Method for estimating battle rope exercise amount based on computer vision

Through computer vision technology, the movement trajectory and arm movement characteristics of the war rope are extracted, and the movement volume model is established, which solves the problems of inaccurate evaluation and complex installation in traditional methods, and realizes the accurate, real-time evaluation and training intensity of the war rope movement volume. Flexible adjustments.

CN119007293APending Publication Date: 2024-11-22SHANGHAI UNIV OF SPORT
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411121987.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2024-04-29
Filing Date
2024-08-15
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The prior art is difficult to accurately evaluate the amount of exercise in combat rope training. The traditional method relies on subjective estimation or simple counting, lacks accuracy, and the sensor installation is complex and inconvenient, and cannot provide real-time feedback, which limits the adjustment of training intensity.

Method used

Using a computer vision-based method, the combat rope movement trajectory images are obtained, the arm movement characteristics are extracted, and the motion quantity model is established. The image processing technology is used for preprocessing and feature extraction, and combined with frequency calculation, the precise, comprehensive and real-time evaluation of the combat rope movement quantity is achieved. .

Benefits of technology

It realizes global multi-dimensional information capture of battle rope swing, reduces system cost and installation difficulty, provides more detailed and accurate exercise data, supports real-time adjustment of training intensity, and improves training effect.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119007293A_ABST
    Figure CN119007293A_ABST
Patent Text Reader

Abstract

The invention discloses a battle rope exercise amount estimation method based on computer vision, and belongs to the technical field of human body posture recognition and physical exercise analysis. The method comprises the following steps of S1, according to the fact that a trainee swings a battle rope in different postures, obtaining a battle rope movement track image in a time interval by utilizing a computer vision technology; s2, extracting an arm action image, and obtaining arm swinging times in a time interval to obtain war rope swinging frequency; and S3, preprocessing the battle rope motion track image, obtaining a black-and-white image after preprocessing in a time interval, extracting features in the black-and-white image, establishing a battle rope motion amount model through the features and the battle rope swing frequency, and obtaining an optimized motion amount through optimization. According to the method, the comprehensive technology of human body key point detection, waveform data analysis and frequency calculation is adopted, the swinging state of the person and the battle rope is analyzed in real time and at multiple angles, a more comprehensive and accurate exercise amount estimation result is provided, and the visual detection technology has remarkable advantages.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a method for estimating the movement amount of a battle rope based on computer vision, and belongs to the technical field of human posture recognition and sports analysis. Background Art

[0002] As an efficient whole-body aerobic training equipment, battle ropes have been widely used in gyms, especially for people with joint injuries or degenerative diseases. It is favored by fitness enthusiasts for its low cost and ease of use, through free-trajectory strength training, without restrictions on age and body shape. It is expected to have potential sales in the Chinese gym market of 39,000 units, with a total output value of approximately RMB 360 million. In addition, battle ropes also have potential in dance physical training, and the sales scale in art and sports colleges is expected to be 5,000-6,000 units, with a total output value of approximately RMB 50 million.

[0003] There are many factors that determine the effectiveness of battle rope training, and the swing of the battle rope is one of them. Currently, it is still a challenge to effectively analyze and evaluate the swing of the battle rope to ensure the effectiveness and safety of training.

[0004] Many existing methods rely on subjective assessment or simple counting techniques, which usually lack sufficient accuracy. For example, some methods rely on manually counting the number of battle rope fluctuations or estimating the amount of exercise by observing the posture of the athlete. Especially in high-intensity training, this subjectivity can easily lead to inaccurate results. Traditional methods usually require specialized equipment or sensors to collect motion data, which not only increases the cost, but also increases the complexity and difficulty of use of the equipment. For example, some systems require the installation of multiple accelerometers or gyroscopes, which require precise calibration and maintenance, increasing the inconvenience of use. In addition, many existing methods cannot provide real-time feedback on the amount of exercise, which limits the athlete's ability to adjust the intensity of training immediately. This is crucial for efficient training and preventing overtraining. Real-time feedback can help athletes better adjust their exercise strategies to achieve better training results. Existing methods usually only focus on a specific aspect of exercise, such as calorie consumption or the activation of a specific muscle group, while ignoring the overall body movement. A comprehensive assessment of exercise volume needs to consider multiple aspects, including strength, endurance, cardiopulmonary function, etc., which is crucial for developing a comprehensive training plan. Some methods require athletes to wear complex equipment during training, which not only affects the exercise experience, but also may limit the naturalness and fluency of the exercise. For example, some systems require athletes to install sensors on battle ropes, which may affect the movement pattern and effect of battle ropes.

