Intelligent rope skipping scoring method based on action recognition
By analyzing video data of jump rope participants using high-definition cameras and deep learning technology, the subjectivity and efficiency problems of traditional manual scoring have been solved. This has enabled objective and precise scoring of movements in jump rope competitions, adapting to movements of different difficulty and styles, and improving the fairness and comprehensiveness of the scoring.
Patent Information
- Application Number
- CN202411274894.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-12
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-09-12
AI Technical Summary
Traditional manual scoring methods in jump rope competitions suffer from significant subjective influences, inconsistent scoring, low efficiency, and difficulty in capturing movement details and quality. Existing electronic scoring systems are complex to install and affected by body size differences, making them difficult to promote in large-scale competitions.
By acquiring video data of rope jumpers through high-definition cameras, the system analyzes indicators such as rope jumping speed, posture stability, spatial position, and movement smoothness. Combined with rope motion data, deep learning and computer vision technologies are used for action recognition and scoring. A unified scoring standard is established to identify action types and errors, and the scoring strategy is dynamically adjusted.
It enables objective, precise, and consistent scoring in jump rope competitions, reduces subjective influence, improves scoring efficiency, captures movement details and quality, adapts to movements of different difficulty and style, and provides scientific and flexible scoring standards.
Smart Images

Figure CN118799964B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of data recognition, and in particular to an intelligent rope skipping scoring method based on motion recognition. Background Art
[0002] As a simple and effective sports activity, rope skipping is widely popular in the field of school education and fitness. It not only strengthens the body, but also cultivates students' perseverance and self-confidence. In order to stimulate students' enthusiasm for sports, many schools hold rope skipping competitions regularly. However, with the increase in the number of participants and the expansion of the scale of competitions, the traditional manual scoring method faces many challenges.
[0003] First, manual scoring is easily affected by subjective factors, and it is difficult to ensure the consistency of scoring standards. Differences in scoring standards between different judges may lead to biased scoring results. For example, for complex rope skipping movements, different judges may give widely different scores due to different focus points. Second, manual scoring is slow and difficult to meet the efficiency requirements of large-scale competitions. This not only prolongs the overall duration of the competition, but also affects the experience of participating rope jumpers and spectators.
[0004] In addition, traditional scoring methods have difficulty capturing and quantifying the subtle differences in rope skipping movements. Rope skipping involves multiple aspects such as speed, posture stability, spatial control, and movement smoothness, and it is difficult for human eyes to accurately evaluate these factors in a short period of time. Especially in high-level competitions, the gap between rope jumpers may be very small, requiring a more sophisticated scoring system to distinguish them.
[0005] Although some existing electronic counters can accurately record the number of rope skipping, they cannot evaluate the quality and difficulty of the movements. Some sensor-based scoring systems, although they can provide some objective data, are complex to install and are easily affected by the body shape of the rope jumpers and the position of the equipment, making them difficult to promote in large-scale competitions. Summary of the invention
[0006] Based on this, it is necessary to provide an intelligent rope skipping scoring method based on motion recognition that can accurately score rope skipping in response to the above technical problems.
[0007] An intelligent rope skipping scoring method based on action recognition comprises the following steps:
[0008] The video data of the rope jumper is acquired through a high-definition camera, and the video data is analyzed to obtain a first data set, a second data set and rope motion state data; the first data set includes rope jumping speed information, posture stability index, motion trajectory data and venue parameter information; the second data set includes the vertical motion amplitude and horizontal displacement information of the rope jumper;
[0009] Based on the first data set, equivalent rope skipping indicators are obtained; the equivalent rope skipping indicators include rope skipping speed, posture stability, spatial position and movement fluency;
[0010] Obtain equivalent movement scores based on rope skipping speed, posture stability, spatial position and movement fluency;
[0011] Obtaining a fused action score based on the second data set and the equivalent action score;
[0012] According to the rope motion state data, rope tension data is obtained, and the current action type is determined according to the rope tension data;
[0013] According to the comparison between the posture stability index in the first data set and the expected posture model in the preset database, if the posture stability index meets the preset condition, the judgment result of the rope jumper making a movement error is obtained; the preset condition is one or any combination of the following conditions: posture instability condition, rhythm inconsistency condition or complete error condition;
[0014] According to the current action type, judgment result and preset scoring strategy, the scoring results of the rope jumper at different periods during the rope skipping process are obtained, and the comprehensive rope skipping score is calculated based on the scoring results.
