A visual-based double jump detection method and system for skipping rope
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGDONG PROPHET BIG DATA CO LTD
- Filing Date
- 2026-02-25
- Publication Date
- 2026-06-05
Smart Images
Figure CN122157094A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer vision technology, and in particular to a vision-based method and system for detecting double jump rope. Background Technology
[0002] With the rapid development of computer vision technology, vision-based sports analysis and assisted training systems have been widely applied. As a basic and efficient fitness exercise, rope skipping, especially the identification and counting of advanced techniques such as "double jumps" (where the rope wraps around the body twice during a single jump), is crucial for quantifying training effectiveness and improving the scientific nature of training.
[0003] In existing technologies, a common approach is to use a camera to capture motion video and extract the sequence of key points on the human skeleton using a pose estimation algorithm. By analyzing the vertical height changes of key points on the feet or ankles, each complete take-off and landing process is identified, and the number of basic jump rope repetitions is counted accordingly. This method achieves non-contact detection and counting of single jump movements.
[0004] However, existing technologies rely solely on vertical positional change information and cannot capture the unique movement characteristics required to complete two rope jumps within a single airborne period, which differ from single jumps (such as shorter airtime distribution and specific joint angle changes). This results in an inability to achieve accurate and robust detection of double jump movements, thus rendering existing visual methods functionally deficient in the analysis of advanced rope skipping skills and the guidance of refined training. Summary of the Invention
[0005] In view of the shortcomings of the prior art, the purpose of the invention is to provide a vision-based method and system for detecting double jump rope movements. The invention addresses the fact that the existing technology relies solely on vertical positional change information, which fails to capture the unique movement characteristics required to complete two jumps within a single airborne period (such as shorter airtime distribution, specific joint angle changes, etc.), which are different from single jumps. This results in the inability to achieve accurate and robust detection of double jump movements, and consequently, the technical problem that existing vision methods have functional deficiencies in the analysis of advanced jump rope skills and refined training guidance.
[0006] In a first aspect, the present invention proposes a vision-based method for detecting double jumps in a jump rope, the method comprising: S1, obtain key information of the entire rope skipping process; S2, based on the changes in the vertical position of key points, determine the start time, highest point time, and landing time for each rope skipping session; S3 extracts multiple jumping features during each jump rope session based on the take-off time, the highest point time, and the landing time. S4. Based on all jumping characteristics of the current jump rope sequence and pre-stored historical jump rope data, cluster analysis is used to calculate the initial boundary threshold and the historical data boundary threshold. S5, weighted fusion of the initial boundary threshold and the historical data boundary threshold to determine the final boundary threshold for each fluctuation feature; S6, calculate the comprehensive deviation value of the jumping characteristics of each jump compared to the final boundary threshold; S7, determine whether the overall deviation value is greater than the judgment threshold; if so, determine that the jump is a double jump; otherwise, determine that the jump is a single jump.
[0007] Furthermore, the key point information specifically includes: left shoulder coordinates, right shoulder coordinates, left elbow coordinates, right elbow coordinates, left hand coordinates, right hand coordinates, left hip coordinates, right hip coordinates, left foot coordinates, and right foot coordinates.
[0008] Furthermore, S1 specifically refers to: acquiring key information from each frame of the entire rope-jumping process of the inspector through a camera.
[0009] Furthermore, S2 specifically includes: S201, Calculate the jump height for each frame based on key point information; S202, based on the jump height, determine the jump state of each frame, where the jump state includes the high jump state, the low jump state, and the airborne state. S203, extract the time frames of all jump high points and construct a set of jump high points; S204, based on the time frames of each high jump point in the high jump point set, find the nearest low jump point state time frame before the high jump point time frame and use it as the start time; S205: Based on the same high jump time frame, find the nearest low jump state time frame after the high jump time frame and use it as the landing time. S206 combines the take-off time, the jump high point time frame, and the landing time to form a jump event; S207 outputs the time series of all jump events, including the take-off time, the highest point time, and the landing time.
[0010] Furthermore, the jumping characteristics specifically include: the hang time height, weighted displacement, and shaking intensity during each jump rope session.
