Multi-dimensional analysis dance training intelligent evaluation system

Abstract

The invention discloses a multi-dimensional analysis dance training intelligent evaluation system, and relates to the field of data analysis, and the main scheme is that a dance training score X is calculated according to an action efficiency index XA, a space and action cooperation index XB, a comprehensive expressive force index XC, an emotional rhythm resonance index XD and a style fusion degree index XE. And subjective factors such as emotion expression and the cooperative influence among the factors are ignored, so that the accuracy of the evaluation result is low. The dance training grade of the user is judged according to the grade threshold value set and the dance training score X, so that the user can know the training condition of the user and make a personalized training plan.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a multi-dimensional analysis and intelligent evaluation system for dance training. Background Art

[0002] In the field of dance training, the dance training intelligent evaluation system can show users the system's professional evaluation results, allowing users to more intuitively understand their own shortcomings and enhance their motivation and confidence in learning.

[0003] The working principle of the traditional dance training intelligent evaluation system is: receive the user's dance video data and collect the user's body posture data in the video, extract various movement features from the user's posture data and analyze them, and evaluate the user's dance training only based on the accuracy of each single movement.

[0004] The existing technology still has the following shortcomings: when conducting dance training evaluation, it is limited to objective factors such as the accuracy of movements, ignoring subjective factors such as emotional expression and the synergistic influence of various factors, resulting in low accuracy of evaluation results. Summary of the invention

[0005] 1. Technical issues to be solved

[0006] In view of the shortcomings of the prior art, the present invention provides a multi-dimensional analysis intelligent dance training evaluation system, which calculates the dance training score X according to the action efficiency index XA, the space and action coordination index XB, the comprehensive expressiveness index XC, the emotional rhythm resonance index XD and the style fusion index XE, thereby solving the problem of ignoring subjective factors such as emotional expression and the synergistic influence of various factors when evaluating dance training, thereby reducing the accuracy of the evaluation results.

[0007] (II) Technical solution

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: a multi-dimensional analysis dance training intelligent evaluation system, comprising:

[0009] A data collection module is used to collect dance training movement position data, movement integrity data, user body condition data, movement time data, coordination data, emotion data and dance style data;

[0010] An index calculation module is used to calculate a posture conformity score A according to the action position data, calculate a movement integrity score B according to the action integrity data, and further calculate a movement accuracy score C; calculate a physical exertion degree N according to the user's physical state data and the action time data, and calculate a movement efficiency index XA according to the movement accuracy score C and the physical exertion degree N;

[0011] Calculate the level change frequency index E according to the action position data, and calculate the action change frequency index F according to the action position data and the action time data; calculate the space and action coordination index XB according to the level change frequency index E, the action change frequency index F and the coordination data;

[0012] The movement amplitude score H and the movement intensity score I are calculated based on the coordination data, and the comprehensive expression index XC is further calculated based on the emotion data;

[0013] The beat synchronization score M is calculated based on the action time data; the emotional rhythm resonance index XD is calculated based on the beat synchronization score M, the rhythm intensity change score MC, and the emotional data;

[0014] Calculate the style fusion index XE based on the dance style data and the total number of movements d2;

[0015] The level judgment module is used to calculate the dance training score X according to the action efficiency index XA, the space and action coordination index XB, the comprehensive expression index XC, the emotional rhythm resonance index XD and the style fusion index XE; preset the level threshold set; and judge the user's dance training level according to the level threshold set and the dance training score X.

[0016] In the preferred embodiment of the multi-dimensional analysis dance training intelligent evaluation system, the method for calculating the movement accuracy score C is:

[0017] The motion position data includes the user's dance motion coordinates (A a , B a , C a ), standard dance movement coordinates (AA a ,BB a , C.C. a ) and the maximum distance deviation value D a ;

[0018] The posture conformity score A is calculated based on the action position data, based on the formula:

[0019]

[0020] Among them, (A a , B a , C a ) is the coordinate of the a-th dance action, a is the serial number corresponding to different dance actions, and its value is [1, b]; b is the total number of different dance actions, and its value is a positive integer; (AA a ,BB a , C.C. a ) is the coordinate of the ath standard dance action; D a is the maximum distance deviation allowed for the a-th dance action;

[0021] The action integrity data includes the number of errors and omissions d1 and the total number of actions d2;

[0022] The action integrity score B is calculated based on the action integrity data according to the formula:

[0023]

[0024] The action accuracy score C is calculated based on the posture compliance score A and the action integrity score B, according to the following formula:

[0025] C=α1×A+α2×B

[0026] Among them, α1 is the weight coefficient of the posture compliance score A, and its value is 0.3~0.7; α2 is the weight coefficient of the action integrity score B, and its value is 0.3~0.7; and α1+α2=1.