[0015] In one embodiment, the following steps are also included: according to the equivalent action score, the number of jumps of the rope jumper between two non-critical actions is obtained; if the number of jumps is less than a preset threshold within a set time period, the comprehensive rope skipping score is used as the final rope skipping score; if the number of jumps is greater than or equal to the preset threshold within the set time period, the comprehensive rope skipping score is adjusted to obtain the final rope skipping score.
[0016] In one embodiment, the step of obtaining an equivalent action score according to the rope skipping speed, posture stability, spatial position and action fluency includes the steps of:
[0017] A unified scoring function is used to map the rope skipping speed, posture stability, spatial position and movement fluency into the same first scoring system, respectively, to obtain a first mapping score corresponding to the rope skipping speed, a second mapping score corresponding to the posture stability, a third mapping score corresponding to the spatial position and a fourth mapping score corresponding to the movement fluency;
[0018] Obtaining a sum of the first mapping score, the second mapping score, the third mapping score, and the fourth mapping score;
[0019] Obtaining a maximum value and a minimum value among the first mapping score, the second mapping score, the third mapping score, and the fourth mapping score, and obtaining a realization difference factor according to the maximum value and the minimum value;
[0020] Obtaining an average value of the first mapping score, the second mapping score, the third mapping score, and the fourth mapping score, obtaining a standard deviation of the first mapping score, the second mapping score, the third mapping score, and the fourth mapping score, and obtaining a consistency factor according to the average value and the standard deviation;
[0021] Obtain the equivalent action score based on the sum value, realization difference factor and consistency factor.
[0022] In one embodiment, the step of obtaining the fused action score based on the second data set and the equivalent action score further includes the steps of:
[0023] The vertical motion amplitude and the horizontal displacement information are standardized and mapped into the second scoring system to obtain a first standardized score corresponding to the vertical motion amplitude and a second standardized score corresponding to the horizontal displacement information;
[0024] Perform weight processing on the first standardized score and the second standardized score to obtain the action difficulty coefficient;
[0025] The fusion movement score is obtained based on the movement difficulty coefficient and the equivalent movement score.
[0026] In one embodiment, the steps of obtaining rope tension data according to rope motion state data and determining the current action type according to the rope tension data include the steps of:
[0027] Based on historical data, expert analysis and competition level, the first and second key thresholds are set;
[0028] When the rope tension data is less than or equal to the first critical threshold, the current action type is determined to be a non-critical action;
[0029] When the rope tension data is greater than the first critical threshold and less than the second critical threshold, the current action type is determined to be a critical action;
[0030] When the rope tension data is greater than or equal to the second critical threshold, the current action type is determined to be a difficult action.
[0031] In one embodiment, the steps are also included:
[0032] A first buffer zone is established with the first critical threshold as the center and the first error value as the radius;
[0033] A second buffer zone is established with the second critical threshold as the center and the second error value as the radius;
[0034] When the rope tension data is in the first buffer or the second buffer, the current action type is determined based on the duration and context of the action corresponding to the current action type.
[0035] In one embodiment, according to the comparison between the posture stability index in the first data set and the expected posture model in the preset database, if the posture stability index meets the preset conditions, the step of obtaining the judgment result that the rope jumper has made a movement error includes the steps of:
[0036] Determine whether the attitude stability index meets the attitude instability condition based on the following steps:
[0037] The sliding time window technique is used to calculate the standard deviation of the attitude stability index within the preset window;
[0038] The standard deviation of the posture stability index is compared with a preset standard deviation threshold of the expected posture model. If the standard deviation of the posture stability index is greater than the preset standard deviation threshold, the posture stability index satisfies the posture instability condition, otherwise it does not satisfy it.
[0039] Determine whether the posture stability index meets the rhythm inconsistency condition based on the following steps:
[0040] Processing the first data set to obtain the jumping height of each jump of the rope jumper;
[0041] Calculate the time intervals between adjacent jump heights to form a time interval sequence;
[0042] Based on the time interval series, the coefficient of variation is calculated;
[0043] The coefficient of variation is compared with a preset coefficient of variation threshold of the expected posture model. If the coefficient of variation is greater than the preset coefficient of variation threshold, the posture stability index satisfies the rhythm inconsistency condition, otherwise it does not satisfy it;
[0044] Determine whether the attitude stability index meets the complete error condition based on the following steps:
[0045] Processing the first data set to obtain the jumping height of each jump of the rope jumper;
[0046] The jumping height is compared with the preset height threshold of the expected posture model. If the jumping height is less than the preset height threshold, the posture stability index meets the complete error condition, otherwise it does not meet the condition.