[0005] In contrast, the use of computer vision technology, through image processing and motion analysis algorithms, has achieved an accurate, comprehensive and real-time assessment of the amount of battle rope exercise, solving many problems of the above traditional methods. This non-contact, automated assessment method not only improves the accuracy and reliability of the data, but also greatly improves the user's training experience.

[0006] At present, the effect of battle rope swing is mainly judged by sports load monitoring, which has been used for a long time in the fields of sports science and sports medicine. It mainly relies on various high-precision sensors, namely accelerometers, gyroscopes, encoders and strain gauges, etc. The above sensors can accurately measure and record key parameters such as acceleration, angular velocity, displacement and stress when the battle rope is swinging. For example, an accelerometer is a sensor that can measure the acceleration of an object and is often used to monitor the linear motion state of the battle rope when it is swinging. Through the accelerometer, the acceleration data of the battle rope swing can be obtained in real time, and the dynamic characteristics of the swing such as frequency, amplitude and speed can be further calculated.

[0007] Although the traditional sensor monitoring method has certain advantages in battle rope swing analysis, it also has obvious limitations and challenges. First, the installation and debugging of sensors are usually complicated and time-consuming, requiring professional skills and experience. Second, due to the fixed position and layout of the sensors, the global and multi-dimensional motion information of the battle rope swing may not be fully captured, resulting in incomplete and inaccurate data. Summary of the invention

[0008] The present invention provides a method for estimating the amount of movement of a battle rope based on computer vision, which solves the problems of incomplete and inaccurate installation and data collection caused by the current use of sensors.

[0009] The technical problem to be solved by the present invention is achieved by adopting the following technical solutions:

[0010] A method for estimating the amount of movement of a battle rope based on computer vision comprises the following steps:

[0011] S1. According to the trainer swinging the battle rope in different postures, the computer vision technology is used to obtain the movement trajectory image of the battle rope within the time interval;

[0012] S2, extracting arm motion images according to the battle rope motion trajectory image in step S1, obtaining the number of arm swings within a time interval, and obtaining the battle rope swing frequency;

[0013] S3. Preprocess the battle rope motion trajectory image in step S1 to obtain a black-and-white image after preprocessing within a time interval, extract features from the black-and-white image, and establish a battle rope motion volume model through the features and the battle rope swing frequency in step S2, and obtain the optimized motion volume after optimization.

[0014] Preferably, before performing step S1, the battle rope swing video input needs to be obtained in the following manner:

[0015] Fix one end of the battle rope on a device and hold the battle rope with the other end. Select a frame rate of 30-120 and a resolution of 1920*1080 to provide sufficient image quality and details for subsequent processing. Make sure that the camera is placed at a 90-degree angle to the battle rope, and the position can fully ensure that the wide angle of the lens can capture the panoramic view of the battle rope movement to avoid obstruction or distortion. The lens is installed by roof hoisting. Furthermore, a retractable fixture is used. There are many retractable fixtures on the market, such as an air cylinder, an electric push rod or an oil cylinder with an up and down telescopic direction. The fixed end is hoisted on the beam or frame on the roof, and the telescopic end is installed with a camera through a bracket. When collecting data, the fixture is extended to a position close to the horizontal position of the battle rope movement to ensure the best viewing angle for collecting sports videos. When not collecting, the fixture is retracted to ensure that its indoor sports space is not occupied. When collecting outdoors, a fixed tripod can be used to support the lens collection.

[0016] The purpose of adopting the above scheme is to ensure that the input video has high definition and high frame rate, so as to ensure that the details and dynamic changes of the battle rope swing can be accurately captured. The video content should include the interaction between the human body and the battle rope in different postures, as well as the whole process of the battle rope swing.

[0017] Preferably, in step S2, the number of arm swings is obtained in the following manner to obtain the swing frequency;

[0018] S2.1 locates the arm action image in step S1 as indexi,y, defines the position of the hand holding the battle rope as the key point, takes this key point as the origin, and the direction of the vertical movement of the arm as the y-axis direction. Calculate the distance difference between the two key points on the y-axis within the time interval (the two key points refer to the positions of the two hands holding the battle rope):

[0019] distance=keypoint index1,y -keypoint index2,y ;

[0020] Among them, keypoint index1,y Indicates one of the key points, keypoint index2,y Indicates another key point;

[0021] S2.2 uses the distance difference calculated in S2.1 to determine the swing direction of the battle rope; define the swing upward as one time and the swing downward as the next time, that is:

[0022]

[0023] From the above, we can know that if distance < 0, the flag is 0, and the battle rope swings downward; if distance ≥ 0, the flag is 1, and the battle rope swings upward;

[0024] S2.3 Use computer vision technology to record the number of times the battle rope in step S2.2 is swung in the same direction within the time interval, and calculate its frequency, which is recorded as

[0025]

[0026] Preferably, in step S3, after the swinging image of the battle rope in the video is extracted, a preprocessing step is required to obtain a black and white image of the movement trajectory of the battle rope within the time interval.

