evaluation method and system based on human posture, and medium

Abstract

The application provides a sit-up quality evaluation method and system based on human posture, and a medium. The method comprises the following steps: obtaining subject number information, performing identity verification on the subject according to the subject number information, and obtaining a verification result; inputting subject identity information and configuring test parameters; testing the subject according to test environment information, collecting human image information of the subject in real time, and preprocessing the human image information of the subject; analyzing human posture data of the subject, extracting key posture features of a sit-up action, constructing a standard action posture model of the sit-up, setting a feature threshold of a compliant action, analyzing the key posture features of the sit-up action, and generating test data; generating a multi-dimensional sit-up action quality evaluation report according to the test data, and transmitting the report to a terminal in real time; and accurately checking action requirements and effectively identifying various types of illegal actions through human posture recognition technology, so as to ensure the accuracy of the test result.

Description

Technical Field

[0001] This application relates to the field of physical fitness testing technology, and more specifically, to a method, system, and medium for assessing sit-up quality based on human posture. Background Technology

[0002] Sit-ups are a core component of national physical fitness tests, student physical fitness assessments, and other fitness evaluations, used to assess abdominal muscle endurance. The accuracy and fairness of these tests directly impact the validity of the fitness assessment results. Traditional sit-up tests rely primarily on manual counting, which has the following significant drawbacks:

[0003] Manual counting has large errors: Testers are prone to visual fatigue during long periods of invigilation, which can lead to problems such as omissions, miscounts, and misjudgments. In addition, the manual standards are not uniform, making it impossible to accurately control compliance requirements such as "elbows past the knees and shoulder blades touching the pad".

[0004] Cheating is difficult to control: test takers are prone to using their elbows to support themselves on the pad, raising and lowering their hips, and making insufficient movements, which are difficult for human proctors to identify in real time and accurately.

[0005] Limitations of traditional equipment: Existing infrared testing instruments rely on through-beam infrared grating detection, which has problems such as complex installation and debugging, susceptibility to ambient light interference, and inability to identify non-standard movement postures; belt-type testing instruments only count waist movement trajectories, which cannot accurately verify the compliance of upper limb movements, are prone to "false movement" counts, and cannot achieve full-dimensional movement quality assessment.

[0006] Insufficient intelligence: Traditional equipment can only perform basic counting and cannot quantitatively assess the standardization of movements, the way of exerting force, the continuity of movements, etc., and cannot provide targeted training guidance for subjects. Summary of the Invention

[0007] The purpose of this application is to provide a method, system, and medium for evaluating sit-up quality based on human posture. By using human posture recognition technology, it can accurately verify action requirements, effectively identify various violations, and ensure the accuracy of test results.

[0008] This application embodiment also provides a sit-up quality assessment method based on human posture, including: obtaining subject ID information, verifying the subject's identity based on the subject ID information, obtaining verification results, entering the subject's identity information based on the verification results, configuring test parameters based on the subject's identity information, and obtaining test environment information;

[0009] The subjects were tested based on the test environment information, and human body image information of the subjects was collected in real time. The human body image information of the subjects was preprocessed to obtain optimized image information.

[0010] Based on the optimized image information analysis of the subject's human posture data, key posture features of the sit-up movement are extracted from the human posture data.

[0011] Construct a standard sit-up posture model, set feature thresholds for compliant movements, analyze key posture features of sit-ups based on the standard posture model and compliant movement feature thresholds, and generate test data.

[0012] A multi-dimensional sit-up quality assessment report is generated based on the test data, and the report is transmitted to the terminal in real time.

[0013] Optionally, in the sit-up quality assessment method based on human posture described in the embodiments of this application, before testing the subject based on the test environment information, the method further includes:

[0014] Obtain the body shape parameter information of the subject, which includes at least waist circumference data and height data;

[0015] The body type classification result of the subject is calculated based on the body shape parameter information;

[0016] Obtain the body shape range mapping database, and based on the body shape classification results, obtain the baseline value of the sit-up range of motion and the trunk tilt angle threshold range data that are suitable for the body shape of the subject from the body shape range mapping database.

[0017] Based on the amplitude range benchmark value and the torso tilt angle threshold range data, set the body shape adaptation compliance action feature threshold.

[0018] Optionally, in the sit-up quality assessment method based on human posture described in this application embodiment, the step of analyzing the key postural features of the sit-up movement based on the compliance movement feature threshold of body shape adaptation to generate test data includes:

[0019] Obtain the range of motion data of the subject during the first few effective sit-ups in the current test;

[0020] Calculate the average effective amplitude value based on the preset effective amplitude of the previous several effective movements;

[0021] The dynamic alert threshold is obtained based on the average effective amplitude value and the preset amplitude attenuation coefficient;

[0022] When it is detected that the amplitude of a subsequent sit-up movement is less than the dynamic reminder threshold, an insufficient amplitude reminder message is generated;

[0023] If the amplitude of the consecutive preset number of sit-ups is less than the dynamic reminder threshold and no amplitude improvement trend information is detected, the consecutive preset number of sit-ups will be marked as invalid movements and will not be included in the valid count.

[0024] Optionally, in the sit-up quality assessment method based on human posture described in the embodiments of this application, the preprocessing of the subject's human image information specifically includes:

[0025] Obtain human body image information of the subject, and perform denoising processing on the human body image information of the subject according to the filtering algorithm to obtain denoised human body image;

[0026] Based on the human target detection algorithm, the denoised human image is analyzed to locate the subject's human body region and background region, the background region is cropped and removed, and the region of interest is extracted.

[0027] The cropped region of interest image is normalized in size and grayscale, and the image size is uniformly adjusted to the preset standard size to obtain standard human image information;

[0028] The human body region is enhanced by applying an image enhancement algorithm to standard human body image information to obtain an enhanced image.

[0029] Optionally, in the sit-up quality assessment method based on human posture described in this application embodiment, the subject's human posture data is analyzed based on optimized image information, and key postural features of the sit-up movement are extracted based on the human posture data, specifically including:

[0030] Optimized image information is acquired, and the optimized image information is analyzed frame by frame according to the pose estimation algorithm to obtain the whole body skeletal joint coordinates, joint angles and motion trajectory of the subject.

[0031] Key posture features of the upper limbs, lower limbs, and trunk were extracted based on the coordinates, angles, and movement trajectories of the subjects' entire skeletal joints.

[0032] Redundancy removal and anomaly analysis were performed on the key posture features of the upper limbs, lower limbs, and trunk to obtain the key posture features of the sit-up movement.

[0033] Optionally, in the sit-up quality assessment method based on human posture described in the embodiments of this application, constructing a standard sit-up posture model specifically includes:

[0034] Video data of multiple subjects who met the testing specifications performing standard sit-ups were collected, and posture frames were sampled for the supine, sit-up, and fall-down transition phases.

[0035] The key pose features in each frame sample are manually annotated, including scapular touch pad state, elbow past knee state, hip distance from the ground, spinal flexion and extension angle, trunk tilt angle, and joint coordinates, to obtain a standard action annotation dataset;

[0036] Based on the national student physical fitness and health standards, quantitative thresholds for compliant actions are set to obtain a rule base.

[0037] The standard action annotation dataset is used as training samples to build an initial model framework. The initial model framework is then trained using machine learning or deep learning based on the training samples to obtain a pose recognition model.

[0038] The posture recognition model is fused with the quantification threshold of compliant actions in the rule base to obtain the standard posture model of sit-up.