[0011] Furthermore, S4 specifically includes: S401, calculate the statistics corresponding to all jumping features in the current jump rope sequence; S402: Based on historical jump rope data, single jump and double jump characteristics are obtained through manual annotation. S403, based on single-hop jump characteristics and double-hop jump characteristics, calculate single-hop statistical characteristics and double-hop statistical characteristics; S404, based on statistics, calculates the initial boundary threshold using the K-means clustering algorithm; S405 calculates the historical data boundary threshold based on single-hop statistical features and double-hop statistical features.
[0012] Furthermore, S5 specifically includes: S501, determine the fusion weight based on the total number of jump events in the current jump rope sequence and the total number of historical samples; S502, based on the fusion weight, performs weighted fusion of the initial boundary threshold and the historical data boundary threshold to obtain the final boundary threshold corresponding to each jump feature.
[0013] Furthermore, S6 specifically includes: S601, calculate the normalized offset of each jitter feature relative to the final boundary threshold; S602 uses the Sigmoid function to convert each normalized offset into a corresponding deviation value; S603, sum and average the deviation values to obtain the comprehensive deviation value.
[0014] In a second aspect, the present invention provides a vision-based double jump rope detection system, comprising: a memory and a processor; The memory contains an application program adapted for execution by a processor to implement the first aspect of the vision-based double jump detection method for rope skipping.
[0015] In a third aspect, the present invention provides a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the vision-based double jump detection method of the first aspect.
[0016] The beneficial effects of this invention are as follows: In this embodiment of the invention, the take-off, highest point, and landing time of each jump are extracted, and multiple refined jumping features are calculated accordingly. These features comprehensively reflect unique movement patterns, including time distribution in the air and speed changes. By combining the current sequence with historical data, the optimal final threshold for different features is dynamically determined, enhancing the system's adaptability to different individuals and jumping styles. By calculating the comprehensive deviation of each jump from the final threshold, minute but crucial feature deviations that conform to the double jump movement pattern can be intelligently identified, achieving high-accuracy double jump judgment. This significantly improves the capabilities of the visual solution in advanced jump rope skill analysis and refined training guidance. Attached Figure Description
[0017] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. It is obvious that the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings.
[0018] Figure 1 This is a flowchart illustrating a vision-based double jump detection method for rope skipping provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of a vision-based double jump rope detection system provided in an embodiment of the present invention. Detailed Implementation
[0019] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.
[0020] To enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0021] Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts disclosed in this invention.
[0022] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The terms "installed," "connected," and "linked" should be interpreted broadly; for example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.
[0023] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of methods and systems consistent with some aspects of the invention as detailed in the appended claims.
[0024] This invention proposes a vision-based method and system for detecting double jumps in rope skipping. It addresses the shortcomings of existing technologies that rely solely on vertical positional changes, failing to capture the unique movement characteristics required to complete two rope jumps within a single airborne period (such as shorter airtime distribution and specific joint angle changes). This results in an inability to accurately and robustly detect double jumps, hindering the application of existing vision-based methods in analyzing advanced rope skipping skills and providing refined training guidance.
[0025] Method Implementation Examples Reference Figure 1 The diagram shows a flowchart of a vision-based double jump rope detection method provided by an embodiment of the present invention.
[0026] This invention provides a vision-based method for detecting double jumps in a jump rope, the method comprising: Specifically, the method includes steps S1 to S7.
[0027] S1, obtain key information of the entire rope skipping process.
[0028] Optionally, the key point information specifically includes: left shoulder coordinates, right shoulder coordinates, left elbow coordinates, right elbow coordinates, left hand coordinates, right hand coordinates, left hip coordinates, right hip coordinates, left foot coordinates, and right foot coordinates.
[0029] In one possible implementation, S1 specifically involves: acquiring key information of each frame of the entire rope-jumping process of the inspector using a camera.
[0030] Specifically, the inspector faces the camera. After the test begins, the camera acquires the key point coordinates of the inspector in each frame, including the coordinates of the left shoulder. Right shoulder coordinates Left elbow coordinates Right elbow coordinates Left-handed coordinates Right-hand coordinates Left hip coordinates Right hip coordinates Left foot coordinates Right foot coordinates To detect double-jumping motion, the selected camera must have a frame rate of at least 60 frames per second.
[0031] S2, based on the vertical position changes of key point information, determine the start time, highest point time, and landing time for each rope skipping session.
