Intelligent playground motion posture monitoring system and method based on visual identification

Abstract

The invention discloses an intelligent playground motion posture monitoring system and method based on visual identification, relates to the technical field of motion posture monitoring, and solves the technical problem that in the prior art, individual differences are ignored, so that the deviation between an actual elbow joint angle and a standard angle is misjudged as a wrong action, and inaccurate feedback is given. The method comprises the following steps: acquiring video data and personnel data of target motion; analyzing the video data to obtain joint data and respiration data; calculating a motion posture evaluation coefficient based on the joint data; calculating a respiration evaluation coefficient based on the respiration data; constructing a correction model based on personnel data to correct the motion posture evaluation coefficient to obtain a target motion posture evaluation coefficient; judging whether the motion posture of the target motion is qualified or not based on the target motion posture evaluation coefficient; and judging whether the breathing frequency of the sportsman is qualified or not based on the breathing evaluation coefficient. The technical problem is solved.

Description

Technical Field

[0001] The present invention belongs to the field of sports posture monitoring, and specifically relates to a monitoring system and method for sports posture in a smart playground based on visual recognition. Background Art

[0002] With the advancement of science and technology and people's emphasis on healthy living, the importance of physical exercise has become increasingly prominent. However, incorrect exercise posture will not only reduce the effect of exercise, but may also cause sports injuries. Traditional sports posture assessment methods mainly rely on the coach's experience or manual video analysis. This method is not only time-consuming and laborious, but also difficult to ensure the consistency and accuracy of the assessment results.

[0003] Existing technologies mainly collect images of athletes, capture and accurately analyze the sports postures in the images in real time, and compare them with standard sports posture models to quickly and accurately identify sports posture deviations and provide targeted error correction suggestions. However, existing technologies ignore individual differences. Everyone's arm and leg lengths are different. For example, for athletes with different arm lengths, when performing certain actions (such as basketball shooting or pull-ups), in order to achieve the same action effect, the angles at which the elbow joints need to be bent will be different. If individual arm length differences are ignored and a unified standard angle is used for evaluation, the deviation between the actual elbow joint angle and the standard angle will be misjudged as an incorrect action, thereby giving inaccurate feedback.

[0004] Therefore, the present invention proposes a monitoring system and method for smart playground motion posture based on visual recognition to solve the above problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a monitoring system and method for smart playground sports posture based on visual recognition, which is used to solve the problem that the prior art ignores individual differences. Everyone's arm length and leg length are different. For example, for athletes with different arm lengths, when performing certain actions, in order to achieve the same action effect, the angles at which the elbow joints need to be bent will be different; if the individual arm length differences are ignored and a unified standard angle is used for evaluation, the deviation between the actual elbow joint angle and the standard angle will be misjudged as an erroneous action, thereby giving inaccurate feedback.

[0006] To achieve the above-mentioned object, the first aspect of the present invention provides a monitoring system for sports postures in a smart playground based on visual recognition, comprising: a data acquisition module, a data analysis module, a correction module and a monitoring module;

[0007] Data acquisition module: used to obtain the video data and personnel data of the target movement in real time;

[0008] Data analysis module: by analyzing the video data, obtain joint data and breathing data; calculate the motion posture evaluation coefficient based on the joint data; and,

[0009] A breathing assessment coefficient is calculated based on the breathing data;

[0010] Correction module: constructs a correction model based on personnel data to correct the motion posture evaluation coefficient and obtain the target motion posture evaluation coefficient;

[0011] Monitoring module: judge whether the target movement posture is qualified based on the target movement posture evaluation coefficient; judge whether the breathing rate of the exerciser is qualified based on the breathing evaluation coefficient.

[0012] Preferably, the acquisition of the target motion video data and personnel data includes:

[0013] The data acquisition equipment is used to obtain in real time the video data of athletes performing target sports on the smart playground and the personnel data of the athletes; the target sports refers to the type of sports performed by the athletes; the personnel data include: forearm length, upper arm length, thigh length and calf length.