[0039] Secondly, embodiments of this application provide a sit-up quality assessment system based on human posture. The system includes a memory and a processor. The memory includes a program for a sit-up quality assessment method based on human posture. When the program for the sit-up quality assessment method based on human posture is executed by the processor, it performs the following steps:

[0040] Obtain the subject ID information, verify the subject's identity based on the subject ID information, obtain the verification result, enter the subject's identity information based on the verification result, configure the test parameters based on the subject's identity information, and obtain the test environment information;

[0041] The subjects were tested based on the test environment information, and human body image information of the subjects was collected in real time. The human body image information of the subjects was preprocessed to obtain optimized image information.

[0042] Based on the optimized image information analysis of the subject's human posture data, key posture features of the sit-up movement are extracted from the human posture data.

[0043] Construct a standard sit-up posture model, set feature thresholds for compliant movements, analyze key posture features of sit-ups based on the standard posture model and compliant movement feature thresholds, and generate test data.

[0044] A multi-dimensional sit-up quality assessment report is generated based on the test data, and the report is transmitted to the terminal in real time.

[0045] Optionally, in the sit-up quality assessment system based on human posture described in this application embodiment, before testing the subject based on the test environment information, the following steps are further included:

[0046] Obtain the body shape parameter information of the subject, which includes at least waist circumference data and height data;

[0047] The body type classification result of the subject is calculated based on the body shape parameter information;

[0048] Obtain the body shape range mapping database, and based on the body shape classification results, obtain the baseline value of the sit-up range of motion and the trunk tilt angle threshold range data that are suitable for the body shape of the subject from the body shape range mapping database.

[0049] Based on the amplitude range benchmark value and the torso tilt angle threshold range data, set the body shape adaptation compliance action feature threshold.

[0050] Optionally, in the sit-up quality assessment system based on human posture described in this application embodiment, the step of analyzing key postural features of sit-up movements based on body shape-adapted compliant movement feature thresholds to generate test data includes:

[0051] Obtain the range of motion data of the subject during the first few effective sit-ups in the current test;

[0052] Calculate the average effective amplitude value based on the preset effective amplitude of the previous several effective movements;

[0053] The dynamic alert threshold is obtained based on the average effective amplitude value and the preset amplitude attenuation coefficient;

[0054] When it is detected that the amplitude of a subsequent sit-up movement is less than the dynamic reminder threshold, an insufficient amplitude reminder message is generated;

[0055] If the amplitude of the consecutive preset number of sit-ups is less than the dynamic reminder threshold and no amplitude improvement trend information is detected, the consecutive preset number of sit-ups will be marked as invalid movements and will not be included in the valid count.

[0056] Thirdly, embodiments of this application also provide a computer-readable storage medium, which includes a sit-up quality assessment method program based on human posture. When the sit-up quality assessment method program based on human posture is executed by a processor, it implements the steps of the sit-up quality assessment method based on human posture as described in any of the above claims.

[0057] As described above, the sit-up quality assessment method, system, and medium based on human posture provided in this application involves: acquiring subject identification information; verifying the subject's identity based on the subject identification information to obtain verification results; recording the subject's identity information based on the verification results; configuring test parameters based on the subject's identity information to obtain test environment information; testing the subject based on the test environment information; collecting the subject's human body image information in real time; preprocessing the subject's human body image information to obtain optimized image information; analyzing the subject's human posture data based on the optimized image information; extracting key posture features of the sit-up movement based on the human posture data; constructing a standard sit-up movement posture model; setting feature thresholds for compliant movements; analyzing the key posture features of the sit-up movement based on the standard movement posture model and the feature thresholds for compliant movements to generate test data; generating a multi-dimensional sit-up movement quality assessment report based on the test data; and transmitting the multi-dimensional sit-up movement quality assessment report to the terminal in real time; and using human posture recognition technology to accurately verify movement requirements, effectively identify various non-compliant movements, and ensure the accuracy of test results. Attached Figure Description

[0058] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 A flowchart of a sit-up quality assessment method based on human posture provided in an embodiment of this application;

[0060] Figure 2 A flowchart illustrating the identity verification process for subjects in the sit-up quality assessment method based on human posture provided in this application embodiment;

[0061] Figure 3 A flowchart illustrating the test environment information acquisition process for the sit-up quality assessment method based on human posture provided in this application embodiment. Detailed Implementation

[0062] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0063] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0064] Please refer to Figure 1 , Figure 1 This is a flowchart of a sit-up quality assessment method based on human posture, as described in some embodiments of this application. This sit-up quality assessment method based on human posture is used in a terminal device and includes the following steps:

[0065] S101, Obtain subject ID information, verify subject identity based on subject ID information, obtain verification results, enter subject identity information based on verification results, configure test parameters based on subject identity information, and obtain test environment information;

[0066] S102, Test the subject according to the test environment information, collect the subject's human body image information in real time, preprocess the subject's human body image information to obtain optimized image information;

[0067] S103, Analyze the subject's human posture data based on optimized image information, and extract key posture features of sit-up movements based on the human posture data.

[0068] S104, Construct a standard sit-up posture model, set feature thresholds for compliant movements, analyze key posture features of sit-ups based on the standard posture model and feature thresholds for compliant movements, and generate test data.

[0069] S105 generates a multi-dimensional sit-up quality assessment report based on the test data and transmits the multi-dimensional sit-up quality assessment report to the terminal in real time.

[0070] It should be noted that the attitude data acquisition method includes at least one of monocular vision acquisition, binocular vision acquisition, depth camera acquisition, or inertial sensor acquisition.

[0071] According to an embodiment of the present invention, before testing the subject based on the test environment information, the method further includes:

[0072] Obtain the subject's body shape parameters, which include at least waist circumference and height data;

[0073] The body type classification results of the subjects are calculated based on body shape parameter information;

[0074] Obtain the body amplitude mapping database, and based on the body type classification results, obtain the baseline values ​​of the sit-up range and the trunk tilt angle threshold range data that are adapted to the body type of the subject from the body amplitude mapping database.

[0075] Set the compliant movement feature thresholds for body shape adaptation based on the amplitude range benchmark value and the torso tilt angle threshold range data.

[0076] It should be noted that body shape parameters can be obtained by having subjects manually input their waist circumference and height data through a pre-test input interface, or by automatically extracting body contours from full-body images and calculating waist circumference and height using image analysis algorithms. Body type classification results can be determined using a combination of BMI and waist-to-body ratio, for example, classifying subjects into four categories: underweight, standard, slightly overweight, and obese. A body shape range mapping database pre-stores the mapping relationships between different body type classifications and corresponding standard sit-up range intervals and trunk tilt angle threshold ranges. This database was established through statistical analysis of a large number of sample test data. For example, the baseline value for the range interval for obese subjects can be set to 80% to 95% of the standard range, with the trunk tilt angle threshold range adjusted accordingly to accommodate the limited movement caused by abdominal fat accumulation in obese individuals.

[0077] Furthermore, a multi-dimensional sit-up performance quality assessment report is generated based on the test data, including:

[0078] A visual posture reconstruction animation of the sit-up movement is generated based on the coordinates and angles of the subject's whole skeletal joints.

[0079] An abdominal muscle function curve is generated based on the peak spinal flexion and extension angle and trunk tilt angle changes of each sit-up.

[0080] The visualization of posture reconstruction animation is synchronously overlaid with the abdominal muscle function curve to generate a visualized motion analysis report that includes posture animation and energy analysis.