[0027] In the preferred embodiment of the multi-dimensional analysis dance training intelligent evaluation system, the method for calculating the action efficiency index XA is:

[0028] User physical status data including average heart rate during training Average breathing rate during training Heart rate basic value N4 and respiratory rate basic value N5;

[0029] The action time data also includes the total dance time N1;

[0030] The physical exertion degree N is calculated based on the user's physical condition data and the total dance time N1, and the formula is as follows:

[0031]

[0032] The action efficiency index XA is calculated based on the action accuracy score C and the physical exertion level N, based on the following formula:

[0033]

[0034] In the preferred embodiment of the multi-dimensional analysis dance training intelligent evaluation system, the method for calculating the level change frequency index E and the action change frequency index F is:

[0035] The action position data also includes the height coordinates E of the body parts (e，f) ;

[0036] According to the height coordinate E of the body part (e，f) The formula for calculating the level change frequency index E is as follows:

[0037]

[0038] Among them, E (e，f) is the height coordinate of the e-th body part at the f-th time point; e is the serial number corresponding to different body parts, and its value is [1, g]; g is the total number of body parts; f is the serial number corresponding to different time points, and its value is [1, h]; h is the total number of time points; E (e，f+1) is the height coordinate of the e-th body part at the f+1-th time point;

[0039] The action time data also includes the action time point M a ;

[0040] According to the user's dance movement coordinates (A a , B a , C a ) and action time point M a The action change frequency index F is calculated based on the formula:

[0041]

[0042] Among them, (A a+1 , B a+1 , C a+1 ) is the coordinate of the a+1th dance action; M a is the time point of the a-th dance action; M a+1 is the time point of the a+1th dance action.

[0043] In the preferred embodiment of the multi-dimensional analysis dance training intelligent evaluation system, the method for calculating the space and action coordination index XB is:

[0044] Coordination data include joint coordination score G1 and body part coordination score G2;

[0045] The spatial and motion coordination index XB is calculated based on the level change frequency index E, the action change frequency index F, the joint coordination score G1 and the body part coordination score G2, and the formula is:

[0046] XB=E×F×(G1+G2).

[0047] In the preferred embodiment of the multi-dimensional analysis dance training intelligent evaluation system, the method for calculating the comprehensive expression index XC is:

[0048] Coordination data also includes limb amplitude H a , target limb amplitude HA a 、Movement Strength I a , Target action intensity IA a and standard deviation of action strength, β;

[0049] According to the limb amplitude Ha and target limb amplitude HA a The formula for calculating the range of motion score H is:

[0050]

[0051] According to the action intensity I a , Target action intensity IA a The action intensity score I is calculated based on the formula:

[0052]

[0053] The emotional data include emotional expressiveness score J1;

[0054] The comprehensive expression index XC is calculated based on the movement amplitude score H, the movement intensity score I and the emotional expression score J1, and the formula is:

[0055] XC=H×I+J1.

[0056] In the preferred embodiment of the multi-dimensional analysis dance training intelligent evaluation system, the method for calculating the emotional rhythm resonance index XD is:

[0057] The action time data also includes the music beat time point MA a and rhythm intensity change score MC;

[0058] According to the action time point M a and music beat time point MA a The beat synchronization score M is calculated based on the formula:

[0059]

[0060] The emotional data also includes the emotional change intensity score J2;

[0061] The emotional rhythm resonance index XD is calculated based on the beat synchronization score M, the rhythm intensity change score MC, the emotional expression score J1 and the emotional change intensity score J2. The formula is:

[0062]

[0063] Among them, γ is the anti-zero correction value.

[0064] In the preferred embodiment of the multi-dimensional analysis dance training intelligent evaluation system, the method for calculating the style fusion index XE is:

[0065] The dance style data includes the number of dance style movements P v and style feature score Q v ;

[0066] The style fusion index XE is calculated based on the dance style data and the total number of movements d2, and the formula is:

[0067]

[0068] Among them, P v is the number of movements of the vth dance style in dance training, v is the serial number corresponding to different dance styles, and its value is [1, u]; u is the total number of dance styles, and its value is a positive integer; Q v Score the style features of the vth dance style in dance training.

[0069] In the preferred embodiment of the multi-dimensional analysis dance training intelligent evaluation system, the method for calculating the dance training score X is:

[0070] The dance training score X is calculated based on the action efficiency index XA, the space and action coordination index XB, the comprehensive expression index XC, the emotional rhythm resonance index XD and the style integration index XE. The formula is:

[0071] X=XA+XB+XC+XD+XE.

[0072] In the preferred embodiment of the multi-dimensional analysis dance training intelligent evaluation system, the method for judging the user's dance training level is:

[0073] The grade threshold set includes an excellent threshold XY and a qualified threshold XH;

[0074] The user's dance training level is determined based on the level threshold set and the dance training score X, and the criteria are as follows:

[0075]

[0076] (III) Beneficial effects

[0077] The present invention provides a multi-dimensional analysis dance training intelligent evaluation system, which has the following beneficial effects:

[0078] (1) By collecting dance training movement position data, movement integrity data, user body condition data, movement time data, coordination data, emotion data and dance style data, it can provide effective data for subsequent calculations and increase the dimension of evaluating the dance training level.

[0079] (2) Calculating the movement efficiency index XA based on the movement accuracy score C and the degree of physical exertion N helps to understand the balance between the user's movement quality and physical exertion during dance training. Calculating the space and movement coordination index XB based on the level change frequency index E, the movement change frequency index F and the coordination data can more comprehensively evaluate the user's ability to utilize space during dance training. Calculating the movement amplitude score H and the movement intensity score I based on the coordination data, and further calculating the comprehensive expression index XC based on the emotional expression score J1 can evaluate the user's coordination between dance movements and expression. Calculating the emotional rhythm resonance index XD helps to understand the user's effectiveness in emotional expression and rhythm coordination. Calculating the style fusion index XE based on the dance style data and the total number of movements d2 helps to evaluate the effectiveness of dance works in integrating different styles.