[0027] Further, according to the order of preprocessing, the preprocessing includes binarization, corrosion and expansion processes.

[0028] Furthermore, the image is preprocessed by binarization, erosion and dilation as follows:

[0029] (1) The color image captured by the camera is converted into a grayscale image in the following manner; that is,

[0030] Gray=R*0.299+G*0.587+B*0.114;

[0031] Where: R, G and B are the pixel values ​​of the red, green and blue channels respectively; the weight coefficient is determined based on the human eye's perception of different colors, where the weight is 0.299 when perceived as red, the weight is 0.587 when perceived as green, and the weight is 0.114 when perceived as blue;

[0032] (2) The gray image is binarized in the following manner; that is,

[0033]

[0034] Where: C1 and C2 represent two color values ​​respectively. If black and white binary is used, C1 and C2 represent black and white colors. Threshold represents the threshold value. The threshold value range is [0,255]. Pixels are divided into two categories, C1 and C2, according to the threshold value range.

[0035] (3) Crop the key area of ​​the binary gray image; perform corrosion and expansion operations in the following manner; that is,

[0036] E(x,y)=min(a,b,c)

[0037] D(x,y)=max(a,b,c)

[0038] Where E is the result image of the erosion operation, D is the result image of the dilation operation, (x, y) is the pixel coordinate, a, b, c are the pixel values ​​in the area covered by the convolution kernel, and the specific values ​​depend on the design and application of the convolution kernel or structural element, and are automatically calculated when performing erosion and dilation operations.

[0039] Using the above scheme, the image processed by binarization, corrosion and expansion plays a vital role in the entire computer vision system.

[0040] First, binarization simplifies the image, highlighting the battle ropes and their important features, making subsequent processing more convenient.

[0041] The erosion operation further removes noise and shrinks the boundaries of foreground objects, ensuring that the battle rope outline is purer and more independent.

[0042] Subsequently, the dilation operation restored the size of the battle ropes and enhanced the important features in the image, making the battle ropes more visible in the image. The final processed image provides a clear basis for feature extraction and data analysis for mathematical models.

[0043] Preferably, in step S3, the battle rope model is established in the following manner:

[0044] S3.1 The black and white image of the battle rope motion trajectory obtained after preprocessing is used to extract the features, including wavelength, amplitude, peak, trough and the area change ΔA between two amplitudes calculated according to the following formula:

[0045] f(x)=Msin(Bx+C)+D;

[0046] g(x) = Ecos(Fx+G)+H;

[0047]

[0048] ΔA=A2-A1;

[0049] Where: A is the area between two amplitudes, f(x) and g(x) are functions between two adjacent peaks or troughs, x1 and x2 are the positions of the peaks or troughs; ΔA is the area change, which reflects the strength of the battle rope swing; M, E are the amplitudes of f(x) and g(x), respectively; B, F are the angular frequencies of f(x) and g(x), respectively, which determine the period of the waveform; C, G are the phase angles of f(x) and g(x), respectively, which determine the starting position of the waveform when the battle rope moves; D, H are the vertical displacements relative to the ground, which determine the position of the waveform on the y-axis when the battle rope moves;

[0050] S3.2 Establish a battle rope swing model based on the wavelength, amplitude, peak, trough and area change ΔA in step S1 and in combination with the frequency P in claim 2, that is:

[0051] Y = β0 + β1 × peak + β2 × trough + β3 × amplitude + β4 × wavelength + β5 × ΔA + β6 × P;

[0052] Among them, Y is the amount of movement, β1, β2, β3, β4, β5, and β6 are parameters of the model, and their values ​​are [0,1] respectively. The crest and trough represent the highest and lowest points reached by the battle rope during the swing process respectively; the amplitude is the difference between the crest and the trough, indicating the amplitude of the swing; the wavelength is obtained by calculating the horizontal distance between adjacent crests or troughs and taking their average value, and A represents the change in area, which reflects the strength of the swing of the battle rope.