[0081] Specifically, the visualized posture reconstruction animation can use skeletal connection animation to recreate the postural changes of the subject during each sit-up in real time, including the continuous trajectory of trunk tilt angle and spinal flexion and extension angle. The abdominal muscle work efficiency curve calculates the peak trunk tilt angle, movement duration, and movement frequency during each sit-up to obtain the relative energy value of the abdominal muscle work, which is presented as a curve. The visualized motion analysis report aligns the posture animation and energy curve on the same timeline, allowing users to intuitively observe the correspondence between the postural performance of each movement and its corresponding work efficiency.

[0082] According to an embodiment of the present invention, key postural features of sit-ups are analyzed based on body shape-adapted compliant movement feature thresholds to generate test data, including:

[0083] Obtain the range of motion data of the subject in the first few effective sit-ups in the current test;

[0084] The average effective amplitude value is calculated based on the preset effective amplitude of the previous few effective movements.

[0085] The dynamic alert threshold is obtained based on the average effective amplitude value and the preset amplitude attenuation coefficient;

[0086] When it is detected that the amplitude of subsequent sit-up movements is less than the dynamic reminder threshold, an insufficient amplitude reminder message is generated;

[0087] If the amplitude of the movement is less than the dynamic reminder threshold for a consecutive preset number of times, and no trend of improvement in amplitude is detected, the consecutive preset number of sit-ups will be marked as invalid movements and will not be included in the valid count.

[0088] It should be noted that the first few preset attempts can use 3-5 valid movements to calculate the average amplitude. The amplitude attenuation coefficient can be set to 5%-15%, and the dynamic alert threshold = average amplitude × (1 - attenuation coefficient). The number of consecutive preset attempts can be 3-5. The amplitude improvement trend refers to the increase in amplitude between adjacent movements exceeding the preset threshold. If there is no improvement, it is judged as fatigue or insufficient amplitude, and the corresponding movement is recorded as invalid to ensure the fairness and accuracy of the test.

[0089] Furthermore, a multi-dimensional sit-up performance quality assessment report is generated based on the test data, including:

[0090] Based on the subjects' body shape parameters, key postural characteristics data of sit-up movements, and test data, the subjects' abdominal muscle strength assessment index, movement standardization index, and physical endurance level were analyzed.

[0091] Based on abdominal muscle strength assessment index, movement standardization index, physical endurance level, and body shape parameters, a personalized fitness suggestion report is generated.

[0092] Specifically, the abdominal muscle strength assessment index is calculated based on the number of effective sit-ups, the average peak trunk inclination angle per repetition, and the speed of movement completion. The movement standardization index is assessed based on the stability of spinal flexion and extension angles, the rate of trunk inclination angle compliance, and the frequency of violations. The physical endurance level is graded based on the total number of effective sit-ups, the average amplitude decay trend of effective movements, and the duration of the test. The personalized fitness recommendation report integrates the above analysis results and provides differentiated training suggestions based on the subject's body type. For example, for obese subjects, it is recommended to start with smaller movement amplitudes and gradually increase them, combined with aerobic exercise for abdominal fat reduction; for underweight subjects, it is recommended to increase the frequency and number of sit-ups to strengthen abdominal muscles. The report may include the abdominal muscle strength assessment conclusions, analysis of movement weaknesses, description of body type characteristics, targeted sit-up training plan suggestions, and daily fitness nutrition guidance.

[0093] like Figure 2 As shown, according to an embodiment of the present invention, obtaining subject identification information, verifying the subject's identity based on the subject identification information, and obtaining the verification result specifically includes:

[0094] Collect subject identification information, which includes at least one of the following: numeric ID, student ID, employee ID, QR code, barcode, IC card information, ID card information, or facial recognition feature information;

[0095] The collected subject identification information is matched and compared with the identity database stored locally or in the cloud to determine whether the matching results are consistent.

[0096] If a match is found, the authentication is considered successful, and the authentication result is obtained.

[0097] If the match is inconsistent, the identity verification is deemed unsuccessful, the test is prohibited from starting, and an identity error message is output.

[0098] It should be noted that identity verification also includes: after successful verification, automatically retrieving historical test data, personal basic information and test permission information corresponding to the subject's ID, which are used for parameter initialization and subsequent result association and storage for this test.

[0099] like Figure 3 As shown in the embodiment of the present invention, the subject's identity information is entered according to the verification results, and the test parameters are configured according to the subject's identity information to obtain the test environment information, specifically including:

[0100] Obtain the verification result and analyze whether the identity verification was successful based on the verification result;

[0101] If the verification is successful, the subject's identity information will be automatically entered. The subject's identity information includes the subject number, name, gender, age, affiliated unit, and test category information.

[0102] Test parameters are matched based on the subject's identity information to obtain test environment information. Test parameters include duration parameters, action threshold parameters, counting rule parameters, and permission parameters.

[0103] If the verification fails, an error message is generated, and the subject number information is re-verified based on the error message.

[0104] It should be noted that, based on the verification results, the subject's identity information is entered, and the test parameters are configured based on the subject's identity information to obtain the test environment information, specifically including:

[0105] Identity information entry: After identity verification, the complete identity information of the subject is automatically entered. The identity information includes subject number, name, gender, age, affiliated unit (school, military, fitness institution, etc.), test category (physical test, assessment, training), and is linked to the subject's historical test data to form a complete subject identity file, ensuring that the identity information is traceable and reproducible.

[0106] Adaptive test parameter configuration: Based on the entered subject identity information, the system automatically matches the corresponding test parameter template, eliminating the need for manual configuration. It also supports manual customization and adjustment, specifically including:

[0107] Duration parameters: Adaptively configured based on the age of the test subject and the test category. For example, the default test duration for primary school students is 1 minute, the default test duration for middle school students is 1 minute, and the training scenario can be customized to be adjusted from 30 seconds to 5 minutes.

[0108] Action threshold parameters: Based on the differences in gender and age of the subjects, the thresholds for compliant action features are adaptively adjusted. For example, the thresholds for judging scapular touch pads and elbows crossing the knees are appropriately relaxed for adolescent subjects.

[0109] Counting rule parameters: Configure counting rules according to the test category. For example, in physical fitness test scenarios, only the number of valid movements is counted, in assessment scenarios, the number of violations and the percentage of invalid movements are recorded, and in training scenarios, movement rhythm reminder parameters can be configured.

[0110] Permission parameters: Configure permissions such as data viewing, report export, and historical data comparison based on the subject's institution and testing permissions. For example, students can only view their individual test reports, while administrators can view batch test data.

[0111] Test environment information calibration and generation: Based on configured test parameters, automatically complete the test environment calibration and generate standardized test environment information, specifically including:

[0112] Acquisition equipment calibration: Perform parameter calibration on the vision acquisition module (camera, depth camera), adjust the focus and exposure to ensure clear and distortion-free attitude data acquisition, and calibrate the sensitivity of the sensor acquisition module to eliminate equipment errors;

[0113] Environmental parameter calibration: Detects the light intensity and background interference at the test site, and automatically adjusts the image preprocessing algorithm parameters (such as noise reduction intensity and image enhancement parameters) to avoid environmental factors affecting detection accuracy;

[0114] Test area calibration: Locate the test pad, delineate the effective area for the subject's movements, and set the area boundary threshold to prevent posture recognition deviation caused by the subject exceeding the test area;

[0115] Environmental information generation: The calibrated equipment parameters, environmental parameters, test area parameters, and test parameters are integrated to generate complete test environment information, which is stored in the data processing unit as the basic environmental basis for this sit-up quality assessment. At the same time, it is synchronized to the terminal interaction module for testers to view and confirm.