[0080] (3) The dance training score X is calculated based on the action efficiency index XA, the space and action coordination index XB, the comprehensive expression index XC, the emotional rhythm resonance index XD and the style fusion index XE, which solves the problem of ignoring subjective factors such as emotional expression and the synergistic influence between various factors when evaluating dance training, thereby reducing the accuracy of the evaluation results. Judging the user's dance training level based on the level threshold set and the dance training score X helps users understand their own training situation and develop personalized training plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 A schematic diagram of the system composition of a multi-dimensional analysis dance training intelligent evaluation system of the present invention; DETAILED DESCRIPTION

[0082] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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.

[0083] See also Figure 1 The present invention provides a multi-dimensional analysis dance training intelligent evaluation system, comprising:

[0084] The data collection module is used to collect dance training movement position data, movement integrity data, user body condition data, movement time data, coordination data, emotion data and dance style data.

[0085] In the above scheme, by collecting dance training movement position data, movement integrity data, user body condition data, movement time data, coordination data, emotion data and dance style data, it can provide effective data for subsequent calculations and increase the dimension of evaluating the dance training level.

[0086] The indicator calculation module is used to calculate the posture conformity score A based on the action position data, calculate the action integrity score B based on the action integrity data, and further calculate the action accuracy score C; calculate the physical exertion degree N based on the user's physical condition data and action time data, and calculate the action efficiency index XA based on the action accuracy score C and the physical exertion degree N.

[0087] Calculate the level change frequency index E according to the action position data, and calculate the action change frequency index F according to the action position data and the action time data; calculate the space and action coordination index XB according to the level change frequency index E, the action change frequency index F and the coordination data;

[0088] The movement amplitude score H and the movement intensity score I are calculated based on the coordination data, and the comprehensive expression index XC is further calculated based on the emotion data;

[0089] The beat synchronization score M is calculated based on the action time data; the emotional rhythm resonance index XD is calculated based on the beat synchronization score M, the rhythm intensity change score MC, and the emotional data;

[0090] The style fusion index XE is calculated based on the dance style data and the total number of movements d2.

[0091] Specifically, the method for calculating the action accuracy score C is:

[0092] The motion position data includes the user's dance motion coordinates (A a , B a , C a ), standard dance movement coordinates (AA a ,BB a , C.C. a ) and the maximum distance deviation value D a .

[0093] It should be noted that the user's dance motion coordinates (A a , B a , C a ) refers to the coordinates of the user's body center of gravity during dance training. Standard dance movement coordinates (AA a ,BB a , C.C. a) refers to the coordinates of the center of gravity of the body of a professional dancer when performing dance movements. The method for obtaining the coordinates of dance movements is: using an inertial measurement unit sensor to obtain the position information of each part of the body when the user or professional dancer performs dance movements, then assigning a certain mass ratio to each part of the body according to the human body mass distribution model, and finally calculating the coordinates of the center of gravity of the body by weighted average method. Maximum distance deviation value D a It is a threshold used to measure the difference between the coordinates of the user's dance movement and the coordinates of the standard dance movement. The maximum distance deviation value D a The method for determining is: collect a large number of users' dance action data, obtain the corresponding dance action coordinates, and calculate the coordinate difference value with the standard dance action coordinates, count the smaller 50% difference value data and calculate the average value as the maximum distance deviation value D a .

[0094] The posture conformity score A is calculated based on the action position data, based on the formula:

[0095]

[0096] Among them, (A a , B a , C a ) is the coordinate of the a-th dance action, a is the serial number corresponding to different dance actions, and its value is [1, b]; b is the total number of different dance actions, and its value is a positive integer; (AA a ,BB a , C.C. a ) is the coordinate of the ath standard dance action; D a is the maximum distance deviation allowed for the a-th dance action.

[0097] It should be noted that the operating principle of this formula is: the numerator is the sum of the Euclidean distances between each dance action coordinate and the standard dance action coordinate, and the denominator is the sum of the maximum distance deviations between all dance actions and the standard dance actions. By calculating the ratio of the numerator to the denominator, and then subtracting the ratio from the fraction value 100, the posture conformity score A is obtained. The higher the posture conformity score A, the more standard the user's movements are.

[0098] The action integrity data includes the number of erroneous and missing actions d1 and the total number of actions d2.

[0099] It should be noted that the number of errors and omissions d1 refers to the total number of movements that do not meet the requirements of standard dance movements and movements that should be performed but not performed during dance training. Errors may include incorrect postures, inaccurate angles, inappropriate strength, etc., and omissions refer to movements that should be performed in a specific dance segment but are ignored by the dancer. The total number of movements d2 refers to the total number of all dance movements during the user's training. The number of errors and omissions d1 and the total number of movements d2 are obtained by: using a motion capture system, such as an optical motion capture system, an inertial motion capture device, etc., to capture the dancer's body movements in real time and obtain motion data; using standard motion videos recorded by experts or professionals, and then using tools such as TensorFlow and PyTorch to train motion recognition models. By learning from a large number of dance videos, the model can recognize specific dance movements and identify the execution of each movement. By comparing the user's motion data with the movements in the standard motion database, it can automatically detect errors and omissions, and count the number of errors and omissions d1 and the total number of movements d2.