[0014] Preferably, the analyzing of the video data includes:

[0015] Extract the image of each frame in the video data;

[0016] Extract human body information from each frame of the image and build a human skeleton model based on the human body information; the human body information includes: shoulders, elbows, wrists, hips, knees, ankles and toes;

[0017] A spatial rectangular coordinate system is constructed by taking any position in the human skeleton model as the origin;

[0018] Obtain the corresponding coordinates of the human body information in each frame of the image based on a spatial rectangular coordinate system;

[0019] Based on the corresponding coordinates of the human body information, the joint data corresponding to the human body information in each frame of the image is calculated by the cosine theorem; wherein the joint data includes: elbow joint angle, knee joint angle and ankle joint angle;

[0020] The breathing data is obtained by analyzing the image of each frame in the video data.

[0021] Preferably, the analyzing of each frame of the video data includes:

[0022] The image of each frame in the video data is analyzed by a stereo matching algorithm to obtain three-dimensional point cloud data of the moving person; wherein the stereo matching algorithm includes: local stereo matching, global stereo matching or deep learning stereo matching algorithm;

[0023] Based on the three-dimensional point cloud data, a three-dimensional model of the moving person is constructed by a three-dimensional reconstruction algorithm; wherein the three-dimensional reconstruction algorithm includes: a Poisson surface reconstruction algorithm or an active three-dimensional reconstruction algorithm;

[0024] The volume change of the target area in the three-dimensional model corresponding to each frame of the image is obtained by a three-dimensional tracking algorithm, and a volume change curve of the target area corresponding to the video data is obtained; wherein the three-dimensional tracking algorithm includes: tracking of feature points, optical flow tracking or tracking algorithm driven by deep learning; the target area refers to the chest of the athlete;

[0025] The volume change curve is converted from the time domain to the frequency domain by fast Fourier transform to obtain the corresponding frequency axis;

[0026] Extract the positive frequency part of the rate axis, and extract the maximum frequency point in the positive frequency part;

[0027] The frequency corresponding to the maximum frequency point is converted into the respiratory frequency and marked as respiratory data.

[0028] It should be noted that the frequency corresponding to the maximum amplitude found in the frequency domain is the breathing frequency, and the Fourier transform converts the signal in the time domain to the frequency domain, so that the frequency components of the signal can be analyzed; breathing is a periodic physiological process, and its characteristic frequency corresponds to the breathing rate (i.e., the number of breaths per minute); through the Fourier transform, the original volume change curve is decomposed into a series of sine waves and cosine waves of different frequencies, and each frequency component has a corresponding amplitude value, indicating the intensity or contribution of the frequency in the original signal; since the respiratory signal has a relatively stable periodicity, it will appear as a significant peak in the frequency domain, and the frequency corresponding to this peak is the main frequency component of breathing; therefore, finding the frequency corresponding to the maximum amplitude in the frequency domain is actually identifying the dominant periodic component, that is, the breathing frequency.

[0029] Preferably, the motion posture evaluation coefficient is calculated based on the joint data, including:

[0030] The elbow joint angle in the joint data is marked as θ1, the knee joint angle is marked as θ2, and the ankle joint angle is marked as θ3;

[0031] By formula The motion posture evaluation coefficient is calculated; wherein, P1 is the motion posture evaluation coefficient, A1, A2, A3 are weight coefficients, and A1+A2+A3=1, M1 is the standard elbow joint angle, M2 is the standard knee joint angle, and M3 is the standard ankle joint angle.

[0032] It should be noted that the weight coefficient is set by experts in the field based on experience according to the specific type of exercise;

[0033] The standard elbow joint angle, standard knee joint angle and standard ankle joint angle refer to the ideal angles of each joint in a specific sport. For example, in a basketball shooting action, the ideal elbow joint angle is usually 90 degrees to ensure that the wrist can release the ball smoothly.