[0116] Specifically, based on the verification results, the subject's identity information is entered, and the test parameters are configured according to the subject's identity information to obtain the test environment information, which includes:

[0117] After identity verification, complete identity information consisting of subject number, name, gender, age, affiliated unit, and test category is entered; test parameter templates are adaptively matched based on identity information, and test duration, action threshold, counting rules, and permission parameters are configured, while also supporting manual customization; the data collection device, environmental parameters, and test area are calibrated, and the calibrated parameters are integrated with the configured test parameters to generate standardized test environment information.

[0118] The test parameter template is preset according to the subject's age, gender, and test category. The action threshold parameters for adolescents and adult subjects are set differently, and the counting rule parameters for physical tests, assessments, and training scenarios are different.

[0119] The test environment information includes the calibration parameters of the acquisition equipment, the calibration parameters of the ambient lighting, the boundary parameters of the test area, the test parameters, and the subject's identity information, which are used to ensure the accuracy of posture data acquisition and motion evaluation.

[0120] According to an embodiment of the present invention, the preprocessing of the subject's human image information specifically includes:

[0121] Obtain human body image information of the subject, and perform denoising processing on the human body image information of the subject according to the filtering algorithm to obtain denoised human body image;

[0122] Based on the human target detection algorithm, the denoised human image is analyzed to locate the subject's human body region and background region, the background region is cropped and removed, and the region of interest is extracted.

[0123] The cropped region of interest image is normalized in size and grayscale, and the image size is uniformly adjusted to the preset standard size to obtain standard human image information;

[0124] The human body region is enhanced by applying an image enhancement algorithm to standard human body image information to obtain an enhanced image.

[0125] It should be noted that the preprocessing of the subject's human image information specifically includes:

[0126] Image denoising: Gaussian filtering, median filtering or bilateral filtering algorithms are used to denoise the acquired subject human images, remove Gaussian noise, salt-and-pepper noise and environmental interference noise in the images, avoid misidentification of pose key points caused by noise, ensure uniform distribution of image gray values, and improve image clarity.

[0127] Image cropping and region of interest extraction: Based on the human target detection algorithm, the human body region of the subject is located, and irrelevant background areas in the image (such as walls outside the test pad, ground debris, other people, etc.) are cropped out. The region of interest (ROI) with the human body of the subject as the core is extracted, which narrows the data processing range, reduces the amount of computation for subsequent feature extraction, and improves processing efficiency.

[0128] Image normalization processing: The cropped region of interest image is normalized in size and grayscale. The image size is uniformly adjusted to a preset standard size (such as 640×480 pixels), and the image grayscale value is normalized to the range of 0-255. This eliminates the influence of differences in image size and grayscale under different acquisition devices and lighting conditions, and ensures the consistency of subsequent pose feature extraction.

[0129] Human target detection and background segmentation: Using target detection algorithms such as YOLO and SSD, the human target of the subject in the image is accurately detected and the bounding box of the human target is generated; through semantic segmentation algorithm, the human target and the background are accurately segmented, and the human area and the background area are completely separated to avoid missed detection and false detection of key pose feature points caused by background interference.

[0130] Image enhancement processing: For scenes with uneven lighting and low image contrast, histogram equalization and adaptive contrast enhancement algorithms are used to improve the contrast and detail clarity of human images, highlight key features such as human joints and contours, and provide a clear image foundation for subsequent key pose feature extraction.

[0131] Image distortion correction: If the acquired human body image has lens distortion (such as barrel distortion or pincushion distortion), the distortion parameters obtained by camera calibration are used to correct the distortion of the image, restore the true posture of the subject's human body, and avoid joint position displacement and angle calculation errors caused by distortion.

[0132] According to an embodiment of the present invention, the subject's human posture data is analyzed based on optimized image information, and key posture features of the sit-up movement are extracted based on the human posture data, specifically including:

[0133] Optimized image information is acquired, and the optimized image information is analyzed frame by frame according to the pose estimation algorithm to obtain the whole body skeletal joint coordinates, joint angles and motion trajectory of the subject.

[0134] Key posture features of the upper limbs, lower limbs, and trunk were extracted based on the coordinates, angles, and movement trajectories of the subjects' entire skeletal joints.

[0135] Redundancy removal and anomaly analysis were performed on the key posture features of the upper limbs, lower limbs, and trunk to obtain the key posture features of the sit-up movement.

[0136] It should be noted that, based on the collected posture data and combined with optimized image information, the subject's human posture data was analyzed, and key posture features of the sit-up movement were extracted from the human posture data, specifically including:

[0137] Human posture data analysis: Frame-by-frame analysis of preprocessed and optimized human images is performed. The coordinates, angles and trajectories of the subject's whole body skeletal joints are analyzed using posture estimation algorithms (such as OpenPose and MediaPipe) to reconstruct the posture changes of the subject throughout the entire process of supine and sit-up. The focus is on capturing posture data in the supine, sit-up and fall-down phases. Interference data remaining after image preprocessing is removed to ensure the accuracy and continuity of posture data.

[0138] Upper limb key posture feature extraction: Focusing on the core force exertion and compliance judgment areas of the upper limbs in sit-ups, the following key features are extracted:

[0139] Features of both elbow joints: Extract the real-time three-dimensional coordinates and elbow flexion and extension angles of both elbow joints, record the movement trajectory of the elbow joints during sitting up and falling back, determine whether the elbow joints meet the compliance requirements of "over the knee", and identify whether there are feature changes corresponding to illegal postures such as supporting pads or using leverage in the elbows.

[0140] Bilateral scapular features: Extract the vertex coordinates of the bilateral scapulae and the vertical distance between the scapulae and the test pad, and capture the posture data of whether the scapulae are fully in contact with the pad during the supine stage and whether they are off the pad during the sit-up stage, as the core judgment features for the compliance of supine movements.

[0141] Overall posture characteristics of the upper limbs: Extract the angle between the line connecting the two shoulders and the horizontal plane, and the angle between the arm and the torso to determine whether there are abnormalities such as excessive leverage or movement deformation in the upper limbs, and to help verify the standardization of the movement;

[0142] Lower limb key postural feature extraction: Based on the characteristics of lower limb fixation and force assistance in sit-ups, the following key features are extracted:

[0143] Features of both knee joints: Extract the three-dimensional coordinates and knee flexion-extension angles of both knee joints, record the posture changes of the knee joints throughout the entire movement process, and determine whether there are excessive bending or abnormal extension of the lower limbs that affect the standardization of the movement;

[0144] Ankle joint fixation characteristics: Extract the coordinate changes of both ankle joints to determine whether the ankle joints remain fixed, identify improper postures such as loosening and sliding of the feet, and ensure that the subjects' movements meet the test standards;

[0145] Hip posture characteristics: Extract the coordinates of the hip joint and the hip flexion and extension angles to capture the force exertion posture of the hip during the sitting-up phase and determine whether there are violations such as lifting the buttocks off the mat or excessive force exertion of the hip.

[0146] Extraction of key trunk posture features: As the core force exertion area in sit-ups, the focus is on extracting the posture features of the trunk, including:

[0147] Spinal flexion and extension angle: By extracting the coordinates of key joints in the cervical, thoracic, and lumbar vertebrae, the flexion and extension angle of the spine is calculated, the bending range of the trunk is quantified, and it is determined whether the range of motion of the sit-up is up to standard, so as to avoid violations of insufficient range of motion.