[0053] Furthermore, the peak, trough, amplitude and wavelength can be calculated by the following steps:

[0054] (1) First, determine the key point data. Assume that the swing data of the battle rope is recorded in an array or list in the following format:

[0055] data=[y1,y2,y3,…,y n ]

[0056] Among them, y i is the position of the battle rope in the i-th frame;

[0057] (2) Re-identify the peaks and troughs, traverse the data, and find the local maximum and minimum values, that is, the highest point reached by the battle rope during the swing process is determined as the peak, and the lowest point reached by the battle rope during the swing process is determined as the trough;

[0058] (3) Finally, calculate the amplitude, which is the difference between the peak and the trough, indicating the amplitude of the swing. The wavelength is calculated by averaging the horizontal distance between adjacent peaks or troughs in the following way:

[0059]

[0060] or

[0061]

[0062] Where Wavelength represents wavelength; peaks represents peaks; valeys represents troughs; n is the number of peaks; m is the number of troughs; i represents the position index in the peak or trough index list;

[0063] S3.3 uses regularization to optimize the battle rope swing model to obtain the optimized battle rope swing model, that is, the battle rope movement volume.

[0064] Using the above scheme, the processed image data ensures the accuracy and reliability of parameters such as amplitude, wavelength and area change, thus providing accurate input data for the estimation of the movement volume of the battle rope swing. This process ensures that the battle rope motion model can accurately analyze and evaluate the movement volume of the battle rope, and realize scientific and effective evaluation of the battle rope training effect.

[0065] A battle rope swing model is established by the above method.

[0066] A computing device comprises a processor and a memory for storing a program executable by the processor. When the processor executes the program stored in the memory, the steps of the method described above are implemented.

[0067] A storage medium stores a program / instruction, wherein the program / instruction implements the steps of the above method when executed by a processor.

[0068] A computer program product comprises a computer program / instruction, which implements the steps of the method described above when executed by a processor.

[0069] The beneficial effects of the present invention are:

[0070] 1. Compared with existing methods, the computer vision method of the present invention can achieve global and multi-dimensional information capture of battle rope swing. Through the existing human key point detection based on the Mediapipe library, the method can capture the complex postures and movements of the human body, thereby providing richer and more detailed swing data. Traditional sensors may only provide limited data points, while computer vision methods can provide more comprehensive and detailed information.

[0071] 2. Compared with the traditional method that requires additional sensor equipment and complicated installation process, the computer vision technology of the present invention does not require additional hardware support. Only an ordinary camera is needed to obtain the required data, which greatly reduces the system cost and installation difficulty. Therefore, this method is more convenient and economical;

[0072] 3. This method uses a comprehensive technology based on human key point detection, waveform data analysis and frequency calculation, which can analyze the swinging state of people and battle ropes in real time and from multiple angles, thereby providing a more comprehensive and accurate estimation result of the amount of exercise, which makes the visual detection technology of the present invention have significant advantages. First, it can capture the information of the battle rope swing in all directions and from multiple angles, providing richer and more detailed data analysis. Secondly, this method does not require additional sensor equipment and complex installation processes, which greatly reduces the cost and maintenance difficulty of the system.

[0073] 4. The use of computer vision technology has strong flexibility and scalability, can adapt to various application scenarios and changes in demand, can be adjusted and optimized according to different training programs, and has strong flexibility and scalability. It can adapt to various application scenarios and changes in demand, can be adjusted and optimized according to different training programs, and through feature extraction, model construction and training evaluation, it can provide scientific training effect analysis and evaluation, and provide a basis for athletes to formulate reasonable training plans;

[0074] In summary, this method uses computer vision technology to achieve accurate, real-time and non-contact monitoring of the amount of battle rope movement. It has significant technical and application advantages and is suitable for a wide range of scenarios and needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 is a flow chart of the present invention;

[0076] Figure 2 A demonstration diagram of rope skipping in the present invention;

[0077] Figure 3 This is a diagram of the device used to configure the camera in the present invention. DETAILED DESCRIPTION

[0078] In order to make the technical means, creative features, objectives and effects of the present invention easier to understand, the present invention is further described below with reference to specific illustrations.

[0079] In response to the defects brought by the use of sensors, visual detection technology has gradually received attention and application in recent years, becoming an emerging trend in battle rope swing monitoring. By capturing the image or video data of the battle rope swing with a high-resolution camera, combined with advanced image processing and computer vision algorithms, real-time, accurate and non-contact monitoring of the battle rope swing can be achieved. Using image processing technology, the outline and morphological information of the battle rope can be clearly extracted, and key parameters such as the morphological changes, speed distribution and spatial position of the battle rope swing can be further analyzed. At the same time, computer vision algorithms can identify and track the key feature points of the battle rope swing, realizing dynamic analysis and pattern recognition of the swing process.