[0034] Preferably, the breathing assessment coefficient is calculated based on the breathing data, comprising:

[0035] Extract the respiratory data and mark the respiratory frequency in the respiratory data as V;

[0036] The breathing assessment coefficient is calculated by the formula P2=α×e^[(V-ZV)^2 / ln(ZV+1)]; wherein P2 is the breathing assessment coefficient, α is the proportional coefficient, and 0<α<1, ZV is the optimal breathing frequency of the target movement, and ln(*) is a logarithmic function with the natural number e as the base.

[0037] It should be noted that the proportional coefficient α is set by technical experts in this field according to the age of the athletes. The older the age, the smaller the proportional coefficient α.

[0038] Preferably, the correction model is constructed based on the personnel data to correct the motion posture evaluation coefficient, including:

[0039] The ratio of forearm length to upper arm length is marked as L, and the ratio of thigh length to calf length is marked as T;

[0040] Taking L and T as independent variables and the corrected target motion posture evaluation coefficient as the dependent variable; constructing the correction model by polynomial fitting;

[0041] The correction model is specifically: ZP=P1×[1+B1×e^[L / (L+1)]+B2×e^[T / (T+1)]]; wherein ZP is the target motion posture evaluation coefficient, B1 and B2 are proportional coefficients, and if L>1, then -1<B1<0, if L<1, then 0<B1<1; if T>1, then -1<B2<0, if T<1, then 0<B2<1.

[0042] It should be noted that the proportionality coefficient is set by experts in the field based on experience according to the specific type of movement;

[0043] The influence of a person's physical data (such as forearm length, upper arm length, thigh length and calf length) on the motion posture evaluation coefficient is mainly reflected in the fact that these proportions will change the deviation between the actual joint angle and the standard angle; specifically, different body proportions will lead to different actual joint angles, thus affecting the evaluation results; for example, individuals with long forearms and short upper arms need a larger elbow flexion angle to achieve the same shooting point when performing a basketball shot, resulting in the actual elbow joint angle being larger than the standard angle, making the motion posture evaluation coefficient larger; similarly, individuals with long thighs and short calves need a larger knee flexion angle when performing a squat to achieve the same effect of the thigh being parallel to the ground, resulting in the actual knee joint angle being larger than the standard angle, making the motion posture evaluation coefficient larger.

[0044] Preferably, judging whether the motion posture of the target motion is qualified based on the target motion posture evaluation coefficient includes:

[0045] Determine whether the target motion posture evaluation coefficient is greater than the preset motion posture evaluation coefficient threshold; if yes, generate a warning message and send it to the client; if no, the motion posture of the target motion is qualified.

[0046] It should be noted that the preset motion posture evaluation coefficient threshold is set by experts in this field based on experience.

[0047] Preferably, judging whether the breathing rate of the athlete is qualified based on the breathing assessment coefficient includes:

[0048] Determine whether the breathing assessment coefficient is greater than the preset breathing assessment coefficient threshold; if yes, generate a warning message and send it to the client; if no, the breathing rate of the athlete is qualified.

[0049] It should be noted that the preset breathing assessment coefficient threshold is set by experts in this field based on experience.

[0050] A second aspect of the present invention provides a method for monitoring sports postures in a smart playground based on visual recognition, comprising:

[0051] Step 1: Obtain the video data and personnel data of the target movement;

[0052] Step 2: Analyze the video data to obtain joint data and breathing data; calculate the motion posture evaluation coefficient based on the joint data; and calculate the breathing evaluation coefficient based on the breathing data;

[0053] Step 3: Construct a correction model based on personnel data to correct the motion posture evaluation coefficient and obtain the target motion posture evaluation coefficient;