[0032] In one possible implementation, S2 specifically includes sub-steps S201 to S207: S201, calculate the jump height for each frame based on key point information.
[0033] Specifically, the jump height is calculated as follows: in, This represents the jump height in the i-th frame. This represents the ordinate of the left hip bone in the i-th frame. This represents the ordinate of the right hip bone in the i-th frame. The ordinate of the left ankle in the i-th frame. This represents the ordinate of the right ankle in the i-th frame.
[0034] S202, based on the jump height, determine the jump state of each frame, where the jump state includes the high jump state, the low jump state, and the airborne state.
[0035] Specifically, obtain the jumping state of the personnel in each frame i (i>3). : in, This indicates the jump state of the i-th frame. This indicates the jump height of the first two frames. This indicates the jump height of the previous frame. Indicates the jump height of the next two frames. Indicates the jump height of the next frame. This represents the function that takes the minimum value.
[0036] It should be noted that, This represents the state at the highest point of the jump. This represents the low point state of the jump. This indicates a jumping, airborne state.
[0037] S203, extract the jump high point time frames of all jump high point states and construct a jump high point set.
[0038] S204: Based on the time frames of each high jump point in the set of high jump points, find the nearest low jump point state time frame before the high jump point time frame and use it as the start time.
[0039] S205: Based on the same high jump time frame, find the nearest low jump state time frame after the high jump time frame and use it as the landing time.
[0040] Specifically, obtaining the jump status of personnel For each time frame of the high jump state Where j represents the sequence number of the jump high point state, This indicates the number of jump high point states. Then, obtain the time frame for each jump high point state. The state of the nearest jump low point ( ) time frame ,when If there is no low point state before, then let and time frames The state of the nearest jump low point (after) ) time frame .
[0041] S206 combines the take-off time, the jump height time frame, and the landing time to form a jump event.
[0042] S207 outputs the time series of all jump events, including the take-off time, the highest point time, and the landing time.
[0043] Specifically, it integrates all jump high point state time frames to form a jump high point set. and the set of jumping times .
[0044] In this embodiment of the invention, by dynamically analyzing the height changes of key points and identifying the jumping state, the take-off, highest point and landing time of each jump are accurately extracted, providing a reliable temporal basis for subsequent double jump detection with multi-feature fusion, while also having good real-time performance, adaptability and fault tolerance.
[0045] S3 extracts multiple jumping features during each jump rope session based on the take-off time, the highest point time, and the landing time.
[0046] Optionally, the jumping characteristics specifically include: the hang time height, weighted displacement, and shaking intensity during each jump.
[0047] Among them, hang time refers to the maximum vertical fluctuation of the human body during a jump, from takeoff to landing.
[0048] The weighted displacement refers to the average displacement intensity of the body (represented by the hip bone) in the horizontal direction during a jump. This displacement is dynamically weighted over time, giving higher weight to the displacement near the highest point of the jump.
[0049] The jitter intensity refers to the degree of drastic change in body displacement during a jump, which is obtained by calculating the weighted average of the relative displacement changes between adjacent frames.
[0050] Specifically, the calculation method for hang time is as follows: in, This represents the hang time height during the j-th jump. This represents the function that takes the maximum value. The x-coordinate of the left ankle in the i-th frame represents the take-off time frame of the j-th jump. The x-coordinate of the right ankle in the i-th frame represents the take-off time frame of the j-th jump. This represents the x-coordinate of the left ankle in the i-th frame, representing the landing time frame of the j-th jump. The x-coordinate of the right ankle in the i-th frame represents the landing time frame of the j-th jump.
[0051] Furthermore, the specific formula for calculating the weighted displacement is as follows: in, This represents the weighted displacement of the j-th jump. This represents the weighted displacement coefficient of the i-th frame. This represents the relative displacement of the i-th frame. This represents the landing time frame of the j-th jump. This represents the start time frame of the j-th jump.
[0052] in, This represents the time frame of the highest point of the j-th jump. This indicates that the current frame i is after the highest point. This represents the x-coordinate of the left hip bone in the i-th frame. This represents the x-coordinate of the left hip bone in the (i-1)th frame. This represents the ordinate of the left hip bone in the i-th frame. This represents the x-coordinate of the left hip bone in the (i-1)th frame. The x-coordinate of the right hip bone in the i-th frame The x-coordinate of the right hip bone in frame i-1. The vertical coordinate of the right hip bone in the i-th frame. The vertical coordinate of the right hip bone in frame i-1.