[0148] Hip position characteristics: Extract the three-dimensional coordinates of the center point of the hip, record the vertical displacement of the hip throughout the entire movement, determine whether the hip is lifted off the test mat, and identify illegal movements such as using leverage to lift the hip;

[0149] Trunk tilt angle: Extract the angle between the trunk and the horizontal plane, capture the speed at which the trunk rises and falls during the sitting-up phase, and help analyze the continuity of the movement and the uniformity of force exertion.

[0150] Posture feature screening and optimization: All extracted key posture features are screened to remove redundant and abnormal features, and retain the core features related to the compliance and quality assessment of sit-up movements. Through feature normalization, feature bias caused by differences in body shape and height of different subjects is eliminated to ensure the consistency of feature data and provide accurate feature basis for subsequent standard posture model construction and movement compliance verification.

[0151] According to an embodiment of the present invention, a standard sit-up posture model is constructed, specifically including:

[0152] Video data of multiple subjects who met the testing specifications performing standard sit-ups were collected, and posture frames were sampled for the supine, sit-up, and fall-down transition phases.

[0153] The key pose features in each frame sample are manually annotated, including scapular touch pad state, elbow past knee state, hip distance from the ground, spinal flexion and extension angle, trunk tilt angle, and joint coordinates, to obtain a standard action annotation dataset;

[0154] Based on the national student physical fitness and health standards, quantitative thresholds for compliant actions are set to obtain a rule base.

[0155] The standard action annotation dataset is used as training samples to build an initial model framework. The initial model framework is then trained using machine learning or deep learning based on the training samples to obtain a pose recognition model.

[0156] The posture recognition model is fused with the quantification threshold of compliant actions in the rule base to obtain the standard posture model of sit-up.

[0157] It should be noted that constructing a standard sit-up posture model specifically includes:

[0158] Standard movement sample collection and annotation: Video data of standard sit-ups were collected from multiple subjects who met the testing specifications. Posture frames were sampled for the supine, sit-up, and return transition phases. Key posture features in each frame were manually annotated, including scapular touch pad status, elbow past knee status, hip distance from the ground, spinal flexion and extension angles, trunk tilt angle, joint coordinates, etc., to form a standard movement annotation dataset.

[0159] The action phase division and feature normalization divides the sit-up action into the following stages according to the movement sequence: the initial stage of supine lying down, the trunk lifting stage, the stage of sitting up to the target, and the trunk falling back down. The key posture features of subjects with different heights and body types are normalized by coordinate and angle to eliminate the influence of individual body type differences on model judgment and form a standard feature space with a unified scale.

[0160] Establish a rule base for judging compliant postures. Based on the National Student Physical Fitness and Health Standards, set quantitative thresholds for compliant actions to form a rule base, including:

[0161] Supine position: If the vertical distance between the scapula and the mat surface is less than the first threshold, it is considered a valid scapular contact with the mat.

[0162] The sitting-up phase is determined when the horizontal coordinate of the elbow joint exceeds the coordinate of the knee joint and the trunk tilt angle is greater than the second threshold.

[0163] Process constraints: If the vertical displacement of the hip is less than the third threshold, it is determined that the hip lift is not in violation of the rules; if the ankle joint displacement is less than the fourth threshold, it is determined that both feet are fixed.

[0164] The pose classification and discrimination model is trained using labeled standard pose features as training samples. Machine learning or deep learning methods are used to train the pose discrimination model so that the model can:

[0165] Distinguish between valid and invalid actions;

[0166] Identify improper postures such as insufficient range of motion, hips off the ground, elbows not extending past the knees, and shoulder blades not touching the mat;

[0167] Output the compliance probability of each frame's pose and the confidence level of the action phase.

[0168] Model fusion and dynamic calibration integrate rule-based hard threshold judgment with data-based soft discriminant model to form the final sit-up standard posture model; the model supports dynamic loading of corresponding threshold parameters according to different test scenarios (student physical tests, daily training) to achieve adaptive adjustment of standard strictness.

[0169] Model validation and iterative optimization use test set samples to validate the model's accuracy and recall. Incremental samples are used to supplement training for scenarios with false positives and false negatives, continuously optimizing the model's discrimination accuracy to ensure that it can stably recognize the standard sit-up posture under different lighting conditions, body types, and movement rhythms.

[0170] According to an embodiment of the present invention, the generation and real-time transmission of the quality assessment report specifically includes:

[0171] By performing multi-dimensional analysis of test data, a comprehensive quality assessment report is generated and transmitted in real time to a designated terminal, meeting the needs for viewing, archiving, and providing training guidance for test results. Specifically, this includes:

[0172] Multi-dimensional Sit-up Performance Quality Assessment Report Generation: Based on complete test data generated by S104, combined with subject identity information and historical test data, a multi-dimensional and personalized quality assessment report is generated. The report covers the following core contents, balancing professionalism and readability:

[0173] Basic test results: Clearly define the test duration, number of valid actions, total number of attempts, and compliance rate, present the test results of the subject in a clear and intuitive way, and compare them with the subject's historical test data to analyze the trend of performance changes;

[0174] Movement standardization assessment: Statistically analyze the frequency and percentage of various violations, identify the main types of violations (such as frequent occurrences of "elbow not extending past the knee" and "scapula not touching the mat"), and clarify the core defects in the subjects' movements; quantify the compliance rate of movement range and joint angle, and accurately describe the degree of movement standardization;

[0175] Movement quality assessment: Analyze the continuity of movement (average duration of a single movement, range of movement rhythm fluctuation) and the uniformity of force exertion (rate of change of trunk tilt angle, trend of joint angle change) to determine the subject's abdominal muscle endurance and movement control ability.

[0176] Personalized training recommendations: Targeted training recommendations are generated based on the subject's motor deficiencies. For example, for "elbows not going past the knees", it is recommended to strengthen upper limb flexibility training; for "insufficient range of motion", it is recommended to strengthen abdominal and core strength training; and for "disordered rhythm", it is recommended to practice steady-speed movements. At the same time, based on historical test data, phased training goals are given.

[0177] Data archiving information: includes test environment information, test data details, and evaluation conclusions, which facilitates subsequent data traceability and batch statistical analysis (such as summarizing school physical test data).

[0178] Real-time transmission of assessment reports to the terminal: After generating the assessment report, the system transmits the multi-dimensional sit-up quality assessment report to the designated terminal in real time according to the configured transmission parameters. The transmission methods support wired transmission and wireless transmission (WiFi, Bluetooth) to adapt to different scenario requirements.

[0179] Tester terminal: Transmits complete evaluation reports, allowing testers to view subject test results and violation details in real time, and complete test record archiving;

[0180] Management terminal (school physical fitness test platform): transmits standardized evaluation reports, facilitating batch statistics, score ranking, and data archiving;

[0181] Subject terminal (personal mobile phone, tablet): Transmits a simplified assessment report (core performance, motor deficiencies, training suggestions) to facilitate subjects' viewing of their individual test results and targeted training;

[0182] Meanwhile, the evaluation report supports export (PDF and Excel formats) and printing, meeting the data archiving and usage needs of different scenarios and ensuring that the test results are traceable and usable.

[0183] In one specific embodiment of the present invention, it is applied to the physical fitness test scenario for middle school students (adapted to the national student physical fitness and health standards).