[0054] Step 4: Determine whether the target movement posture is qualified based on the target movement posture evaluation coefficient; determine whether the exerciser's breathing rate is qualified based on the breathing evaluation coefficient.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] 1. The prior art ignores individual differences. Everyone's arm length and leg length are different. For example, for athletes with different arm lengths, in order to achieve the same action effect when performing certain actions, the angles at which the elbow joints need to be bent will be different; if the individual arm length differences are ignored and a unified standard angle is used for evaluation, the deviation between the actual elbow joint angle and the standard angle will be misjudged as an erroneous action, thereby giving inaccurate feedback. The present invention obtains video data and personnel data of athletes in real time through a data acquisition module, analyzes the video data through a data analysis module, extracts human body information in each frame of the image, constructs a human skeleton model, and obtains joint data and respiratory data; calculates based on the joint data and respiratory data The motion posture evaluation coefficient and the breathing evaluation coefficient are obtained; the correction module constructs a correction model according to the personnel data, adjusts the motion posture evaluation coefficient to take into account the individual body proportion differences, and uses the forearm to upper arm length ratio and the thigh to calf length ratio as independent variables through polynomial fitting, and the corrected target motion posture evaluation coefficient as the dependent variable, so as to more accurately reflect the deviation between the actual joint angle and the standard angle; the monitoring module determines whether the motion posture is qualified based on the corrected motion posture evaluation coefficient, and determines whether the breathing frequency is qualified based on the breathing evaluation coefficient, which solves the technical problem that the prior art ignores individual differences, resulting in the deviation between the actual elbow joint angle and the standard angle being misjudged as an erroneous action, thereby giving inaccurate feedback.

[0057] 2. The present invention uses a stereo matching algorithm to analyze each frame of image in the video data to generate three-dimensional point cloud data of the athlete, ensuring that the images captured from multiple perspectives can accurately reconstruct the three-dimensional model of the athlete; through a Poisson surface reconstruction algorithm or an active three-dimensional reconstruction algorithm, a high-precision three-dimensional model is constructed based on the three-dimensional point cloud data, providing a solid foundation for subsequent volume change analysis; through feature point tracking, optical flow tracking or deep learning-driven tracking algorithms, the system can track the volume changes of the chest area in real time and generate detailed volume change curves, thereby accurately reflecting the dynamic changes during the breathing process; the volume change curve is converted from the time domain to the frequency domain using a fast Fourier transform, the positive frequency part is extracted and the frequency corresponding to the maximum amplitude is found as the breathing frequency. This method can not only effectively identify the main frequency components of breathing, but also avoid noise interference and improve the accuracy of frequency analysis. Based on the extracted breathing rate data, the breathing assessment coefficient is calculated through a formula; by judging whether the breathing assessment coefficient is greater than the preset threshold, it is determined whether the breathing rate is qualified, and if necessary, a warning message is generated and sent to the client. This not only improves the accuracy and reliability of breathing rate assessment, but also provides personalized feedback and timely warnings to help athletes optimize breathing patterns, improve training effects, and reduce the risk of sports injuries. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 A schematic diagram of a system module according to an embodiment of the present invention;

[0060] Figure 2 A schematic diagram of a specific process of an embodiment of the present invention;

[0061] Figure 3 Schematic diagram of the method steps of an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] See also Figure 1-Figure 2, the first aspect of the present invention provides a monitoring system for sports postures in a smart playground based on visual recognition, including: a data acquisition module, a data analysis module, a correction module and a monitoring module;

[0064] Data acquisition module: used to obtain the video data and personnel data of the target movement in real time;

[0065] Data analysis module: by analyzing the video data, obtain joint data and breathing data; calculate the motion posture evaluation coefficient based on the joint data; and,

[0066] A breathing assessment coefficient is calculated based on the breathing data;

[0067] Correction module: constructs a correction model based on personnel data to correct the motion posture evaluation coefficient and obtain the target motion posture evaluation coefficient;

[0068] Monitoring module: judge whether the target movement posture is qualified based on the target movement posture evaluation coefficient; judge whether the breathing rate of the exerciser is qualified based on the breathing evaluation coefficient.

[0069] Obtain target motion video data and personnel data, including:

[0070] The data acquisition equipment is used to obtain in real time the video data of athletes performing target sports on the smart playground and the personnel data of the athletes; the target sports refers to the type of sports performed by the athletes; the personnel data include: forearm length, upper arm length, thigh length and calf length.