[0080] like Figure 1-3 As shown, a method for estimating the amount of movement of a battle rope based on computer vision comprises the following steps:

[0081] S1. According to the trainer swinging the battle rope in different postures, the computer vision technology is used to obtain the movement trajectory image of the battle rope within the time interval;

[0082] S2, extracting arm motion images according to the battle rope motion trajectory image in step S1, obtaining the number of arm swings within a time interval, and obtaining the battle rope swing frequency;

[0083] S3. Preprocess the battle rope motion trajectory image in step S1 to obtain a black-and-white image after preprocessing within a time interval, extract features from the black-and-white image, and establish a battle rope motion volume model through the features and the battle rope swing frequency in step S2, and obtain the optimized motion volume after optimization.

[0084] In step S1, according to the trainer swinging the battle rope in different postures, the computer vision technology is used to obtain the battle rope movement trajectory image within the time interval:

[0085] Fix one end of the battle rope on a device and hold the battle rope with the other end. Select a frame rate of 30-120 and a resolution of 1920*1080 to provide sufficient image quality and details for subsequent processing. Make sure that the camera is placed at a 90-degree angle to the battle rope, and the position can fully ensure that the wide angle of the lens can capture the panoramic view of the battle rope movement to avoid obstruction or distortion. The lens is installed by roof hoisting. Furthermore, a retractable fixture is used. There are many retractable fixtures on the market, such as an air cylinder, an electric push rod or an oil cylinder with an up and down telescopic direction. The fixed end is hoisted on the beam or frame on the roof, and the telescopic end is installed with a camera through a bracket. When collecting data, the fixture is extended to a position close to the horizontal position of the battle rope movement to ensure the best viewing angle for collecting sports videos. When not collecting, the fixture is retracted to ensure that its indoor sports space is not occupied. When collecting outdoors, a fixed tripod can be used to support the lens collection.

[0086] The purpose of adopting the above scheme is to ensure that the input video has high definition and high frame rate, so as to ensure that the details and dynamic changes of the battle rope swing can be accurately captured. The video content should include the interaction between the human body and the battle rope in different postures, as well as the whole process of the battle rope swing.

[0087] In step S2, the arm motion image is extracted according to the battle rope motion trajectory image in step S1, and the number of arm swings within the time interval is obtained to obtain the battle rope swing frequency;

[0088] S2.1 locates the arm action image in step S1 as indexi,y, defines the position of the hand holding the battle rope as the key point, takes this key point as the origin, and the direction of the vertical movement of the arm as the y-axis direction. Calculate the distance difference between the two key points on the y-axis within the time interval (the two key points refer to the positions of the two hands holding the battle rope):

[0089] distance=keypoint index1,y -keypoint index2,y ;

[0090] Among them, keypoint index1,y represents one of the key points,

[0091] keypoint index2,y Indicates another key point;

[0092] S2.2 uses the distance difference calculated in S2.1 to determine the swing direction of the battle rope; define the swing upward as one time and the swing downward as the next time, that is:

[0093]

[0094] From the above, we can know that if distance < 0, the flag is 0, and the battle rope swings downward; if distance ≥ 0, the flag is 1, and the battle rope swings upward;

[0095] S2.3 Use computer vision technology to record the number of times the battle rope in step S2.2 is swung in the same direction within the time interval, and calculate its frequency, which is recorded as

[0096]

[0097] It should be noted that the time interval refers to the time period between t1 and t2, for example.

[0098] In step S2, image binarization and smoothing are performed in the following manner:

[0099] S2.1 converts the color image captured by the camera into a grayscale image in the following manner; namely,

[0100] Gray=R*0.299+G*0.587+B*0.114;

[0101] Where: R, G and B are the pixel values ​​of the red, green and blue channels respectively; the weight coefficient is determined based on the human eye's perception of different colors, where the weight is 0.299 when perceived as red, the weight is 0.587 when perceived as green, and the weight is 0.114 when perceived as blue;

[0102] S2.2 The gray image is binarized in the following manner; namely,

[0103]

[0104] Where: C1 and C2 represent two color values ​​respectively. If black and white binary is used, C1 and C2 represent black and white colors. Threshold represents the threshold value. The threshold value range is [0,255]. Pixels are divided into two categories, C1 and C2, according to the threshold value range.