[0184] This embodiment is applied to the physical fitness test of ordinary middle school students. The subjects are middle school students aged 14-16. The purpose of the test is to complete the semester physical fitness test, generate a physical fitness test report, and synchronize it to the school's physical fitness test management platform. The specific implementation process is as follows:

[0185] The system collects the student ID (number information) of the test taker using a campus card reader, compares the student ID with the school's pre-stored student identity information database (including name, gender, grade, class, etc.), and automatically enters the complete identity information of the test taker after successful verification. It also retrieves the student's sit-up test data from the previous semester. Based on the characteristics of middle school students, the system adaptively matches the test parameter template: the timing parameter is set to 1 minute, the action judgment threshold is appropriately relaxed (the vertical distance between the shoulder blade and the mat surface ≤8mm is considered touching the mat, and the angle between the torso and the horizontal plane ≥25° is considered sitting up to the correct position), the counting rule only counts the number of valid movements, and the transmission parameters are set to synchronize to the school's physical testing platform, the tester's terminal, and the student's parent's mobile terminal. The system automatically completes the test environment calibration, adjusts the camera focal length to 1.5m, adapts the exposure to the natural light environment of the classroom or playground, delineates the single-person test area (1.2m×0.8m), and generates complete test environment information.

[0186] After the tester confirmed that the test environment was correct, they issued a start command. The subject lay supine on the test mat with their hands behind their head and their knees bent. The system simultaneously started a 1-minute timer and image acquisition (acquisition frequency 30 frames / second). During the acquisition process, if the subject went out of the test area, the system immediately issued a voice prompt "Please return to the test area". The acquired raw images were preprocessed: Gaussian filtering was used to remove noise, the human body area was cropped based on the YOLO algorithm, normalized to 640×480 pixels, semantic segmentation was used to separate the human body from the playground background, histogram equalization was used to enhance image contrast, and lens distortion was corrected based on camera calibration parameters to finally obtain optimized image information.

[0187] The MediaPipe pose estimation algorithm was used to analyze the optimized images frame by frame, extracting the coordinates and angles of key joints such as the elbows, scapulae, knees, ankles, and spine of the subjects. The key pose features of the upper limbs (elbows above the knee, scapulae touching the pad), trunk (spinal flexion and extension angles, trunk tilt angles), and lower limbs (ankle displacement) were extracted. Abnormal features caused by fluctuations in the range of student movements were removed, and after normalization, standardized feature data was formed.

[0188] A standard posture model for sit-ups was constructed for middle school students. Standard movement samples were collected from 500 middle school students who met the national student physical fitness standards. Postural characteristics at each stage were labeled, dividing the exercise into four stages: initial supine position, trunk raising, sitting up to standard, and trunk lowering. A compliance threshold rule base was established. The extracted subject characteristics were compared with the model and thresholds to determine the compliance of the movements in real time: Valid movements were defined as follows: elbows extending past the knees during sitting up, shoulder blades touching the pad during supine position, hips not lifted (vertical displacement ≤10mm), and feet fixed (ankle joint displacement ≤3mm), with the count incremented by 1. Invalid movements were defined as shoulder blades not touching the pads or elbows not extending past the knees, with no count and a real-time voice prompt: "The movement is not standard, please adjust." After one minute of timing, the system locked the test data, including the number of valid movements (e.g., 42), the number of invalid movements (e.g., 3), the compliance rate (93.3%), the type of violation (2 instances of shoulder blades not touching the pads, 1 instance of elbows not extending past the knees), and quantitative data for each posture.

[0189] Based on the test data and the student's physical fitness test scores (38 times) from the previous semester, a multi-dimensional assessment report was generated: The basic test results showed 42 effective movements, reaching a good level, an improvement of 10.5% compared to the previous semester; the movement standardization assessment pointed out that the core defect was that the scapula did not fully touch the mat when lying supine; the movement quality assessment showed that the movement rhythm was basically even, with an average single movement duration of 1.4 seconds; the personalized training suggestion was to practice supine leg raises daily to strengthen abdominal strength and correct the scapula-touching posture; the report was transmitted in real time to the school's physical fitness test platform (for batch statistics), the tester's terminal (for archiving), and the parents' mobile phones (for viewing), and supported PDF export and printing, completing the closed loop of the physical fitness test process.

[0190] This embodiment is adapted to the needs of large-scale and efficient physical fitness testing in middle schools. It can complete 2,000 error-free tests per day, and the data reporting efficiency is 60% higher than that of traditional manual testing, effectively solving the pain points of large errors in manual counting and cumbersome recording.

[0191] In one specific embodiment of the present invention, it is applied to a community fitness scenario (to meet the personalized training needs of the public).

[0192] This embodiment was applied in a community fitness center, with participants being community residents (aged 18-60 years). The purpose of the test was to provide personalized training guidance to help residents standardize their sit-up movements and improve training effectiveness. The specific implementation process is as follows:

[0193] By scanning the QR code (number information) of a resident's fitness card with a barcode scanner, identity verification is completed, and information such as the resident's name, age, gender, and fitness level is automatically entered, and their historical training data is retrieved. Based on the personalized needs of mass fitness, test parameters are configured: timing parameters are selected by the resident (1 minute, 2 minutes, or 3 minutes), the action judgment threshold is adapted to the resident's age (young residents ≤5mm, middle-aged and elderly residents ≤10mm), the counting rules are configured with action rhythm reminders (an action speed prompt is given every 10 seconds), and the transmission parameters are set to synchronize to the resident's mobile phone WeChat mini-program and the community fitness center management terminal. The system adopts lightweight calibration, adapts to the lighting environment of the community fitness center, and the camera quickly completes focus and exposure calibration, generates test environment information, and supports rapid deployment in 3 minutes.

[0194] Residents initiate the test independently via the terminal, entering the test mat and maintaining an initial supine posture. The system then starts timing and image acquisition (acquisition frequency 30 frames / second). During the acquisition process, if the image is blurry (e.g., the resident's movements are too fast), the system automatically adjusts the exposure to ensure clear acquisition. The original image undergoes preprocessing: median filtering is used for noise reduction, the human body area is cropped, and normalization is performed to complete background segmentation and image enhancement, adapting to the complex background environment of the community fitness center, resulting in optimized image information.

[0195] The MediaPipe lightweight pose estimation algorithm is used to analyze and optimize images, extract key pose features, focus on common movement defects among residents (such as excessive arching of the waist and elbows not reaching the knees), optimize the feature data, eliminate bias caused by body shape differences, and generate standardized feature data.

[0196] A standard posture model for sit-ups in mass fitness was constructed, and standard movement samples of residents of different ages and body types were collected. Differentiated compliance thresholds were set, and the model supports adaptive adjustment of movement rhythm. The residents' characteristics were compared with the model and thresholds to determine the compliance of the movements in real time. If the waist was excessively arched or the elbows did not go past the knees, a voice prompt was immediately issued saying "Relax your waist and elbows past your knees". After the test, test data was generated: number of effective movements, type of violation, compliance rate of movement range, average movement rhythm, etc., and muscle force-related posture data of residents during training were recorded simultaneously.

[0197] The system generates personalized training assessment reports, analyzes movement deficiencies based on residents' fitness goals (such as fat loss and core strength enhancement), and provides targeted training suggestions (e.g., for excessive arching of the lower back, it suggests strengthening lower back stretching exercises; for disordered movement rhythm, it suggests using a metronome for assisted practice). The reports are transmitted to residents' mobile phone mini-programs in real time, supporting the viewing and comparison of historical training data and the creation of training plans. The community fitness center management terminal can view residents' training status and provide targeted guidance. The reports can be exported, allowing residents to easily review training priorities at any time.