[0047] In one embodiment, according to the current action type, the judgment result and the preset scoring strategy, the scoring results of the rope jumper at different time periods during the rope skipping process are obtained, and the comprehensive rope skipping score is calculated according to the scoring results, including the steps of:
[0048] The preset scoring strategies include non-critical action scoring strategy, critical action scoring strategy and high-difficulty action scoring strategy;
[0049] Divide the video data into N video sub-data;
[0050] According to the current action type in the video sub-data, a corresponding scoring strategy is selected from a non-critical action scoring strategy, a critical action scoring strategy, and a difficult action scoring strategy;
[0051] Calculate the score of the video sub-data based on the corresponding scoring strategy and the judgment result in the video sub-data;
[0052] The comprehensive rope skipping score is calculated based on the score results and the duration of the video sub-data.
[0053] In one embodiment, the step of obtaining the number of jumps of the rope jumper between two non-critical actions according to the equivalent action score includes the steps of:
[0054] Set the fusion action scoring threshold and the second action scoring threshold;
[0055] When the equivalent action score rises from below the fusion action score threshold to above the second action score threshold, a positive jump is recorded;
[0056] When the equivalent action score drops from above the second action score threshold to below the fused action score threshold, a reverse jump is recorded;
[0057] Count the total number of positive and reverse transitions within the set time period.
[0058] In one embodiment, if the number of jumps is greater than or equal to a preset threshold within a set time period, the comprehensive rope skipping score is adjusted, and the steps of obtaining the final rope skipping score include the following steps:
[0059] According to the action difficulty coefficient, the score result corresponding to the action difficulty coefficient and the total number of actions in the high dynamic segment, a modified fusion score is obtained; the high dynamic segment is the video sub-data with the number of jumps greater than or equal to the preset threshold; the video data is divided into N video sub-data;
[0060] The score result corresponding to the action difficulty coefficient and the modified fusion score are weighted to obtain an adjusted score;
[0061] The comprehensive rope skipping score is adjusted based on the adjusted score to obtain the final rope skipping score.
[0062] One of the above technical solutions has the following advantages and beneficial effects:
[0063] The present application comprises the following steps: obtaining video data of a rope jumper through a high-definition camera, analyzing the video data, obtaining a first data set, a second data set and rope motion state data; obtaining equivalent rope skipping indicators based on the first data set: obtaining equivalent motion scores based on rope skipping speed, posture stability, spatial position and motion fluency; obtaining fusion motion scores based on the second data set and equivalent motion scores; obtaining rope tension data based on rope motion state data, and judging the current motion type based on the rope tension data; comparing the posture stability index in the first data set with the expected posture model in the preset database, if the posture stability index meets the preset conditions, obtaining the judgment result of the rope jumper's motion error; obtaining the scoring results of the rope jumper at different time periods during the rope skipping process based on the current motion type, the judgment result and the preset scoring strategy, and calculating the comprehensive rope skipping score based on the scoring results. A unified scoring standard is achieved, the subjective factors of traditional manual scoring are avoided, the details of rope skipping movements can be captured, and the rope skipping can be scored more finely based on the quality and difficulty of the rope skipping movements. BRIEF DESCRIPTION OF THE DRAWINGS
[0064] Figure 1 A flow chart of an intelligent rope skipping scoring method based on motion recognition in an embodiment of the present application.
[0065] Figure 2 Another flowchart of the intelligent rope skipping scoring method based on motion recognition in an embodiment of the present application. DETAILED DESCRIPTION
[0066] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0067] In order to achieve the above object, in one embodiment, Figure 1 As shown, an intelligent rope skipping scoring method based on action recognition is provided, comprising the following steps:
[0068] Step S110, acquiring video data of the rope jumper through a high-definition camera, analyzing the video data, and obtaining a first data set, a second data set, and rope motion state data.
[0069] Among them, the first data set includes rope skipping speed information, posture stability index, motion trajectory data and venue parameter information.