[0100] The action integrity score B is calculated based on the action integrity data according to the formula:

[0101]

[0102] It should be noted that the formula obtains the movement integrity score B by calculating the ratio of the number of correct dance movements to the total number of movements d2.

[0103] The action accuracy score C is calculated based on the posture compliance score A and the action integrity score B, according to the following formula:

[0104] C=α1×A+α2×B

[0105] Among them, α1 is the weight coefficient of the posture compliance score A, and its value is 0.3~0.7; α2 is the weight coefficient of the action integrity score B, and its value is 0.3~0.7; and α1+α2=1.

[0106] It should be noted that this formula comprehensively considers the influence of posture conformity score A and movement integrity score B on movement accuracy score C in a weighted manner. The higher the movement accuracy score C and movement integrity score B, the higher the movement accuracy score C, and the better the user's dance training effect.

[0107] Specifically, the method for calculating the action efficiency index XA is:

[0108] User physical status data including average heart rate during training Average breathing rate during training The heart rate basic value is N4 and the respiratory rate basic value is N5.

[0109] It should be noted that the average heart rate during training Refers to the average number of heart beats per minute during dance training. The method of determination is: before dance training, wear a heart rate bracelet and continuously monitor your heart rate during training. After training, count the heart rate data and calculate its average. Average breathing rate during training It refers to the average number of times the user breathes per minute during dance training. The determination method is: before dance training, wear an existing wireless respiratory monitor and continuously monitor the respiratory rate during training. After the training, count the respiratory rate data and calculate its average value. The heart rate basal value N4 refers to the average number of times the dancer's heart beats per minute in a quiet state. The acquisition method is: use a heart rate bracelet to measure the user's heart rate data in a relaxed and calm state and calculate its average value. The respiratory rate basal value N5 refers to the number of times the dancer breathes per minute in a quiet state. The acquisition method is: use a breathing belt to measure the user's respiratory rate data in a relaxed and calm state and calculate its average value.

[0110] The action time data includes the total dance time N1.

[0111] It should be noted that the total dance time N1 refers to the total time of the user's dance training this time, which is determined by calculating the difference between the time when the user ends the dance training and the time when the user starts the dance training.

[0112] The physical exertion degree N is calculated based on the user's physical condition data and the total dance time N1, and the formula is as follows:

[0113]

[0114] It should be noted that the operating principle of this formula is: by calculating the user's average heart rate during training The difference between the heart rate base value N4 and the total dance time N1 reflects the degree of physical exertion caused by the heart rate increase during the user's training. The more the heart rate increases and the longer the duration, the greater the physical exertion caused by the heart rate factor. By calculating the average breathing rate of the user during training The difference between the heart rate and the basic breathing rate value N5 multiplied by the total dance time N1 reflects the degree of physical exertion caused by the increase in breathing rate during the user's training. The more the breathing rate increases and the longer the duration, the greater the physical exertion caused by the breathing rate factor. The physical exertion related to heart rate and the physical exertion related to breathing rate are added together to obtain the degree of physical exertion N.

[0115] The action efficiency index XA is calculated based on the action accuracy score C and the physical exertion level N, based on the following formula:

[0116]

[0117] It should be noted that the formula obtains the action efficiency index XA by calculating the ratio of the action accuracy score C and the physical exertion degree N. The higher the action accuracy score C, the lower the physical exertion degree N, the higher the action efficiency index XA, and the better the user's dance training effect.

[0118] Specifically, the method for calculating the level change frequency index E and the action change frequency index F is:

[0119] The action position data also includes the height coordinates E of the body parts (e，f) .

[0120] It should be noted that the height coordinate E of the body part (e，f) During the user's training process, the vertical distances of different body parts relative to the ground at different time points are obtained by: obtaining the three-dimensional coordinates of different body parts of the user at each time point through the optical motion capture system, and extracting the vertical coordinate values ​​as the height coordinates of the body parts. Body parts include the head, shoulders, elbows, wrists, hips, knees, ankles, etc.

[0121] According to the height coordinate E of the body part (e，f) The formula for calculating the level change frequency index E is as follows:

[0122]

[0123] Among them, E (e，f) is the height coordinate of the e-th body part at the f-th time point; e is the serial number corresponding to different body parts, and its value is [1, g]; g is the total number of body parts; f is the serial number corresponding to different time points, and its value is [1, h]; h is the total number of time points; E (e，f+1) is the height coordinate of the e-th body part at the f+1-th time point.

[0124] It should be noted that the operating principle of this formula is: the numerator reflects the total amplitude of height changes of all body parts during the entire dance process by calculating the sum of the height differences between different body parts at different time points. The denominator is the product of the total number of body parts and the total number of time points. By calculating the ratio of the numerator to the denominator, the average amplitude of height changes of body parts during the dance process is obtained, that is, the level change frequency index E. The higher the level change frequency index E, the more frequent and drastic the height changes of body parts during the dance process.

[0125] The action time data also includes the action time point Ma .