[0105] S2.3 cuts out the key area from the binarized gray image, and finally performs corrosion and expansion operations in the following manner; that is,

[0106] E(x,y)=min(a,b,c)

[0107] D(x,y)=max(a,b,c)

[0108] Where E is the result image of the erosion operation, D is the result image of the dilation operation, (x, y) is the pixel coordinate, a, b, c are the pixel values ​​in the area covered by the convolution kernel, and the specific values ​​depend on the design and application of the convolution kernel or structural element, and are automatically calculated when performing erosion and dilation operations.

[0109] The images processed by binarization, corrosion and dilation play a vital role in the entire computer vision system.

[0110] First, binarization simplifies the image to highlight the battle ropes and their important features, making subsequent processing more convenient.

[0111] The erosion operation further removes noise and shrinks the boundaries of foreground objects, ensuring that the battle rope outline is purer and more independent.

[0112] Subsequently, the dilation operation restores the size of the battle rope and enhances important features in the image, making the battle rope more visible in the image.

[0113] The final processed images provide a clear basis for feature extraction and data analysis for mathematical models. Using these processed images, key points and feature data such as peaks, troughs and amplitudes can be extracted;

[0114] As the input variable of the model; used to train the multivariate linear regression model and Lasso regularization model, optimize the model parameters and improve the prediction accuracy.

[0115] In specific applications, the processed image data ensures the accuracy and reliability of parameters such as amplitude, wavelength and area change, thus providing accurate input data for estimating the amount of motion of the battle rope swing. This process ensures that the mathematical model can accurately analyze and evaluate the amount of motion of the battle rope, and achieve scientific and effective evaluation of the effect of battle rope training.

[0116] In step S3, a battle rope swing model is established in the following manner;

[0117] S3.1 The black and white image of the battle rope motion trajectory obtained after preprocessing is used to extract the features, including wavelength, amplitude, peak, trough and the area change ΔA between two amplitudes calculated according to the following formula:

[0118] f(x)=Msin(Bx+C)+D;

[0119] g(x) = Ecos(Fx+G)+H;

[0120]

[0121] ΔA=A2-A1;

[0122] Where: A is the area between two amplitudes, f(x) and g(x) are functions between two adjacent peaks or troughs, x1 and x2 are the positions of the peaks or troughs; ΔA is the area change, which reflects the strength of the battle rope swing; M, E are the amplitudes of f(x) and g(x), respectively; B, F are the angular frequencies of f(x) and g(x), respectively, which determine the period of the waveform; C, G are the phase angles of f(x) and g(x), respectively, which determine the starting position of the waveform when the battle rope moves; D, H are the vertical displacements relative to the ground, which determine the position of the waveform on the y-axis when the battle rope moves;

[0123] S3.2 Establish a battle rope swing model based on the wavelength, amplitude, peak, trough and area change ΔA in step S1 and in combination with the frequency P in claim 2, that is:

[0124] Y = β0 + β1 × peak + β2 × trough + β3 × amplitude + β4 × wavelength + β5 × ΔA + β6 × P;

[0125] Where Y is the amount of movement, β1, β2, β3, β4, β5, and β6 are parameters of the model, and their values ​​are [0, 1], respectively. The peak and trough represent the highest and lowest points reached by the battle rope during the swing process, respectively; the amplitude is the difference between the peak and the trough, indicating the amplitude of the swing; the wavelength is obtained by calculating the horizontal distance between adjacent peaks or troughs and taking their average value, and A represents the area change, which reflects the strength of the swing of the battle rope;

[0126] Furthermore, the peak, trough, amplitude and wavelength can be calculated by the following steps:

[0127] (1) First, determine the key point data. Assume that the swing data of the battle rope is recorded in an array or list in the following format:

[0128] data=[y1,y2,y3,…,y n ]

[0129] Among them, y i is the position of the battle rope in the i-th frame;

[0130] (2) Re-identify the peaks and troughs, traverse the data, and find the local maximum and minimum values, that is, the highest point reached by the battle rope during the swing process is determined as the peak, and the lowest point reached by the battle rope during the swing process is determined as the trough;

[0131] (3) Finally, calculate the amplitude, which is the difference between the peak and the trough, indicating the amplitude of the swing. The wavelength is calculated by averaging the horizontal distance between adjacent peaks or troughs in the following way:

[0132]

[0133] or

[0134]

[0135] Where Wavelength represents wavelength; peaks represents peaks; valeys represents troughs; n is the number of peaks; m is the number of troughs; i represents the position index in the peak or trough index list;

[0136] S3.3 According to the Y obtained in step S3.2, the model Y is trained, and the model is optimized using the following method based on Lasso regularization to predict the actual movement of the battle rope, which is used to reduce the complexity of the model, prevent overfitting, and select the most important feature variables:

[0137] Right now:

[0138]

[0139] in: is the residual sum of squares of the model, which represents the difference between the observed value and the model prediction value. The observed value is obtained by actually measuring the amount of exercise during battle rope training, and the model prediction value is calculated by the battle rope swing model.