[0070] In one example, rope skipping speed information is obtained based on the following steps: using a convolutional neural network (CNN) to preprocess video data and enhance rope and athlete features; using an optical flow estimation network to calculate the optical flow field between consecutive video frames; using an AI-optimized fast Fourier transform (FFT) to identify the main frequency components; and combining a recurrent neural network (RNN) to calculate rope skipping speed information based on the main frequency components.
[0071] The extraction of rope skipping speed information utilizes an AI-enhanced motion detection algorithm. The system first applies a pre-trained CNN to pre-process the video frames to highlight the key features of the rope and the athlete. Then, an optical flow estimation network is used to calculate the precise optical flow field between consecutive video frames, focusing on the movement of the rope passing through the area of the rope jumper's feet. This approach significantly improves the accuracy of motion detection in complex backgrounds.
[0072] The system then applies an optimized FFT to this AI-enhanced optical flow field to identify the main frequency component f. AI-optimized FFT can filter noise more effectively and extract key frequency features. The basic calculation formula for the skipping speed V is V=f*60 (times / minute), where f is the main frequency obtained by AI-assisted FFT analysis, in Hz.
[0073] The system also integrates an RNN model to fine-tune the basic speed calculation results. The RNN takes into account the timing characteristics of the action and provides a more accurate speed estimate. For example, when a dominant frequency of 2Hz is identified, the RNN may adjust the basic estimate of 120 beats per minute to a more accurate value based on the pattern of continuous actions.
[0074] This approach, which combines deep learning and computer vision techniques, not only improves the accuracy of speed estimation, but also enhances the system's ability to adapt to complex rope skipping movements.
[0075] In one example, a posture stability index is obtained based on the following steps: a multi-stage posture estimation network is used to process video data to accurately identify the key points of the body of a rope jumper; a graph neural network (GNN) is applied to analyze the spatial relationship between the key points and calculate motion indicators; and a temporal convolutional network (TCN) is used to process continuous frames to obtain dynamic motion trajectory data.
[0076] Analyzing the posture stability index of rope jumpers uses advanced computer vision technology. The system uses a multi-stage posture estimation network based on deep learning, such as an improved version of HRNet or OpenPose, which can accurately identify and locate 17 to 25 key points of the human body in complex backgrounds and occlusions. These models are trained on large-scale human posture datasets and have strong generalization capabilities.
[0077] Based on the high-quality key point data recognized by AI, the system uses GNN to calculate a series of important indicators. GNN can effectively capture the spatial relationship between various parts of the human body and provide more accurate posture analysis. The torso tilt angle θ is calculated by analyzing the relative positions of the shoulders and hips recognized by AI, reflecting the body stability of the jumper during the rope skipping process. The calculation formula remains unchanged: θ=arccos((y_hip-y_shoulder) / √((x_hip-x_shoulder)²+(y_hip-y_shoulder)²)), where (x_hip, y_hip) and (x_shoulder, y_shoulder) represent the two-dimensional coordinates of the hip and shoulder precisely located by AI, respectively.
[0078] The arm swing amplitude A is quantified by TCN tracking the movement trajectory of the wrist key point recognized by AI in a complete cycle. TCN can effectively process time series data and capture the continuity characteristics of the action. The calculation formula is A=max(√((x_t-x_0)²+(y_t-y_0)²)), where (x_0, y_0) is the wrist position at the beginning of the AI recognition cycle, and (x_t, y_t) is the wrist position at time t.
[0079] The jump height H is calculated by the vertical displacement of the ankle key point tracked by AI, H=max(y_ankle)-min(y_ankle), where y_ankle represents the vertical coordinate of the ankle key point accurately located by AI. The system uses a recurrent neural network (RNN) to process these time series data, improving the accuracy of height estimation.
[0080] These AI-driven indicators are ultimately integrated into a deep learning-based regression model to output a comprehensive posture stability index S. The calculation formula is S=1-(θ / 90°+A / A_ref+H / H_ref) / 3, where A_ref and H_ref are reference values dynamically adjusted by the machine learning algorithm. This AI-optimized comprehensive index fully reflects the rope jumper's technical level and body control ability.
[0081] In one example, motion trajectory data is obtained based on the following steps: a deep learning-driven multi-view 3D posture estimation network is used to process video data to obtain position information of key points of the rope jumper's body changing over time in 3D space; a spatiotemporal graph convolutional network (ST-GCN) is used to analyze the position information to generate accurate motion trajectory data.