[0126] It should be noted that the action time point M a Refers to the time point when the user practices each dance move during dance training. It is determined by looking at the timestamp corresponding to each frame of the image in the optical capture system.

[0127] According to the user's dance movement coordinates (A a , B a , C a ) and action time point M a The action change frequency index F is calculated based on the formula:

[0128]

[0129] Among them, (A a+1 , B a+1 , C a+1 ) is the coordinate of the a+1th dance action; M a is the time point of the a-th dance action; M a+1 is the time point of the a+1th dance action.

[0130] It should be noted that the formula calculates the distance between the coordinates of different dance movements and divides it by the corresponding time difference to obtain the change frequency of different dance movements. The movement change frequency index F is obtained by calculating the sum of the change frequencies and dividing it by the total number of dance movements.

[0131] Specifically, the method for calculating the space and action coordination index XB is:

[0132] The coordination data includes a joint coordination score G1 and a body part coordination score G2.

[0133] It should be noted that the joint coordination score G1 is used to measure the degree of coordination between the movements of the various joints of the human body during the dance process. It reflects whether the movements of different joints of the dancer can be coordinated, smooth and natural when performing dance movements, so as to achieve the harmony and beauty of the overall movement. The specific determination method is: according to the optical motion capture system, the start time and stop time of each joint of each dance movement during the user's training are obtained, the difference between the start time and stop time of each joint of each dance movement is calculated, and the number of dance movements with a difference of less than 1 second is further calculated. The proportion of the total number of dance movements. When the proportion is greater than 90%, the joint coordination score G1 is 100. When the proportion is greater than 70% and less than 90%, the joint coordination score G1 is 70. When the proportion is greater than 50% and less than 70%, the joint coordination score G1 is 50. When the proportion is less than 50%, the joint coordination score G1 is 10.

[0134] The body part synergy score G2 mainly evaluates the degree of coordination and cooperation between different parts of the body during the dance process. It examines whether the various parts of the body, such as limbs, torso, head, etc., can work together when the dancer performs dance movements to achieve the smoothness and integrity of the overall movement. The specific determination method is: according to the optical motion capture system, the start time and stop time of each part of the body of each dance movement of the user during training are obtained, and the difference between the start time and stop time of each part of the body of the user for each dance movement is calculated, and the number of dance movements with a difference of less than 1 second is further calculated. The proportion of the total number of dance movements. When the proportion is greater than 90%, the body part synergy score G2 is 100. When the proportion is greater than 70% and less than 90%, the body part synergy score G2 is 70. When the proportion is greater than 50% and less than 70%, the body part synergy score G2 is 50. When the proportion is less than 50%, the body part synergy score G2 is 10.

[0135] The spatial and motion coordination index XB is calculated based on the level change frequency index E, the action change frequency index F, the joint coordination score G1 and the body part coordination score G2, and the formula is:

[0136] XB=E×F×(G1+G2)

[0137] It should be noted that this formula obtains the space and movement coordination index XB by considering the factors of space utilization and movement change frequency index F and the coordination degree scores between various parts and joints of the user's body during dance training.

[0138] Specifically, the method for calculating the comprehensive performance index XC is:

[0139] Coordination data also includes limb amplitude H a , target limb amplitude HA a 、Movement Strength I a , Target action intensity IA a and the standard deviation of action strength β.

[0140] It should be noted that the limb amplitude H a It refers to the amplitude of the user's limb movement during the dance. It can reflect the degree of stretch of the dancer's movements. The acquisition method is: upload the user's dance training video to the Openpose posture estimation model, identify the displacement and angle changes of the user's limbs in different directions according to the Openpose posture estimation model, input the displacement and angle changes into the limb movement model, and obtain the limb amplitude H a The limb movement model is a machine learning-based model that is trained by collecting a large amount of dancers’ limb movement data. The model calculates the user’s limb amplitude based on the learned rules and patterns.a Refers to the degree of strength displayed by the user when performing dance moves. Target Action Strength IA a It refers to the degree of strength displayed by professional dancers when performing dance moves. Movement strength is reflected in many aspects: for example, in a jumping action, the strength of the leg force when taking off, and in an arm swinging action, the strength of the arm muscle swinging. The acquisition method is: obtain the pressure value applied to the ground by the user or professional dancer when performing dance moves through a mechanical measurement sensor, obtain the resultant force on the limbs of the user or professional dancer when performing dance moves through an inertial measurement unit sensor, calculate the sum of the pressure value applied by the user and the resultant force, and use it as the movement strength I. a Calculate the pressure value applied by the professional dancer and the sum of the resultant force received as the target action intensity IA a The standard deviation of movement intensity β is a statistic used to measure the stability of the dancer's movement intensity during the dance process. It is determined by collecting the movement intensity data of each dance movement during the user's dance training, adding all the collected movement intensity data, and then dividing it by the number of data to get the average value. Calculate the sum of the squares of the difference between each movement intensity data and the average value, and then divide it by the number of data to get the variance. Calculate the square root of the variance, which is the standard deviation of movement intensity β.

[0141] According to the limb amplitude H a and target limb amplitude HA a The formula for calculating the range of motion score H is:

[0142]

[0143] It should be noted that the operating principle of this formula is: Used to measure limb amplitude H a HA with target limb amplitude a The degree of deviation between the limbs is calculated by the sum of the limb deviations of each dance movement and divided by the total number of dance movements to obtain the movement amplitude score H. The greater the degree of deviation, the more non-standard the movement amplitude and the lower the movement amplitude score H.