[0053] Furthermore, the specific formula for calculating the jitter intensity is as follows: in, This represents the jitter intensity of the j-th jump.
[0054] In this embodiment of the invention, by extracting and quantifying three complementary motion features during the rope skipping process—hang height, weighted displacement, and shaking intensity—a multi-dimensional and fine-grained basis for accurately distinguishing between single and double jumps is provided, effectively improving the accuracy and robustness of motion classification.
[0055] S4. Based on all jumping characteristics of the current jump rope sequence and pre-stored historical jump rope data, cluster analysis is used to calculate the initial boundary threshold and the historical data boundary threshold.
[0056] In one possible implementation, S4 specifically includes sub-steps S401 to S405: S401, calculate the statistics corresponding to all jumping features in the current rope skipping sequence.
[0057] Specifically, the statistics used to calculate the eigenvalues are as follows: The specific calculation method for the average values of all fluctuation characteristics is as follows: in, This represents the average hang time. This represents the average of the weighted displacements. This represents the average value of the jitter intensity.
[0058] The specific calculation methods for the standard deviations of all fluctuation characteristics are as follows: in, The standard deviation of hang time is represented by the following: The standard deviation of the weighted displacement is represented by the following: The standard deviation represents the intensity of the jitter.
[0059] The specific 25th quantiles of all jump characteristics are as follows: in, Represents the 25th percentile of airborne height, This represents the 25th quantile of the weighted displacement. This represents the 25th quantile of the jitter intensity.
[0060] The 25th quartile, also known as the "lower quartile," refers to the value that is in the top 25% after a set of data is sorted from smallest to largest.
[0061] The medians for all jump characteristics are as follows: in, Represents the median of the airborne height, This represents the median of the weighted displacements. This represents the median of the jitter intensity.
[0062] The specific 75th quantiles of all jump characteristics are as follows: in, Represents the 75th percentile of airborne height, This represents the 75th quantile of the weighted displacement. This represents the 75th quantile of the jitter intensity.
[0063] The 75th quartile, also known as the "upper quartile," refers to the value that is in the top 75% after a set of data has been sorted.
[0064] S402, based on historical jump rope data, obtains single jump and double jump characteristics through manual annotation.
[0065] S403 calculates single-hop statistical features and double-hop statistical features based on single-hop and double-hop features.
[0066] Specifically, based on collected historical jump rope videos, each jump is manually labeled as either a single jump or a double jump. The hang time, weighted displacement, and jitter intensity of each single jump are obtained, and corresponding statistical characteristics are calculated, including: Average hang time of a single jump Average value of single-hop weighted displacement Average intensity of single jump .
[0067] Standard deviation of single-jump hang time Standard deviation of single-jump weighted displacement Standard deviation of single-jump jitter intensity .
[0068] Furthermore, the hang time, weighted displacement, and jitter intensity of each double jump are obtained, and the corresponding statistical characteristics are calculated, specifically including: Average hang time of double jump Double-jump weighted average displacement Average intensity of double jump jitter .
[0069] Double jump hang time standard deviation Standard deviation of double-jump weighted displacement Standard deviation of double-jump jitter intensity .
[0070] S404 calculates the initial boundary threshold based on statistics using the K-means clustering algorithm.
[0071] K-means clustering is an unsupervised machine learning algorithm used to divide a set of data points into K mutually exclusive clusters, such that data points within the same cluster are as similar as possible, and data points between different clusters are as dissimilar as possible.
[0072] Specifically, K-means clustering (K=2) was used to identify the centers of the low-value and high-value clusters for the three feature sets of this rope skipping exercise. The centers of the low-value clusters for hang time were: The high value cluster center of the hangar height Low-value cluster centers of weighted displacement The high-value cluster center of the weighted displacement Low-value cluster center of jitter intensity The high value cluster center of the jitter intensity .
[0073] Further, calculate the corresponding boundary threshold: in, The initial threshold representing the hang time. This represents the initial boundary threshold for the weighted displacement. The initial threshold representing the intensity of jitter.