[0082] The acquisition of motion trajectory data M makes full use of advanced AI and computer vision technologies. The system deploys a multi-view 3D pose estimation network based on deep learning, such as VoxelPose or VIBE, which can accurately capture the 3D spatial position of key points of the human body in complex environments. These models overcome the depth ambiguity problem of traditional monocular estimation through the collaborative work of multiple cameras and provide highly accurate 3D pose information.
[0083] The key points that the AI system identifies and tracks include but are not limited to the head, neck, shoulders, elbows, wrists, hips, knees, and ankles. By combining convolutional neural networks (CNN) and timing models (such as LSTM), the system can accurately locate the 3D coordinates of these key points in each frame of video, and maintain stable tracking even in high-speed motion and partial occlusion.
[0084] These AI-generated data are stored in the form of time series, represented as M={(x_i(t), y_i(t), z_i(t))|i∈key point set, t∈[0, total video duration]}, where i represents a specific body key point, t represents the timestamp of the video, ranging from 0 to the total video duration, and (x_i(t), y_i(t), z_i(t)) represents the AI-estimated 3D coordinates of key point i at time t. The system uses a high-speed image processing unit (GPU) to accelerate and achieve high-frequency data acquisition of 60 frames per second, ensuring that fast and complex action details are captured.
[0085] In one example, venue parameter information is obtained based on the following steps: processing video data to obtain motion trajectory data of the rope jumper's ankle key points; calculating the rope jumper's maximum movement range on the horizontal plane based on the motion trajectory data of the ankle key points; and obtaining venue reference information based on the maximum movement range.
[0086] After obtaining the venue parameter information, the system tracks the motion trajectory data of the key points of the rope jumper's ankle and calculates the maximum range of movement of the rope jumper in the horizontal plane. The maximum range of movement is calculated as follows: Area = (max(x_ankle)-min(x_ankle))*(max(y_ankle)-min(y_ankle)), where x_ankle and y_ankle represent the horizontal and vertical coordinates of the key points of the ankle, respectively.
[0087] The second data set includes the vertical motion amplitude and horizontal displacement information of the rope jumper.
[0088] When acquiring the second data set, the system deployed a deep learning-enhanced motion analysis module, focusing on the overall performance and difficulty assessment of the rope jumper's movements. For the analysis of vertical motion amplitude, the system uses a human posture estimation model based on a convolutional neural network (CNN) to accurately locate the hip position as an approximation of the body's center of gravity. By continuously tracking the trajectory of the hip in each jump cycle through a recurrent neural network (RNN), the system identifies the highest and lowest points and calculates the vertical amplitude V_amp=max(y_hip)-min(y_hip), where y_hip is the vertical coordinate of the hip identified by AI. The system also applies a temporal convolutional network (TCN) to analyze continuous frame data and calculates the average vertical amplitude V_avg as an indicator of the vertical motion amplitude of the overall performance.
[0089] The system continuously captures the movement trajectory of the rope jumper through AI-enhanced high-precision computer vision technology to analyze horizontal displacement information. The deep learning model accurately locates the hip coordinates (x_hip, y_hip) in each frame, and can maintain stable tracking even in complex backgrounds and high-speed movements. The coordinate system takes the fixed point of the rope skipping area as the origin, and the x-axis and y-axis represent two orthogonal directions of the horizontal plane. The calculation of the total horizontal displacement D_total integrates the motion trajectory analysis driven by the graph neural network (GNN). For two adjacent moments at time t and t-1, the system calculates the change in hip position √((x_hip(t)-x_hip(t-1))²+(y_hip(t)-y_hip(t-1))²). By accumulating all small displacements, the system obtains the total horizontal displacement D_total of the rope jumper during the entire rope skipping process.
[0090] To calculate the average horizontal displacement per jump D_avg, the system deploys a periodic motion detection model based on the long short-term memory network (LSTM) to accurately record the number of jumps of the rope jumper. This AI model analyzes the periodic motion in the vertical direction and accurately identifies the peak point of each jump for counting. The total horizontal displacement D_total is divided by the number of jumps detected by the AI to obtain the average horizontal displacement per jump D_avg (i.e. horizontal displacement information).