[0140] Specifically, the predicted amount of motion is calculated based on the given characteristic values ​​(peaks, troughs, amplitudes, wavelengths, and area changes); is the L1 regularization term, which is used to limit the complexity of the model and feature selection; Y i is the motion of the ith observation; β0 is the intercept term, which indicates the predicted value when all features (peak, trough, amplitude, wavelength, area change) are 0; β1, β2, ..., β p are the parameters of the model; X 1i ,X 2i ,……,X pi is the eigenvalue of the ith observation; n is the number of observations; λ is the regularization parameter; and p is the number of features.

[0141] A battle rope swing model is established by the above method.

[0142] A computing device comprises a processor and a memory for storing a program executable by the processor. When the processor executes the program stored in the memory, the steps of the above method are implemented.

[0143] A storage medium stores a program / instruction, wherein the program / instruction implements the steps of the above method when executed by a processor.

[0144] A computer program product comprises a computer program / instruction, wherein the computer program / instruction implements the steps of the above method when executed by a processor.

[0145] In summary, the specific process of evaluating exercise volume by this model is as follows:

[0146] First, a high-resolution camera is used to capture video data of the battle rope movement. The video content should include the interaction between the human body and the battle rope in different postures, as well as the whole process of the battle rope swinging. The collected video data is then preprocessed, including grayscale, binarization, corrosion and dilation operations, to obtain a battle rope motion trajectory image with obvious characteristics. The key feature data of the battle rope, such as peaks, troughs, amplitude, wavelength and area change (ΔA), are extracted from the preprocessed image. These feature data reflect the morphological changes, velocity distribution and spatial position of the battle rope swing. Computer vision technology is used to extract arm action images from the battle rope motion trajectory image, calculate the number of arm swings, and thus obtain the battle rope swing frequency. The specific method includes locating the position of the hand holding the battle rope as the key point, calculating the distance difference of the key point in the time interval, judging the swing direction of the battle rope by the distance difference, recording the number of swings in the same direction and calculating its frequency. The battle rope motion model is established using the extracted feature data. A multivariate linear regression model is adopted and optimized using the Lasso regularization method. The regularization method prevents overfitting by reducing the complexity of the model. The input variables of the model include characteristic data such as amplitude, wavelength and area change, and the output variable is the movement of the battle rope, that is, the energy consumed or power generated by the swing. In practical applications, the video of the battle rope swing is captured in real time, and new feature data is generated through image processing and feature extraction to input into the model, and the model outputs the current movement of the battle rope. By continuously collecting and analyzing data, the model can be further adjusted and optimized to meet the needs of different trainers and the requirements of different training stages.

[0147] In the specific calculation process, the model equation obtains the predicted value of the motion amount by inputting characteristics such as amplitude, wavelength and area change;

[0148] Lasso regularization introduces a regularization term to draw a regularization path during the optimization process. By observing the regularization path, the sparse model coefficient corresponding to the regularization parameter value is found, so as to select the best regularization parameter and achieve the parameter β0-β p The identification of the Y model is finally completed to improve the interpretability and prediction accuracy of the Y model.

[0149] The Y model evaluates the intensity and effect of battle rope training by outputting the amount of exercise value. This amount of exercise can guide trainers to adjust the training intensity, optimize the training effect, and provide a scientific basis for training evaluation.

[0150] By continuously collecting and analyzing data, the model can be further adjusted and optimized to meet the needs of different trainers and the requirements of different training stages. Therefore, the mathematical model comprehensively evaluates the amount of movement of the battle rope through feature extraction, model construction, training and predictive analysis, thereby providing accurate analysis and evaluation of training effects.

[0151] Compared with the traditional sensor monitoring method, the present invention adopts a comprehensive technology based on human key point detection, waveform data analysis and frequency calculation, which can analyze the swinging state of people and battle ropes in real time and from multiple angles, thereby providing more comprehensive and accurate motion estimation results, so that the visual detection technology of the present invention has significant advantages. First, it can capture the information of the swing of the battle rope in an all-round and multi-angle manner, providing richer and more detailed data analysis. Secondly, this method does not require additional sensor equipment and complex installation processes, which greatly reduces the cost and maintenance difficulty of the system. Finally, the visual detection technology of the present invention has strong flexibility and scalability, can adapt to various application scenarios and changes in demand, and has broad application potential and market prospects.