[0071] By analyzing the video data, including:

[0072] Extract the image of each frame in the video data;

[0073] Extract human body information from each frame of the image and build a human skeleton model based on the human body information; the human body information includes: shoulders, elbows, wrists, hips, knees, ankles and toes;

[0074] A spatial rectangular coordinate system is constructed by taking any position in the human skeleton model as the origin;

[0075] Obtain the corresponding coordinates of the human body information in each frame of the image based on a spatial rectangular coordinate system;

[0076] Based on the corresponding coordinates of the human body information, the joint data corresponding to the human body information in each frame of the image is calculated by the cosine theorem; wherein the joint data includes: elbow joint angle, knee joint angle and ankle joint angle;

[0077] The breathing data is obtained by analyzing the image of each frame in the video data.

[0078] By analyzing the image of each frame in the video data, including:

[0079] The image of each frame in the video data is analyzed by a stereo matching algorithm to obtain three-dimensional point cloud data of the moving person; wherein the stereo matching algorithm includes: local stereo matching, global stereo matching or deep learning stereo matching algorithm;

[0080] Based on the three-dimensional point cloud data, a three-dimensional model of the moving person is constructed by a three-dimensional reconstruction algorithm; wherein the three-dimensional reconstruction algorithm includes: a Poisson surface reconstruction algorithm or an active three-dimensional reconstruction algorithm;

[0081] The volume change of the target area in the three-dimensional model corresponding to each frame of the image is obtained by a three-dimensional tracking algorithm, and a volume change curve of the target area corresponding to the video data is obtained; wherein the three-dimensional tracking algorithm includes: tracking of feature points, optical flow tracking or tracking algorithm driven by deep learning; the target area refers to the chest of the athlete;

[0082] The volume change curve is converted from the time domain to the frequency domain by fast Fourier transform to obtain the corresponding frequency axis;

[0083] Extract the positive frequency part of the rate axis, and extract the maximum frequency point in the positive frequency part;

[0084] The frequency corresponding to the maximum frequency point is converted into the respiratory frequency and marked as respiratory data.

[0085] The motion posture evaluation coefficients are calculated based on the joint data, including:

[0086] The elbow joint angle in the joint data is marked as θ1, the knee joint angle is marked as θ2, and the ankle joint angle is marked as θ3;

[0087] By formula The motion posture evaluation coefficient is calculated; wherein, P1 is the motion posture evaluation coefficient, A1, A2, A3 are weight coefficients, and A1+A2+A3=1, M1 is the standard elbow joint angle, M2 is the standard knee joint angle, and M3 is the standard ankle joint angle.

[0088] The breathing assessment coefficients are calculated based on the breathing data, including:

[0089] Extract the respiratory data and mark the respiratory frequency in the respiratory data as V;

[0090] The breathing assessment coefficient is calculated by the formula P2=α×e^[(V-ZV)^2 / ln(ZV+1)]; wherein P2 is the breathing assessment coefficient, α is the proportional coefficient, and 0<α<1, ZV is the optimal breathing frequency of the target movement, and ln(*) is a logarithmic function with the natural number e as the base.

[0091] A correction model is constructed based on personnel data to correct the motion posture evaluation coefficient, including:

[0092] The ratio of forearm length to upper arm length is marked as L, and the ratio of thigh length to calf length is marked as T;

[0093] Taking L and T as independent variables and the corrected target motion posture evaluation coefficient as the dependent variable; constructing the correction model by polynomial fitting;

[0094] The correction model is specifically: ZP=P1×[1+B1×e^[L / (L+1)]+B2×e^[T / (T+1)]]; wherein ZP is the target motion posture evaluation coefficient, B1 and B2 are proportional coefficients, and if L>1, then -1<B1<0, if L<1, then 0<B1<1; if T>1, then -1<B2<0, if T<1, then 0<B2<1.

[0095] Whether the target motion posture is qualified is judged based on the target motion posture evaluation coefficient, including:

[0096] Determine whether the target motion posture evaluation coefficient is greater than the preset motion posture evaluation coefficient threshold; if yes, generate a warning message and send it to the client; if no, the motion posture of the target motion is qualified.