[0091] This AI-driven analysis method not only improves the accuracy of data acquisition, but also enhances the system's ability to understand complex action sequences.
[0092] In one example, rope motion state data is obtained based on the following steps: using Hough transform technology to locate the rope in continuous video frames to obtain the rotation radius of the rope; after locating the rope, calculating the instantaneous angular velocity of each position on the rope between continuous video frames; and obtaining the rope motion state data based on the rotation radius and the instantaneous angular velocity.
[0093] The system analyzes the rope motion state data of the rope. First, the system uses the Hough transform technology to locate the rope in the continuous video frames. The advantage of the Hough transform is that it can accurately identify the outline of the rope in a complex background and maintain a stable tracking effect even in high-speed movement and light changes. In this way, the system can not only track the position change of the rope in real time, but also accurately measure its rotation radius r.
[0094] After locating the rope, the system analyzes the change in rope position in consecutive frames. This step uses a high-precision image processing algorithm, and the system calculates the displacement vector of the rope endpoint between adjacent frames. Combined with the camera's frame rate information, through a complex mathematical model, the system can accurately calculate the rope's instantaneous angular velocity ω. This angular velocity data reflects the real-time state of the rope's rotation and is a key parameter for subsequent tension calculations.
[0095] After obtaining the rotation radius r and angular velocity ω, the system applies the formula T=k*(ω²*r) to estimate the rope tension T. This formula is based on the principles of classical mechanics and takes into account the contribution of centrifugal force to rope tension. Among them, k is a comprehensive parameter, which is not only related to the physical properties of the rope (such as mass, length, material elasticity), but also needs to consider the influence of environmental factors (such as air resistance).
[0096] Step S120, based on the first data set, obtaining equivalent rope skipping indicators, wherein the equivalent rope skipping indicators include rope skipping speed, posture stability, spatial position and movement fluency.
[0097] For the skipping speed index, the system uses a standardized formula for processing:
[0098] V_normalized=(V-V_min) / (V_max-V_min)
[0099] Here, V_min and V_max are preset values that are dynamically adjusted according to different competition levels and age groups of rope jumpers. For example, in adult professional competitions, V_min may be set to 120 times / minute and V_max may be set to 240 times / minute; while in the junior youth group, these values may be 60 times / minute and 180 times / minute, respectively. This standardization process ensures that the speed indicators of rope jumpers of different levels are comparable. It is worth noting that this method can not only reflect the basic physical fitness level of rope jumpers, but also their technical proficiency. For example, a professional rope jumper may be able to maintain a stable rhythm when jumping rope at high speeds, while a beginner may become unstable at a relatively low speed.
[0100] The posture stability index directly adopts the posture stability index S calculated in step S110. This index has comprehensively considered the trunk inclination angle, arm swing amplitude and jumping height, and is an indicator that comprehensively reflects the technical level and body control ability of the rope jumper. The S value range is between 0 and 1, and the closer to 1, the more stable the posture. The importance of this indicator lies in that it not only reflects the static balance ability of the rope jumper, but also reflects the dynamic balance ability. For example, a high-level rope jumper can maintain a high S value even when jumping at high speed or performing complex movements, showing excellent body control ability.
[0101] The calculation of the spatial position index is based on the effective jump area (Area), and the spatial control index is calculated using the following formula:
[0102] P=1-(Area / Area_max)
[0103] Area_max is usually set to 1 square meter. This indicator reflects the control ability of the rope jumper in a limited space. The closer the P value is to 1, the stronger the spatial control ability is. The significance of this indicator is to evaluate the ability of the rope jumper to maintain stable rope skipping in a small space, which is particularly important in certain competitions or training scenarios. For example, in a group rope skipping competition, each rope jumper needs to complete the action in a limited space, and a high P value is particularly important at this time.
[0104] The calculation of the movement fluency index uses the rope jumper's motion trajectory data M to evaluate it through Fourier analysis and spectrum entropy calculation. First, the system performs Fourier analysis on the motion trajectory of key points (such as the hip) to obtain the main frequency components. Then, the energy distribution of these frequency components is calculated to obtain the spectrum entropy value E:
[0105] E=-Σ(p_i*log2(p_i))
[0106] Where p_i is the energy proportion of the i-th frequency component. The system normalizes this entropy value to obtain the final action fluency index:
[0107] F=1-(E-E_min) / (E_max-E_min)
[0108] Where E_min and E_max are the preset minimum and maximum entropy values, respectively. The closer the F value is to 1, the smoother the action. This indicator can capture the coherence and rhythm of the action, which is very important when evaluating the beauty and technical level of rope skipping. For example, an experienced rope jumper can maintain the smoothness of the action even when performing complex rope skipping techniques, thus obtaining a high F value.