[0144] According to the action intensity I a , Target action intensity IA a The action intensity score I is calculated based on the formula:

[0145]

[0146] It should be noted that this formula obtains the movement intensity score I by comprehensively considering the deviation between the actual movement intensity and the target movement intensity and the fluctuation of the movement intensity. If the actual movement intensity is close to the target movement intensity and the standard deviation of the movement intensity is small, then the movement intensity score I will be relatively high, indicating that the dance training effect is good. e is the base of the natural logarithm in the exponential function.

[0147] The emotion data includes an emotion expression score J1.

[0148] It should be noted that the emotional expression score J1 is used to measure the user's ability to convey emotions through facial expressions during the dance. The determination method is: use a large-scale data set containing various facial expressions to train a facial expression recognition model, such as a convolutional neural network model and a Resnet model, input the user's dance training video into the trained expression recognition model, and use the model to identify the number of basic expressions and the number of complex emotional state combinations correctly expressed by the user, J1 = 0.3 × the number of basic expressions + 0.7 × the number of complex emotional state combinations.

[0149] The comprehensive expression index XC is calculated based on the movement amplitude score H, the movement intensity score I and the emotional expression score J1, and the formula is:

[0150] XC=H×I+J1

[0151] It should be noted that the formula obtains the comprehensive expression index XC by comprehensively considering the influence of the movement amplitude score H, the movement intensity score I and the emotional expression score J1. The comprehensive expression index XC is used to measure the comprehensive expression of the user in terms of movement and emotion during dance training. The higher the comprehensive expression index XC, the better the dance training effect.

[0152] Specifically, the method for calculating the emotional rhythm resonance index XD is:

[0153] The action time data also includes the music beat time point MA a and rhythm intensity change score MC.

[0154] It should be noted that the music beat time point MA a Refers to the specific moment when the beat that needs to change dance movements appears in the dance music. The determination method is: identify the beats in the music through audio analysis software, mark the time points corresponding to each beat, and count the time points corresponding to the beats that need to change dance movements. The rhythm intensity change score MC is an indicator used to measure the change in rhythm intensity of the dance music during playback. The determination method is: based on the Zwicker loudness model, the loudness change rate MCA and the loudness peak occurrence frequency MCB of the dance music are obtained, MC = 100×(MCA+MCB).

[0155] According to the action time point M a and music beat time point MA a The beat synchronization score M is calculated based on the formula:

[0156]

[0157] It should be noted that this formula is based on the action time point M of each dance action. a and music beat time point MA a The difference between the two values ​​is divided by the total number of dance moves to obtain the beat synchronization score M. The smaller the difference value, the higher the beat synchronization score M, and the better the dance training effect.

[0158] The emotion data also includes an emotion change intensity score J2.

[0159] It should be noted that the emotional change intensity score J2 is used to measure the degree of emotional fluctuation conveyed by the user during dance training. The determination method is: divide the user's dance training video into five segments, input the five segments into the trained expression recognition model, and use the model to identify the number of basic expressions and the number of complex emotional state combinations correctly expressed by the user in each segment, calculate the sum of the two numbers, and obtain the number of expressions correctly expressed by the user JAA, JAB, JAC, JAD, and JAE, where JAA is the number of expressions correctly expressed by the user in the first segment, JAB is the number of expressions correctly expressed by the user in the second segment, JAC is the number of expressions correctly expressed by the user in the third segment, JAD is the number of expressions correctly expressed by the user in the fourth segment, and JAE is the number of expressions correctly expressed by the user in the fifth segment.

[0160] The emotional rhythm resonance index XD is calculated based on the beat synchronization score M, the rhythm intensity change score MC, the emotional expression score J1 and the emotional change intensity score J2. The formula is:

[0161]

[0162] It should be noted that the formula obtains the emotional rhythm resonance index XD by comprehensively considering the comprehensive performance between rhythm and emotion. The higher the emotional rhythm resonance index XD, the better the user's balance between emotion and rhythm, and the better the dance training effect.

[0163] Among them, γ is the anti-zero correction value, which is a very small positive number. The anti-zero correction value is set to avoid the situation where the denominator is zero.

[0164] Specifically, the method for calculating the style fusion index XE is:

[0165] The dance style data includes the number of dance style movements P v and style feature score Q v .

[0166] It should be noted that the number of dance style movements P v It refers to the number of signature moves that a user can perform in a specific dance style during dance training. v It is a quantitative assessment of the degree to which a dance reflects a specific style feature as a whole. The acquisition method is: collect a large number of high-quality videos of different dance styles. For each video, it is necessary to manually annotate the bounding box and style feature category corresponding to each dance move. The YOLO model is trained through these videos. The user's dance training video is input into the trained YOLO model. The model will detect the dance style movements frame by frame and generate a series of bounding boxes. Each bounding box corresponds to a possible dance style and gives the probability that the movement belongs to each dance style. The dance style corresponding to the maximum probability value is determined as the dance style to which the movement belongs. Each dance move is classified according to the dance style, and the number of moves contained in each dance style is counted as the number of dance style moves P. v . Calculate the sum of the probability values ​​corresponding to each action in each dance style as the style feature score Q v .