[0198] This embodiment is adapted to the lightweight and personalized needs of community fitness. The equipment is portable, easy to deploy, and simple to operate. It can effectively improve the standardization of residents' training and help implement the national fitness policy. After pilot application, the residents' training retest rate increased by 40%.

[0199] In summary, the method of this invention is feasible and adaptable. It can flexibly adjust parameter configuration, model threshold and report content according to different scenario requirements, realize intelligent and personalized sit-up testing and evaluation, cover mainstream application scenarios such as student physical testing and community fitness, and has broad practicality and promotion value.

[0200] Secondly, embodiments of this application provide a sit-up quality assessment system based on human posture. The system includes a memory and a processor. The memory includes a program for a sit-up quality assessment method based on human posture. When the program for the sit-up quality assessment method based on human posture is executed by the processor, it performs the following steps:

[0201] Obtain the subject ID information, verify the subject's identity based on the subject ID information, obtain the verification result, enter the subject's identity information based on the verification result, configure the test parameters based on the subject's identity information, and obtain the test environment information;

[0202] The subjects were tested based on the test environment information, and human body image information of the subjects was collected in real time. The human body image information of the subjects was preprocessed to obtain optimized image information.

[0203] Based on the optimized image information analysis of the subject's human posture data, key posture features of the sit-up movement are extracted from the human posture data.

[0204] Construct a standard sit-up posture model, set feature thresholds for compliant movements, analyze key posture features of sit-ups based on the standard posture model and compliant movement feature thresholds, and generate test data.

[0205] A multi-dimensional sit-up quality assessment report is generated based on the test data, and the report is transmitted to the terminal in real time.

[0206] According to an embodiment of the present invention, obtaining subject identification information, verifying the subject's identity based on the subject identification information, and obtaining the verification result specifically includes:

[0207] Collect subject identification information, which includes at least one of the following: numeric ID, student ID, employee ID, QR code, barcode, IC card information, ID card information, or facial recognition feature information;

[0208] The collected subject identification information is matched and compared with the identity database stored locally or in the cloud to determine whether the matching results are consistent.

[0209] If a match is found, the authentication is considered successful, and the authentication result is obtained.

[0210] If the match is inconsistent, the identity verification is deemed unsuccessful, the test is prohibited from starting, and an identity error message is output.

[0211] According to an embodiment of the present invention, subject identity information is entered based on the verification results, test parameters are configured based on the subject identity information, and test environment information is obtained, specifically including:

[0212] Obtain the verification result and analyze whether the identity verification was successful based on the verification result;

[0213] If the verification is successful, the subject's identity information will be automatically entered. The subject's identity information includes the subject number, name, gender, age, affiliated unit, and test category information.

[0214] Test parameters are matched based on the subject's identity information to obtain test environment information. Test parameters include duration parameters, action threshold parameters, counting rule parameters, and permission parameters.

[0215] If the verification fails, an error message is generated, and the subject number information is re-verified based on the error message.

[0216] It should be noted that the system consists of the following modules:

[0217] Identity verification and parameter configuration module: used for subject identity verification, number entry, and test parameter configuration, supporting multiple identity entry methods and adapting to different testing scenarios;

[0218] Posture data acquisition module: including a visual acquisition unit (high-definition camera / depth camera) and / or a multi-sensor fusion unit, used to acquire the subject's full-body posture data and key joint motion trajectories in real time;

[0219] Posture feature extraction and modeling module: used to extract the core feature points of sit-up movements, build a standard movement posture model, and define feature thresholds for compliant movements;

[0220] Action verification and counting module: Based on the standard posture model, it performs frame-by-frame compliance verification of real-time actions, automatically completes the counting of valid actions, and identifies and marks illegal actions;

[0221] Timing and Process Control Module: Used for test timing, automatically controlling the test process, issuing prompts and locking results when the test ends;

[0222] Quality assessment and report generation module: Generates multi-dimensional motion quality assessment reports based on test data and provides personalized training suggestions;

[0223] Data storage and management module: used for storing, tracing, and statistically analyzing all test data, supporting cloud synchronization and multi-terminal access;

[0224] Terminal interaction module: includes a touch operation terminal, a display unit, and a voice prompt unit, used for tester operation, subject status display, and output of test instructions and prompts.

[0225] Furthermore, the attitude data acquisition module is also compatible with existing infrared and belt-type testers, enabling multi-source data fusion verification and improving detection accuracy.

[0226] Integration with infrared testing instruments: Cross-validation of infrared grating detection data with human posture recognition data avoids misjudgments caused by ambient light interference;

[0227] Integration with belt-type testing device: Combine waist movement trajectory data with upper limb posture data to achieve full-dimensional movement verification and eliminate "false movement" counting.

[0228] According to an embodiment of the present invention, it also includes an anti-cheating identification unit for real-time identification of violations such as hips off the ground, elbows supporting the pad, insufficient range of motion, and loose feet.

[0229] Furthermore, the system also includes an anti-cheating control unit, which uses gesture recognition to identify subjects' violations in real time, issues voice warnings in real time, records the number of violations, and ensures the fairness of the test.

[0230] According to an embodiment of the present invention, a real-time prompting module is further included, which includes a voice broadcasting unit, a display screen unit, and an indicator light unit.

[0231] According to an embodiment of the present invention, a data synchronization module is also included, which is used to upload test data to the school's physical fitness testing platform or cloud management platform.

[0232] In summary, the beneficial effects of the present invention are as follows:

[0233] Significantly improved detection accuracy: Through human posture recognition technology, core action requirements such as "elbows past the knees and shoulder blades touching the pad" are accurately verified, completely solving the problems of manual counting errors and misjudgment by traditional equipment, effectively identifying various violations and ensuring the accuracy of test results;

[0234] Intelligent management and control throughout the entire process: It realizes full automation of identity verification, data collection, counting, evaluation and report generation, eliminating the need for manual proctoring, greatly reducing labor costs and improving testing efficiency;

[0235] Multi-dimensional quality assessment: Breaking through the limitations of traditional equipment that can only count, it realizes multi-dimensional assessment of movement standardization, continuity, force exertion, etc., providing targeted training guidance for subjects and empowering physical training;

[0236] Strong compatibility and scalability: It is compatible with existing infrared and belt-type testers, enabling multi-source data fusion, while supporting cloud data management and multi-terminal access, and is suitable for various application scenarios such as schools and fitness institutions;

[0237] Significantly enhanced anti-cheating capabilities: Through real-time posture recognition, various cheating actions are accurately identified, and real-time warnings and records are provided, eliminating violations from a technical perspective and ensuring the fairness of the test.

[0238] A third aspect of the present invention provides a computer-readable storage medium including a sit-up quality assessment method program based on human posture. When the sit-up quality assessment method program based on human posture is executed by a processor, it implements the steps of the sit-up quality assessment method based on human posture as described above.

[0239] This invention discloses a method, system, and medium for assessing sit-up quality based on human posture. The method involves: acquiring subject identification information; verifying subject identity based on subject identification information to obtain verification results; recording subject identity information based on verification results; configuring test parameters based on subject identity information to obtain test environment information; conducting tests on subjects based on the test environment information; acquiring subject human body image information in real time; preprocessing subject human body image information to obtain optimized image information; analyzing subject human posture data based on optimized image information; extracting key posture features of sit-up movements based on human posture data; constructing a standard sit-up movement posture model; setting feature thresholds for compliant movements; analyzing key posture features of sit-up movements based on the standard movement posture model and compliant movement feature thresholds to generate test data; generating a multi-dimensional sit-up movement quality assessment report based on the test data; and transmitting the multi-dimensional sit-up movement quality assessment report to the terminal in real time. Through human posture recognition technology, the method accurately verifies movement requirements, effectively identifies various non-compliant movements, and ensures the accuracy of test results.