[0097] Determine whether the athlete's breathing rate is qualified based on the breathing assessment coefficient, including:

[0098] Determine whether the breathing assessment coefficient is greater than the preset breathing assessment coefficient threshold; if yes, generate a warning message and send it to the client; if no, the breathing rate of the athlete is qualified.

[0099] For example, if a basketball player is doing shooting training, the smart playground sports posture monitoring system of the present invention will be used to evaluate his sports posture and breathing rate. The details are as follows:

[0100] 1. Data acquisition module;

[0101] Target sports: basketball shooting;

[0102] Personnel data:

[0103] Forearm length: 30cm;

[0104] Upper arm length: 25cm;

[0105] Thigh length: 45cm;

[0106] Calf length: 40cm;

[0107] The video data of the athlete's shooting action on the smart playground is obtained in real time through the data acquisition equipment.

[0108] 2. Data analysis module;

[0109] Extract human body information from each frame of image and construct a human skeleton model;

[0110] Extract each frame of the video data and build a human skeleton model based on the human body information (shoulders, elbows, wrists, hips, knees, ankles and toes) in the image. Construct a spatial rectangular coordinate system with any position in the human skeleton model as the origin, and obtain the corresponding coordinates of the human body information in each frame of the image based on the coordinate system.

[0111] Calculate joint data;

[0112] The joint data corresponding to the human body information in each frame of the image is calculated by the cosine theorem:

[0113] Elbow joint angle θ1: the actual measured value is 95 degrees;

[0114] Knee joint angle θ2: The actual measured value is 170 degrees;

[0115] Ankle joint angle θ3: the actual measured value is 85 degrees;

[0116] Analyze respiratory data;

[0117] Generate 3D point cloud data through stereo matching algorithm, build a 3D model using Poisson surface reconstruction algorithm, and obtain the volume change curve of the chest area through feature point tracking algorithm. Use fast Fourier transform to convert the volume change curve from time domain to frequency domain, and find the frequency corresponding to the maximum amplitude as the respiratory frequency V;

[0118] Respiratory rate V: The actual measured value is 16 times / minute;

[0119] 3. Calculate the motion posture evaluation coefficient;

[0120] By formula Calculate the motion posture evaluation coefficient;

[0121] Assume the standard angle is:

[0122] Standard elbow joint angle M1 = 90 degrees;

[0123] Standard knee joint angle M2 = 180 degrees;

[0124] Standard ankle joint angle M3 = 90 degrees;

[0125] Assume weight coefficients: A1 = 0.6, A2 = 0.2, A3 = 0.2;

[0126] Substituting the data, we get P1 = 1.00148;

[0127] 4. Modify the model;

[0128] According to the ratio of forearm length to upper arm length, L, and the ratio of thigh length to calf length, T:

[0129] L = 30 / 25 = 1.2;

[0130] T = 45 / 40 = 1.125;

[0131] Corrected model formula: ZP = P1 × [1 + B1 × e^[L / (L+1)] + B2 × e^[T / (T+1)]];

[0132] Assume that the proportionality coefficients B1 = -0.5 and B2 = -0.4 (because L>1, T>1);

[0133] Substituting the data, we get ZP = 0.4593;

[0134] 5. Calculate the respiratory assessment coefficient;

[0135] The breathing assessment coefficient is calculated by the formula P2 = α × e^[(V-ZV)^2 / ln(ZV+1)];

[0136] Assuming the optimal breathing rate ZV = 15 times / minute, the proportionality coefficient α = 0.8 (assuming the athlete is older):

[0137] Substituting the data, we get P2 = 1.147;

[0138] 6. Determine whether it is qualified;

[0139] Assume that the preset motion posture evaluation coefficient threshold is 0.5 and the preset breathing evaluation coefficient threshold is 1.2:

[0140] Movement posture evaluation coefficient ZP: 0.4593<0.5, the movement posture is qualified.