[0109] Through this detailed distinction, the system is able to provide more accurate and targeted assessments for jumpers of different levels. This not only helps with competition scoring, but also provides valuable feedback to jumpers and coaches, pointing out specific areas that need improvement. For example, if a jumper has high V_normalized and S values, but a low P value, this may mean that the jumper needs to strengthen training in spatial control ability. Similarly, if the F value is low, it may be necessary to focus on improving the coherence and rhythm of the movements.
[0110] Step S130, obtaining an equivalent action score according to the rope skipping speed, posture stability, spatial position and action fluency.
[0111] In one example, the step of obtaining an equivalent action score according to the skipping speed, posture stability, spatial position and movement fluency includes the steps of: using a unified scoring function to map the skipping speed, posture stability, spatial position and movement fluency into the same first scoring system, respectively, to obtain a first mapping score corresponding to the skipping speed, a second mapping score corresponding to the posture stability, a third mapping score corresponding to the spatial position and a fourth mapping score corresponding to the movement fluency; obtaining the sum of the first mapping score, the second mapping score, the third mapping score and the fourth mapping score; obtaining the maximum and minimum values of the first mapping score, the second mapping score, the third mapping score and the fourth mapping score, and obtaining a realization difference factor according to the maximum and minimum values; obtaining the average of the first mapping score, the second mapping score, the third mapping score and the fourth mapping score, obtaining the standard deviation of the first mapping score, the second mapping score, the third mapping score and the fourth mapping score, and obtaining a consistency factor according to the average and the standard deviation; obtaining the equivalent action score according to the sum, the realization difference factor and the consistency factor.
[0112] Specifically, the system uses a unified scoring function to map each equivalent skipping index to a score range of 0 to 10 (the first scoring system). This scoring function uses an improved sigmoid form, expressed as:
[0113] Score=10 / (1+e^(-k*(x-x0)))
[0114] Among them, x is the original indicator value, x0 is the median of the indicator, and k is the parameter that controls the curvature of the function. This function form can provide good discrimination in the middle range, while having a proper saturation effect at the extreme values. By adjusting the parameters k and x0, the system can flexibly adjust the scoring for different indicators and different levels of competition.
[0115] For the skipping speed index (V_normalized), the system sets x0 to 0.5 and k to 10. This means that when the normalized speed reaches a medium level, the score is 5 points. The faster the speed, the closer the score is to 10 points; the slower the speed, the closer the score is to 0 points. This setting reflects the importance of speed in skipping scoring, while also giving skippers the opportunity to excel in other aspects. The speed score is calculated as:
[0116] Speed_Score=10 / (1+e^(-10*(V_normalized-0.5)))
[0117] This formula ensures that even extremely fast speeds will not be scored above 10, while slower speeds will not be scored too low. For example, if a jumper has a V_normalized value of 0.7, indicating an above-average speed, then this formula would give a Speed_Score of about 7.31, reflecting excellent speed performance.
[0118] The x0 of the posture stability index (S) is set to 0.8 and k is 15, which reflects that the posture of high-level rope jumpers is usually relatively stable, and only those with very stable posture can get high scores. The purpose of this setting is to encourage rope jumpers to maintain a good posture while jumping rope quickly. The calculation formula for the posture stability score is:
[0119] Stability_Score=10 / (1+e^(-15*(S-0.8)))
[0120] A larger k value (15) makes the score curve steeper around 0.8, which means that a small improvement in posture stability may lead to a significant score improvement. For example, if a rope jumper's S value is improved from 0.75 to 0.85, his Stability_Score may jump from about 4.98 to about 8.81, reflecting a huge score increase brought by improved posture stability.
[0121] The spatial position index (P) has an x0 of 0.7 and a k of 12, which provides good discrimination near the medium level of spatial control ability. The spatial position score is calculated as:
[0122] Position_Score=10 / (1+e^(-12*(P-0.7)))
[0123] The score of this indicator reflects the ability of the jumper to control the body position in a limited space, which is particularly important for evaluating complex movements and combination skills. For example, if a jumper's P value is 0.8, indicating that his spatial control ability is excellent, then the calculated Position_Score is about 8.32 points, showing that the jumper performs well in this aspect.