[0167] The style fusion index XE is calculated based on the dance style data and the total number of movements d2, and the formula is:

[0168]

[0169] Among them, P v is the number of movements of the vth dance style in dance training, v is the serial number corresponding to different dance styles, and its value is [1, u]; u is the total number of dance styles, and its value is a positive integer; Q v Score the style features of the vth dance style in dance training.

[0170] It should be noted that this formula calculates the number of different dance style movements P v The proportion of the total number of actions d2 and the sum of the corresponding style feature scores are then divided by the total number of dance styles to obtain the style fusion index XE. The higher the style fusion index XE, the better the user integrates different dance styles and the better the dance training effect.

[0171] In the above scheme, the action efficiency index XA is calculated according to the action accuracy score C and the physical exertion degree N, which helps to understand the balance between the user's action quality and physical exertion during dance training. The space and action coordination index XB is calculated according to the level change frequency index E, the action change frequency index F and the coordination data, which can more comprehensively evaluate the user's space utilization ability during dance training. The action amplitude score H and the action strength score I are calculated according to the coordination data, and the comprehensive expressiveness index XC is further calculated in combination with the emotional expressiveness score J1, which can evaluate the user's coordination between dance movements and expressiveness. By calculating the emotional rhythm resonance index XD, it is helpful to understand the user's effect in emotional expression and rhythm coordination. The style fusion index XE is calculated according to the dance style data and the total number of actions d2, which helps to evaluate the effect of dance works in integrating different styles.

[0172] The level judgment module is used to calculate the dance training score X according to the action efficiency index XA, the space and action coordination index XB, the comprehensive expression index XC, the emotional rhythm resonance index XD and the style fusion index XE; preset the level threshold set; and judge the user's dance training level according to the level threshold set and the dance training score X.

[0173] Specifically, the method for calculating the dance training score X is:

[0174] The dance training score X is calculated based on the action efficiency index XA, the space and action coordination index XB, the comprehensive expression index XC, the emotional rhythm resonance index XD and the style integration index XE. The formula is:

[0175] X=XA+XB+XC+XD+XE

[0176] It should be noted that this formula obtains the dance training score X by comprehensively considering indicators of multiple dimensions.

[0177] Specifically, the method for determining the user's dance training level is:

[0178] The grade threshold set includes an excellent threshold XY and a qualified threshold XH.

[0179] It should be noted that the method for determining the excellent threshold value XY and the qualified threshold value XH is as follows: collect historical data of different dancers in the past and calculate the corresponding dance training scores according to the above method to form a historical data set. Count the 50% with higher dance training scores in the historical data set and calculate their average and standard deviation. Calculate the value of the average plus the standard deviation as the reference data for the excellent threshold value XY. Calculate the value of the average minus the standard deviation as the reference data for the qualified threshold value XH.

[0180] The user's dance training level is determined based on the level threshold set and the dance training score X, and the criteria are as follows:

[0181]

[0182] In the above scheme, the dance training score X is calculated based on the action efficiency index XA, the space and action coordination index XB, the comprehensive expression index XC, the emotional rhythm resonance index XD and the style fusion index XE, which solves the problem of ignoring subjective factors such as emotional expression and the synergistic influence between various factors when evaluating dance training, thereby reducing the accuracy of the evaluation results. Judging the user's dance training level based on the level threshold set and the dance training score X helps users understand their own training situation and develop personalized training plans.

[0183] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. A person of ordinary skill in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented in electronic hardware or in combination with computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0184] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0185] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A multi-dimensional analysis dance training intelligent evaluation system, characterized by: include: A data collection module is used to collect dance training movement position data, movement integrity data, user body condition data, movement time data, coordination data, emotion data and dance style data; An index calculation module, used to calculate a posture conformity score A according to the action position data, calculate a movement integrity score B according to the movement integrity data, and further calculate a movement accuracy score C; The physical exertion degree N is calculated according to the user's physical state data and the action time data, and the action efficiency index XA is calculated according to the action accuracy score C and the physical exertion degree N; Calculate the level change frequency index E according to the action position data, and calculate the action change frequency index F according to the action position data and the action time data; calculate the space and action coordination index XB according to the level change frequency index E, the action change frequency index F and the coordination data; The movement amplitude score H and the movement intensity score I are calculated based on the coordination data, and the comprehensive expression index XC is further calculated based on the emotion data; The beat synchronization score M is calculated based on the action time data; the emotional rhythm resonance index XD is calculated based on the beat synchronization score M, the rhythm intensity change score MC, and the emotional data; Calculate the style fusion index XE based on the dance style data and the total number of movements d2; A level judgment module is used to calculate the dance training score X based on the action efficiency index XA, the space and action coordination index XB, the comprehensive expression index XC, the emotional rhythm resonance index XD and the style integration index XE; A set of level threshold values ​​is preset; and the user's dance training level is determined according to the level threshold value set and the dance training score X.