[0074] When a clear bimodal distribution cannot be identified, the cutoff threshold is: in, This represents the adjustment coefficient, and its value range is... .
[0075] Among them, bimodal distribution refers to a statistical distribution pattern in which two obvious peaks appear in the frequency histogram of the data distribution.
[0076] S405 calculates the historical data boundary threshold based on single-hop statistical features and double-hop statistical features.
[0077] Specifically, the historical boundary thresholds corresponding to each fluctuation feature are calculated as follows: in, Historical reference thresholds representing hang time. This represents the historical reference threshold for weighted displacement. This represents the historical reference threshold for jitter intensity. This represents the historical adjustment coefficient, with a range of values. .
[0078] In this embodiment of the invention, by combining the statistical clustering analysis of the current sequence with the distribution characteristics of historical labeled data, an adaptive boundary threshold is dynamically generated. This not only utilizes the specificity of the current data to improve the real-time discrimination accuracy, but also introduces historical experience to enhance the robustness and generalization ability of the system when the data is sparse or the distribution is ambiguous.
[0079] S5, weighted fusion of the initial boundary threshold and the historical data boundary threshold to determine the final boundary threshold for each fluctuation feature.
[0080] In one possible implementation, S5 specifically includes sub-steps S501 and S502: S501, determine the fusion weight based on the total number of jump events in the current jump rope sequence and the total number of historical samples.
[0081] Specifically, the formula for calculating the fusion weight is as follows: in, Indicates the fusion weight. This represents the sum of the number of single-hop and double-hop sequences in the historical samples. This represents the total number of jump events in the current jump rope sequence. Represents the natural constant. This represents the attenuation coefficient.
[0082] S502, based on the fusion weight, performs weighted fusion of the initial boundary threshold and the historical data boundary threshold to obtain the final boundary threshold corresponding to each jump feature.
[0083] Specifically, the formula for calculating the final threshold corresponding to each jump feature is as follows: in, The final threshold representing the hang time. This represents the final threshold for weighted displacement. The final threshold representing the intensity of jitter.
[0084] In this embodiment of the invention, by dynamically adjusting the fusion weight based on the current data volume and the historical data volume, a smooth fusion of the current sequence adaptive threshold and the historical experience threshold is achieved. This ensures that the specificity of the current sequence is fully utilized when the data is sufficient, and that historical experience is effectively utilized when the data is sparse. As a result, a stable and reliable boundary threshold can be obtained under different data conditions, thereby improving the overall adaptability and classification accuracy of the system.
[0085] S6, calculate the overall deviation value of the jumping characteristics of each jump compared to the final boundary threshold.
[0086] In one possible implementation, S6 specifically includes sub-steps S601 to S603: S601, calculate the normalized offset of each jitter feature relative to the final boundary threshold.
[0087] The normalized offset refers to the relative distance between the current jump feature value and the final boundary threshold after standardization.
[0088] S602 uses the Sigmoid function to convert each normalized offset into a corresponding deviation value.
[0089] The Sigmoid function is a mathematical function with an "S"-shaped curve that can map any real number to the interval (0,1). It is often used to convert linear outputs into probability values.
[0090] Specifically, the deviation value of each feature in each hop j is calculated as follows: in, This represents the deviation of the hang time height during the j-th jump. This represents the deviation value of the weighted displacement of the j-th jump. This represents the deviation value of the j-th jump jitter intensity. This represents the steepness coefficient of the Sigmoid function, with +0.001 indicating a smoothing term.
[0091] S603, sum and average the deviation values to obtain the comprehensive deviation value.
[0092] Specifically, the calculation method for the overall deviation value is as follows: in, This indicates the overall deviation value.
[0093] In this embodiment of the invention, by normalizing the relative distance between each feature value and the adaptive threshold and converting it into the degree of deviation in probabilistic form, and then fusing it into a single comprehensive index, the quantification and normalization of multi-feature joint discrimination is realized, which significantly improves the decision interpretability, stability and accuracy of single and double jump classification.
[0094] S7. Determine if the overall deviation value is greater than the judgment threshold. If yes, determine the jump as a double jump. Otherwise, determine the jump as a single jump.
[0095] It should be noted that those skilled in the art can set the value of the judgment threshold according to actual needs, and this invention does not limit it.