[0152] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for estimating the amount of movement of a battle rope based on computer vision, characterized in that: The following steps are involved: S1. According to the trainer swinging the battle rope in different postures, the computer vision technology is used to obtain the movement trajectory image of the battle rope within the time interval; S2, extracting arm motion images according to the battle rope motion trajectory image in step S1, obtaining the number of arm swings within a time interval, and obtaining the battle rope swing frequency; S3. Preprocess the battle rope motion trajectory image in step S1 to obtain a black-and-white image after preprocessing within a time interval, extract features from the black-and-white image, and establish a battle rope motion volume model through the features and the battle rope swing frequency in step S2, and obtain the optimized motion volume after optimization.

2. The method for estimating the amount of movement of a battle rope based on computer vision according to claim 1, characterized in that: In step S2, the arm swing times are obtained in the following manner to obtain the swing frequency; S2.1 locates the arm action image in step S1 as indexi,y, defines the position of the hand holding the battle rope as the key point, takes this key point as the origin, and the direction of the vertical movement of the arm as the y-axis direction. Calculate the distance difference between the two key points on the y-axis within the time interval (the two key points refer to the positions of the two hands holding the battle rope): distamce=keypoint index1,y -keypoint index2,y ; Among them, keypoint index1,y Indicates one of the key points, keypoint index2,y Indicates another key point; S2.2 uses the distance difference calculated in S2.1 to determine the swing direction of the battle rope; define the swing upward as one time and the swing downward as the next time, that is: From the above, we can know that if distance < 0, the flag is 0, and the battle rope swings downward; if distance ≥ 0, the flag is 1, and the battle rope swings upward; S2.3 Use computer vision technology to record the number of times the battle rope in step S2.2 is swung in the same direction within the time interval, and calculate its frequency, which is recorded as 3. The method for estimating the amount of movement of a battle rope based on computer vision according to claim 2, characterized in that: In step S3, the battle rope model is established in the following manner: S3.1 The black and white image of the battle rope motion trajectory obtained after preprocessing is used to extract the features, including wavelength, amplitude, peak, trough and the area change ΔA between two amplitudes calculated according to the following formula: f(x)=Msin(Bx+C)+D; g(x) = Ecos(Fx+G)+H; ΔA=A2-A1; Where: A is the area between two amplitudes, f(x) and g(x) are functions between two adjacent peaks or troughs, x1 and x2 are the positions of these two peaks or troughs; ΔA is the area change, which reflects the strength of the battle rope swing; M, E are the amplitudes of f(x) and g(x), respectively; B, F are the angular frequencies of f(x) and g(x), respectively, which determine the period of the waveform; C, G are the phase angles of f(x) and g(x), respectively, which determine the starting position of the waveform when the battle rope moves; D, H are the vertical displacements relative to the ground, which determine the position of the waveform on the y-axis when the battle rope moves; S3.2 Establish a battle rope swing model based on the wavelength, amplitude, peak, trough and area change ΔA in step S1 and in combination with the frequency P in claim 2, that is: Y = β0 + β1 × peak + β2 × trough + β3 × amplitude + β4 × wavelength + β5 × ΔA + β6 × P; where Y is the amount of movement, β1, β2, β3, β4, β5, β6 are parameters of the model, and their values ​​are [0,1], respectively. The peak and trough represent the highest and lowest points reached by the battle rope during the swing process; the amplitude is the difference between the peak and the trough, indicating the amplitude of the swing; the wavelength is obtained by calculating the horizontal distance between adjacent peaks or troughs and taking their average value, and A represents the area change, which reflects the strength of the swing of the battle rope; S3.3 uses regularization to optimize the battle rope swing model to obtain the optimized battle rope swing model, that is, the battle rope movement volume.

4. A computing device, characterized in that The invention comprises a processor and a memory for storing a program executable by the processor, and when the processor executes the program stored in the memory, the steps of the method according to any one of claims 1 to 3 are implemented.

5. A storage medium storing a program / instruction, characterized in that: When the program / instruction is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

6. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method described in any one of claims 1 to 3 are implemented.

Citation Information

Patent Citations

  • CPR intelligent training management method and system using computer vision

    CN115350081A

  • Physical education teaching test method and system based on human body posture estimation

    CN116503954A

  • Intelligent battle rope training device, system and method

    CN116585656A

  • Sports Swing Mechanics Training Device

    US20150087446A1

  • Battle Rope Training Device

    US20150265869A1