[0141] Respiration assessment coefficient P2: 1.147<1.2, the breathing rate is qualified.

[0142] Through the above examples, we can see that the smart playground sports posture monitoring system of the present invention can accurately evaluate the sports posture and breathing rate of athletes, and make corrections based on individual body proportion differences to provide personalized feedback. In this example, the athlete's sports posture and breathing rate both meet the standards, and the system does not generate warning information. This process demonstrates the efficiency and accuracy of the system, ensuring that each athlete can exercise in the best condition, improving the training effect and reducing the risk of sports injuries.

[0143] See also Figure 3 The second aspect of the present invention provides a method for monitoring sports postures in a smart playground based on visual recognition, comprising:

[0144] Step 1: Obtain the video data and personnel data of the target movement;

[0145] Step 2: Analyze the video data to obtain joint data and breathing data; calculate the motion posture evaluation coefficient based on the joint data; and calculate the breathing evaluation coefficient based on the breathing data;

[0146] Step 3: Construct a correction model based on personnel data to correct the motion posture evaluation coefficient and obtain the target motion posture evaluation coefficient;

[0147] Step 4: Determine whether the target movement posture is qualified based on the target movement posture evaluation coefficient; determine whether the exerciser's breathing rate is qualified based on the breathing evaluation coefficient.

[0148] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0149] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A monitoring system for sports postures in a smart playground based on visual recognition, characterized in that: include: Data acquisition module, data analysis module, correction module and monitoring module; Data acquisition module: used to obtain the video data and personnel data of the target movement in real time; Data analysis module: obtain joint data and respiratory data by analyzing video data; The motion posture evaluation coefficient is calculated based on the joint data; as well as, A breathing assessment coefficient is calculated based on the breathing data; Correction module: constructs a correction model based on personnel data to correct the motion posture evaluation coefficient and obtain the target motion posture evaluation coefficient; Monitoring module: judge whether the target movement posture is qualified based on the target movement posture evaluation coefficient; judge whether the breathing rate of the exerciser is qualified based on the breathing evaluation coefficient.

2. The monitoring system for sports posture in a smart playground based on visual recognition according to claim 1 is characterized in that: The step of acquiring the video data and personnel data of the target movement includes: The data acquisition equipment is used to obtain in real time the video data of athletes performing target sports on the smart playground and the personnel data of the athletes; the target sports refers to the type of sports performed by the athletes; the personnel data include: forearm length, upper arm length, thigh length and calf length.

3. The monitoring system of sports posture in smart playground based on visual recognition according to claim 1 is characterized in that: The analysis of the video data includes: Extract the image of each frame in the video data; Extract human body information from each frame of the image and build a human skeleton model based on the human body information; the human body information includes: shoulders, elbows, wrists, hips, knees, ankles and toes; A spatial rectangular coordinate system is constructed by taking any position in the human skeleton model as the origin; Obtain the corresponding coordinates of the human body information in each frame of the image based on a spatial rectangular coordinate system; Based on the corresponding coordinates of the human body information, the joint data corresponding to the human body information in each frame of the image is calculated by the cosine theorem; wherein the joint data includes: elbow joint angle, knee joint angle and ankle joint angle; The breathing data is obtained by analyzing the image of each frame in the video data.

4. The monitoring system for sports posture in a smart playground based on visual recognition according to claim 3 is characterized in that: The analysis of each frame of the video data includes: The image of each frame in the video data is analyzed by a stereo matching algorithm to obtain three-dimensional point cloud data of the moving person; wherein the stereo matching algorithm includes: local stereo matching, global stereo matching or deep learning stereo matching algorithm; Based on the three-dimensional point cloud data, a three-dimensional model of the moving person is constructed by a three-dimensional reconstruction algorithm; wherein the three-dimensional reconstruction algorithm includes: a Poisson surface reconstruction algorithm or an active three-dimensional reconstruction algorithm; The volume change of the target area in the three-dimensional model corresponding to each frame of the image is obtained by a three-dimensional tracking algorithm, and a volume change curve of the target area corresponding to the video data is obtained; wherein the three-dimensional tracking algorithm includes: tracking of feature points, optical flow tracking or tracking algorithm driven by deep learning; the target area refers to the chest of the athlete; The volume change curve is converted from the time domain to the frequency domain by fast Fourier transform to obtain the corresponding frequency axis; Extract the positive frequency part of the rate axis, and extract the maximum frequency point in the positive frequency part; The frequency corresponding to the maximum frequency point is converted into the respiratory frequency and marked as respiratory data.