[0096] In this embodiment of the invention, by comparing the comprehensive deviation value with the judgment threshold in a concise and efficient manner, the clear and rapid classification of single-hop and double-hop actions is achieved. While ensuring that the core judgment logic of the algorithm is clear and reliable, it also leaves room for flexible adjustment of performance in practical applications.
[0097] Specifically, when If the jump is not double, it is considered a double jump. Otherwise, it is considered a single jump.
[0098] In practical application, the key points of the entire jump rope process are first recorded. Based on the vertical positional changes of these key points, the start and end times of each jump, as well as the time to reach the highest point, are determined. After knowing the specific time interval for each jump, features are extracted to determine whether it's a single or double jump. These features include hang time (the vertical distance between the highest and lowest points during each jump, as double jumps are higher than single jumps), weighted displacement (the distance the wrist moves; double jumps involve an extra rope swing, increasing displacement compared to single jumps; weighting is added because a faster wrist shake at the top of the jump completes the double jump, highlighting this wrist movement), and shaking intensity (the wrist's acceleration is faster in double jumps than in single jumps; weighting is the same as the previous feature). Once the features for each jump are available, a method combining the current jump with historical data is used for judgment.
[0099] Furthermore, for jumps where three characteristics exceed the threshold for the current jump (the threshold calculation is an innovative approach), there is a possibility of a double jump, corresponding to a high jump, frequent shaking, and rapid shaking changes. Similarly, for historical data (historical data with similar jump counts, which can be categorized into four types based on the number of excellent, good, passable, and fail grades according to the national physical fitness test standards for sixth-grade boys, and the corresponding historical data set selected based on the current jump count), jumps where three characteristics exceed the threshold for the historical data also suggest a double jump. The threshold used is the weighted sum of both thresholds, with the weights based on the sample size of the historical data; a larger sample size indicates more convincing historical data, and the weights can be appropriately increased. Finally, the deviation of each jump's three characteristics from the threshold is calculated, and the average deviation is used to determine whether the jump is a double or single jump.
[0100] The beneficial effects of this invention are as follows: In this embodiment of the invention, the take-off, highest point, and landing time of each jump are extracted, and multiple refined jumping features are calculated accordingly. These features comprehensively reflect unique movement patterns, including time distribution in the air and speed changes. By combining the current sequence with historical data, the optimal final threshold for different features is dynamically determined, enhancing the system's adaptability to different individuals and jumping styles. By calculating the comprehensive deviation of each jump from the final threshold, minute but crucial feature deviations that conform to the double jump movement pattern can be intelligently identified, achieving high-accuracy double jump judgment. This significantly improves the capabilities of the visual solution in advanced jump rope skill analysis and refined training guidance.
[0101] System Implementation Examples refer to Figure 2The diagram shows a structural schematic of a vision-based double jump rope detection system provided in an embodiment of the present invention.
[0102] The present invention proposes a vision-based double jump rope detection system 30, comprising: a memory 303 and a processor 301.
[0103] The memory 303 stores an application program adapted to be executed by the processor 301 to implement the vision-based double jump detection method for the method embodiment.
[0104] The vision-based double jump rope detection system 30 includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, for example, via a bus 302.
[0105] The structure of the vision-based double jump rope detection system 30 does not constitute a limitation on the embodiments of the present invention.
[0106] Processor 301 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in connection with this disclosure. Processor 301 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.
[0107] Bus 302 may include a pathway for transmitting information between the aforementioned components. Bus 302 may be a PCI bus or an EISA bus, etc. Bus 302 may be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the figure, but this does not mean that there is only one bus or one type of bus.
[0108] The memory 303 may be a ROM or other type of static storage device capable of storing static information and instructions, RAM or other type of dynamic storage device capable of storing information and instructions, or it may be an EEPROM, CD-ROM or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.
[0109] Computer-readable storage medium embodiments The present invention proposes a computer-readable storage medium having a computer program stored thereon, the computer program being loadable and executed by a processor for a vision-based double jump detection method for the first aspect.