[0240] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0241] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0242] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0243] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0244] Alternatively, if the integrated units of the present invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A method for assessing the quality of sit-ups based on human posture, characterized in that, include: Obtain the subject ID information, verify the subject's identity based on the subject ID information, obtain the verification result, enter the subject's identity information based on the verification result, configure the test parameters based on the subject's identity information, and obtain the test environment information; The subjects were tested based on the test environment information, and human body image information of the subjects was collected in real time. The human body image information of the subjects was preprocessed to obtain optimized image information. Based on the optimized image information analysis of the subject's human posture data, key posture features of the sit-up movement are extracted from the human posture data. Construct a standard sit-up posture model, set feature thresholds for compliant movements, analyze key posture features of sit-ups based on the standard posture model and compliant movement feature thresholds, and generate test data. A multi-dimensional sit-up quality assessment report is generated based on the test data, and the report is transmitted to the terminal in real time.

2. The sit-up quality assessment method based on human posture according to claim 1, characterized in that, Before testing the subject based on the test environment information, the method further includes: Obtain the body shape parameter information of the subject, which includes at least waist circumference data and height data; The body type classification result of the subject is calculated based on the body shape parameter information; Obtain the body shape range mapping database, and based on the body shape classification results, obtain the baseline value of the sit-up range of motion and the trunk tilt angle threshold range data that are suitable for the body shape of the subject from the body shape range mapping database. Based on the amplitude range benchmark value and the torso tilt angle threshold range data, set the body shape adaptation compliance action feature threshold.

3. The sit-up quality assessment method based on human posture according to claim 2, characterized in that, The key postural features of sit-ups are analyzed based on the body shape-adapted compliant movement feature threshold to generate test data, including: Obtain the range of motion data of the subject during the first few effective sit-ups in the current test; Calculate the average effective amplitude value based on the preset effective amplitude of the previous several effective movements; The dynamic alert threshold is obtained based on the average effective amplitude value and the preset amplitude attenuation coefficient; When it is detected that the amplitude of a subsequent sit-up movement is less than the dynamic reminder threshold, an insufficient amplitude reminder message is generated; If the amplitude of the consecutive preset number of sit-ups is less than the dynamic reminder threshold and no amplitude improvement trend information is detected, the consecutive preset number of sit-ups will be marked as invalid movements and will not be included in the valid count.

4. The sit-up quality assessment method based on human posture according to claim 3, characterized in that, Preprocessing of the subject's human image information specifically includes: Obtain human body image information of the subject, and perform denoising processing on the human body image information of the subject according to the filtering algorithm to obtain denoised human body image; Based on the human target detection algorithm, the denoised human image is analyzed to locate the subject's human body region and background region, the background region is cropped and removed, and the region of interest is extracted. The cropped region of interest image is normalized in size and grayscale, and the image size is uniformly adjusted to the preset standard size to obtain standard human image information; The human body region is enhanced by applying an image enhancement algorithm to standard human body image information to obtain an enhanced image.

5. The sit-up quality assessment method based on human posture according to claim 4, characterized in that, Based on the optimized image information analysis of the subject's human posture data, key postural features of the sit-up movement are extracted from the human posture data, specifically including: Optimized image information is acquired, and the optimized image information is analyzed frame by frame according to the pose estimation algorithm to obtain the whole body skeletal joint coordinates, joint angles and motion trajectory of the subject. Key posture features of the upper limbs, lower limbs, and trunk were extracted based on the coordinates, angles, and movement trajectories of the subjects' entire skeletal joints. Redundancy removal and anomaly analysis were performed on the key posture features of the upper limbs, lower limbs, and trunk to obtain the key posture features of the sit-up movement.

6. The sit-up quality assessment method based on human posture according to claim 5, characterized in that, Constructing a standard sit-up posture model, specifically including: Video data of multiple subjects who met the testing specifications performing standard sit-ups were collected, and posture frames were sampled for the supine, sit-up, and fall-down transition phases. The key pose features in each frame sample are manually annotated, including scapular touch pad state, elbow past knee state, hip distance from the ground, spinal flexion and extension angle, trunk tilt angle, and joint coordinates, to obtain a standard action annotation dataset; Based on the national student physical fitness and health standards, quantitative thresholds for compliant actions are set to obtain a rule base. The standard action annotation dataset is used as training samples to build an initial model framework. The initial model framework is then trained using machine learning or deep learning based on the training samples to obtain a pose recognition model. The posture recognition model is fused with the quantification threshold of compliant actions in the rule base to obtain the standard posture model of sit-up.

7. A sit-up quality assessment system based on human posture, characterized in that, The system includes a memory and a processor. The memory contains a program for a sit-up quality assessment method based on human posture. When the program for the sit-up quality assessment method based on human posture is executed by the processor, it performs the following steps: Obtain the subject ID information, verify the subject's identity based on the subject ID information, obtain the verification result, enter the subject's identity information based on the verification result, configure the test parameters based on the subject's identity information, and obtain the test environment information; The subjects were tested based on the test environment information, and human body image information of the subjects was collected in real time. The human body image information of the subjects was preprocessed to obtain optimized image information. Based on the optimized image information analysis of the subject's human posture data, key posture features of the sit-up movement are extracted from the human posture data. Construct a standard sit-up posture model, set feature thresholds for compliant movements, analyze key posture features of sit-ups based on the standard posture model and compliant movement feature thresholds, and generate test data. A multi-dimensional sit-up quality assessment report is generated based on the test data, and the report is transmitted to the terminal in real time.

8. The sit-up quality assessment system based on human posture according to claim 7, characterized in that, Before testing the subject based on the test environment information, the method further includes: Obtain the body shape parameter information of the subject, which includes at least waist circumference data and height data; The body type classification result of the subject is calculated based on the body shape parameter information; Obtain the body shape range mapping database, and based on the body shape classification results, obtain the baseline value of the sit-up range of motion and the trunk tilt angle threshold range data that are suitable for the body shape of the subject from the body shape range mapping database. Based on the amplitude range benchmark value and the torso tilt angle threshold range data, set the body shape adaptation compliance action feature threshold.

9. The sit-up quality assessment system based on human posture according to claim 8, characterized in that, The key postural features of sit-ups are analyzed based on the body shape-adapted compliant movement feature threshold to generate test data, including: Obtain the range of motion data of the subject during the first few effective sit-ups in the current test; Calculate the average effective amplitude value based on the preset effective amplitude of the previous several effective movements; The dynamic alert threshold is obtained based on the average effective amplitude value and the preset amplitude attenuation coefficient; When it is detected that the amplitude of a subsequent sit-up movement is less than the dynamic reminder threshold, an insufficient amplitude reminder message is generated; If the amplitude of the consecutive preset number of sit-ups is less than the dynamic reminder threshold and no amplitude improvement trend information is detected, the consecutive preset number of sit-ups will be marked as invalid movements and will not be included in the valid count.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a sit-up quality assessment method program based on human posture. When the sit-up quality assessment method program based on human posture is executed by a processor, it implements the steps of the sit-up quality assessment method based on human posture as described in any one of claims 1 to 6.

Images