5. The monitoring system of sports posture in smart playground based on visual recognition according to claim 1 is characterized in that: The motion posture evaluation coefficient is calculated based on the joint data, including: The elbow joint angle in the joint data is marked as θ1, the knee joint angle is marked as θ2, and the ankle joint angle is marked as θ3; By formula The motion posture evaluation coefficient is calculated; wherein, P1 is the motion posture evaluation coefficient, A1, A2, A3 are weight coefficients, and A1+A2+A3=1, M1 is the standard elbow joint angle, M2 is the standard knee joint angle, and M3 is the standard ankle joint angle.

6. The monitoring system of sports posture in smart playground based on visual recognition according to claim 1 is characterized in that: The step of calculating the breathing assessment coefficient based on the breathing data includes: Extract the respiratory data and mark the respiratory frequency in the respiratory data as V; The breathing assessment coefficient is calculated by the formula P2=α×e^[(V-ZV)^2 / ln(ZV+1)]; wherein P2 is the breathing assessment coefficient, α is the proportional coefficient, and 0<α<1, ZV is the optimal breathing frequency of the target movement, and ln(*) is a logarithmic function with the natural number e as the base.

7. The monitoring system of sports posture in smart playground based on visual recognition according to claim 1 is characterized in that: The method of constructing a correction model based on personnel data to correct the motion posture evaluation coefficient includes: The ratio of forearm length to upper arm length is marked as L, and the ratio of thigh length to calf length is marked as T; Taking L and T as independent variables and the corrected target motion posture evaluation coefficient as the dependent variable; constructing the correction model by polynomial fitting; The correction model is specifically: ZP=P1×[1+B1×e^[L / (L+1)]+B2×e^[T / (T+1)]]; wherein ZP is the target motion posture evaluation coefficient, B1 and B2 are proportional coefficients, and if L>1, then -1<B1<0, if L<1, then 0<B1<1; if T>1, then -1<B2<0, if T<1, then 0<B2<1.

8. The monitoring system of sports posture in smart playground based on visual recognition according to claim 1 is characterized in that: The step of judging whether the target motion posture is qualified based on the target motion posture evaluation coefficient includes: Determine whether the target motion posture evaluation coefficient is greater than the preset motion posture evaluation coefficient threshold; if yes, generate a warning message and send it to the client; if no, the motion posture of the target motion is qualified.

9. The monitoring system of sports posture in smart playground based on visual recognition according to claim 1 is characterized in that: The method of judging whether the breathing rate of the athlete is qualified based on the breathing assessment coefficient includes: Determine whether the breathing assessment coefficient is greater than the preset breathing assessment coefficient threshold; if yes, generate a warning message and send it to the client; if no, the breathing rate of the athlete is qualified.

10. A method for monitoring sports postures in a smart playground based on visual recognition, applied to a monitoring system for sports postures in a smart playground based on visual recognition as claimed in any one of claims 1 to 9, characterized in that: include: Step 1: Obtain the video data and personnel data of the target movement; Step 2: Analyze the video data to obtain joint data and breathing data; The motion posture evaluation coefficient is calculated based on the joint data; and, calculating a breathing assessment coefficient based on the breathing data; Step 3: Construct a correction model based on personnel data to correct the motion posture evaluation coefficient and obtain the target motion posture evaluation coefficient; Step 4: Determine whether the target movement posture is qualified based on the target movement posture evaluation coefficient; determine whether the exerciser's breathing rate is qualified based on the breathing evaluation coefficient.

Images