[0110] The applicant of this invention has provided a detailed description of the embodiments of the invention in conjunction with the accompanying drawings. However, those skilled in the art should understand that the above embodiments are merely preferred embodiments of the invention. The detailed description is only intended to help readers better understand the spirit of the invention and is not intended to limit the scope of protection of the invention. On the contrary, any improvements or modifications made based on the inventive spirit of the invention should fall within the scope of protection of the invention.
[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the protection scope of the present invention.
Claims
1. A vision-based method for detecting double jumps in a jump rope, characterized in that, The vision-based double jump rope detection method includes: S1, obtain key information of the entire rope skipping process; S2, based on the positional changes of the key points in the vertical direction, determine the start time, highest point time, and landing time for each rope skipping session; S3, Based on the take-off time, the highest point time, and the landing time, extract multiple jumping features during each rope skipping process; S4. Based on all the jumping features of the current jump rope sequence and the pre-stored historical jump rope data, cluster analysis is used to calculate the initial boundary threshold and the historical data boundary threshold. S5, the initial boundary threshold and the historical data boundary threshold are weighted and fused to determine the final boundary threshold of each of the jumping features; S6, calculate the comprehensive deviation value of the jumping feature of each jump relative to the final boundary threshold; S7, determine whether the comprehensive deviation value is greater than the determination threshold; if so, determine that the jump is a double jump; otherwise, determine that the jump is a single jump.
2. The vision-based double jump rope detection method according to claim 1, characterized in that, The key point information specifically includes: left shoulder coordinates, right shoulder coordinates, left elbow coordinates, right elbow coordinates, left hand coordinates, right hand coordinates, left hip coordinates, right hip coordinates, left foot coordinates, and right foot coordinates.
3. The vision-based double jump rope detection method according to claim 1, characterized in that, Specifically, S1 involves using a camera to acquire key point information for each frame of the entire rope-jumping process of the inspector.
4. The vision-based double jump rope detection method according to claim 1, characterized in that, S2 specifically includes: S201, Calculate the jump height of each frame based on the key point information; S202, based on the jump height, determine the jump state of each frame, wherein the jump state includes the high jump state, the low jump state, and the airborne state. S203, extract the time frames of all the jump high point states and construct a jump high point set; S204, based on each of the jump high point time frames in the jump high point set, find the nearest jump low point state time frame before the jump high point time frame and use it as the take-off time; S205, based on the same high jump time frame, find the nearest low jump state time frame after the high jump time frame and use it as the landing time; S206, combine the take-off time, the jump high point time frame, and the landing time to form a jump event; S207, output the time sequence of all the jump events, wherein the time sequence includes the take-off time, the highest point time and the landing time.
5. The vision-based double jump rope detection method according to claim 1, characterized in that, The jumping characteristics specifically include: the hang time height, weighted displacement, and shaking intensity during each jump rope session.
6. The vision-based double jump rope detection method according to claim 1, characterized in that, S4 specifically includes: S401, Calculate the statistics corresponding to all jumping features in the current rope skipping sequence; S402, Based on the historical jump rope data, obtain single jump characteristics and double jump characteristics through manual annotation; S403, Based on the single-hop jumping feature and the double-hop jumping feature, calculate the single-hop statistical feature and the double-hop statistical feature; S404, Based on the statistics, calculate the initial boundary threshold using the K-means clustering algorithm; S405, Calculate the historical data boundary threshold based on the single-hop statistical features and the double-hop statistical features.
7. The vision-based double jump rope detection method according to claim 6, characterized in that, S5 specifically includes: S501, determine the fusion weight based on the total number of jump events in the current jump rope sequence and the total number of historical samples; S502, based on the fusion weight, the initial boundary threshold and the historical data boundary threshold are weighted and fused to obtain the final boundary threshold corresponding to each of the jumping features.
8. The vision-based double jump rope detection method according to claim 1, characterized in that, S6 specifically includes: S601, calculate the normalized offset of each of the jumping features relative to the final boundary threshold; S602, using the Sigmoid function, convert each of the normalized offsets into a corresponding deviation value; S603, sum and average the deviation values to obtain the comprehensive deviation value.
9. A vision-based double jump rope detection system, characterized in that, include: Memory and processor; The memory stores an application program adapted to be executed by the processor to implement the vision-based double jump rope detection method according to any one of claims 1 to 8.
10. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the vision-based double jump rope detection method as described in any one of claims 1 to 8.