2. The multi-dimensional analysis dance training intelligent evaluation system according to claim 1 is characterized by: The method for calculating the movement accuracy score C is: The motion position data includes the user's dance motion coordinates (A a , B a , C a ), standard dance movement coordinates (AA a ,BB a , C.C. a ) and the maximum distance deviation value D a ; The posture conformity score A is calculated based on the action position data, based on the formula: Among them, (A a , B a , C a ) is the coordinate of the a-th dance action, a is the serial number corresponding to different dance actions, and its value is [1, b]; b is the total number of different dance actions, and its value is a positive integer; (AA a ,BB a , C.C. a ) is the coordinate of the ath standard dance action; D a is the maximum distance deviation allowed for the a-th dance action; The action integrity data includes the number of errors and omissions d1 and the total number of actions d2; The action integrity score B is calculated based on the action integrity data according to the formula: The action accuracy score C is calculated based on the posture compliance score A and the action integrity score B, according to the following formula: C=α1×A+α2×B Among them, α1 is the weight coefficient of the posture compliance score A, and its value is 0.3~0.7; α2 is the weight coefficient of the action integrity score B, and its value is 0.3~0.7; and α1+α2=1.

3. The multi-dimensional analysis dance training intelligent evaluation system according to claim 2 is characterized by: The method for calculating the action efficiency index XA is: User physical status data including average heart rate during training Average breathing rate during training Heart rate basic value N4 and respiratory rate basic value N5; The movement time data include the total dance time N1; The physical exertion degree N is calculated based on the user's physical condition data and the total dance time N1, and the formula is as follows: The action efficiency index XA is calculated based on the action accuracy score C and the physical exertion level N, based on the following formula:

4. The multi-dimensional analysis dance training intelligent evaluation system according to claim 3 is characterized by: The method for calculating the level change frequency index E and the action change frequency index F is: The action position data also includes the height coordinates E(e, f) of the body parts; According to the height coordinate E of the body part (e，f) The formula for calculating the level change frequency index E is as follows: Among them, E (e，f) is the height coordinate of the e-th body part at the f-th time point; e is the serial number corresponding to different body parts, and its value is [1, g]; g is the total number of body parts; f is the serial number corresponding to different time points, and its value is [1, h]; h is the total number of time points; E (e，f+1) is the height coordinate of the e-th body part at the f+1-th time point; The action time data also includes the action time point M a ; According to the user's dance movement coordinates (A a , B a , C a ) and action time point M a The action change frequency index F is calculated based on the formula: Among them, (A a+1 , B a+1 , C a+1 ) is the coordinate of the a+1th dance action; M a is the time point of the a-th dance action; M a+1 is the time point of the a+1th dance action.

5. The multi-dimensional analysis dance training intelligent evaluation system according to claim 4 is characterized by: The method for calculating the space and action coordination index XB is: Coordination data include joint coordination score G1 and body part coordination score G2; The spatial and motion coordination index XB is calculated based on the level change frequency index E, the action change frequency index F, the joint coordination score G1 and the body part coordination score G2, and the formula is: XB=E×F×(G1+G2).

6. The multi-dimensional analysis dance training intelligent evaluation system according to claim 5 is characterized by: The method for calculating the comprehensive performance index XC is: Coordination data also includes limb amplitude H a , target limb amplitude HA a 、Movement Strength I a , Target action intensity IA a and standard deviation of action strength, β; According to the limb amplitude H a and target limb amplitude HA a The formula for calculating the range of motion score H is: According to the action intensity I a , Target action intensity IA a The action intensity score I is calculated based on the formula: The emotional data include emotional expressiveness score J1; The comprehensive expression index XC is calculated based on the movement amplitude score H, the movement intensity score I and the emotional expression score J1, and the formula is: XC=H×I+J1.

7. The multi-dimensional analysis dance training intelligent evaluation system according to claim 6 is characterized by: The method for calculating the emotional rhythm resonance index XD is: The action time data also includes the music beat time point MA a and rhythm intensity change score MC; According to the action time point M a and music beat time point MA a The beat synchronization score M is calculated based on the formula: The emotional data also includes the emotional change intensity score J2; The emotional rhythm resonance index XD is calculated based on the beat synchronization score M, the rhythm intensity change score MC, the emotional expression score J1 and the emotional change intensity score J2. The formula is: Among them, γ is the anti-zero correction value.

8. The multi-dimensional analysis dance training intelligent evaluation system according to claim 7 is characterized by: The method for calculating the style fusion index XE is: The dance style data includes the number of dance style movements P v and style feature score Q v ; The style fusion index XE is calculated based on the dance style data and the total number of movements d2, and the formula is: Among them, P v is the number of movements of the vth dance style in dance training, v is the serial number corresponding to different dance styles, and its value is [1, u]; u is the total number of dance styles, and its value is a positive integer; Q v Score the style features of the vth dance style in dance training.

9. The multi-dimensional analysis dance training intelligent evaluation system according to claim 8, characterized in that: The method for calculating the dance training score X is: The dance training score X is calculated based on the action efficiency index XA, the space and action coordination index XB, the comprehensive expression index XC, the emotional rhythm resonance index XD and the style fusion index XE. The formula is: X=XA+XB+XC+XD+XE.

10. The multi-dimensional analysis dance training intelligent evaluation system according to claim 9, characterized in that: The method for determining the user's dance training level is: The grade threshold set includes an excellent threshold XY and a qualified threshold XH; The user's dance training level is determined based on the level threshold set and the dance training score X, and the criteria are as follows:

Images