[0124] The x0 of the movement fluency index (F) is set to 0.75 and k is 13, which reflects the importance of fluency in high-level rope skipping. Only very smooth movements can get a score close to full marks. The calculation formula for the fluency score is:
[0125] Fluency_Score=10 / (1+e^(-13*(F-0.75)))
[0126] This indicator not only reflects the beauty of the movement, but also indirectly reflects the technical proficiency and physical fitness level of the rope jumper. For example, if a rope jumper's F value is 0.85, which means that his movements are very smooth, then the calculated Fluency_Score is about 9.18 points, reflecting the rope jumper's excellent performance in movement fluency.
[0127] In order to obtain the final equivalent action score BasicScore, the system adopts a multi-factor comprehensive scoring method. This method not only takes into account the basic performance of the rope jumper in all aspects, but also introduces two key factors, performance difference and consistency, to comprehensively evaluate the rope jumper's rope skipping level. The scoring formula is as follows:
[0128] BasicScore=(Speed_Score+Stability_Score+Position_Score+Fluency_Score)*(1+(Max_Score-Min_Score) / 10)*(1+(1-Standard_Deviation / Average_Score) / 5)
[0129] In this formula, Speed_Score, Stability_Score, Position_Score, and Fluency_Score represent the scores of speed, stability, position, and fluency, respectively. Max_Score and Min_Score are the highest and lowest scores of these four scores, respectively. Standard_Deviation is the standard deviation of the four scores, and Average_Score is their average.
[0130] The first part of the formula is a simple sum of four basic scores, which ensures that the scoring system takes into account all the key technical elements of rope skipping. Each score reflects the jumper's performance level in that aspect, and their sum provides a solid foundation for scoring.
[0131] The second part introduces the performance difference factor (1+(Max_Score-Min_Score) / 10). The design concept of this factor is to encourage rope jumpers to pursue excellence in certain aspects, while not overly punishing relatively weak aspects. The difference between the highest score and the lowest score reflects the fluctuation range of the rope jumper's performance. If this difference is large, it means that the rope jumper has performed well in certain aspects, and the system will give appropriate rewards. For example, if the difference between the highest score and the lowest score is 5 points, this factor will be 1.5, increasing the score by 50%. This design helps to discover and cultivate the strengths of rope jumpers, while also encouraging them to challenge themselves and pursue the ultimate in their strengths.
[0132] The third part is the consistency factor (1+(1-Standard_Deviation / Average_Score) / 5). This factor is introduced to reward the balanced development of rope jumpers in all aspects. The coefficient of variation is obtained by dividing the standard deviation by the mean, which reflects the degree of dispersion of the score distribution. When the four scores are very close, the standard deviation will be small, making this factor close to 1.2, providing up to 20% additional points. This design encourages rope jumpers to develop all skills in an all-round way instead of over-relying on a certain strength. Balanced development is not only beneficial to the long-term progress of rope jumpers, but also improves their stability and adaptability in competitions.
[0133] This scoring method cleverly balances multiple scoring goals while keeping the calculations simple. It encourages jumpers to pursue excellence by considering the difference between the highest and lowest scores. At the same time, it also values all-round development by rewarding balanced performance through a consistency factor.
[0134] Step S140: Obtain a fused action score based on the second data set and the equivalent action score.
[0135] In one example, the step of obtaining a fused action score based on a second data set and an equivalent action score also includes the steps of: normalizing the vertical movement amplitude and horizontal displacement information, mapping them into a second scoring system, and obtaining a first standardized score corresponding to the vertical movement amplitude and a second standardized score corresponding to the horizontal displacement information; weighting the first standardized score and the second standardized score to obtain an action difficulty coefficient; and obtaining a fused action score based on the action difficulty coefficient and the equivalent action score.
[0136] First, the system normalizes the vertical motion amplitude:
[0137] V_norm=(V_amp-V_min) / (V_max-V_min)
[0138] Among them, V_amp represents the vertical jump amplitude of the rope jumper during the entire performance, which is calculated in step 110. V_min and V_max are preset reference values, representing the minimum and maximum jump heights expected in the competition at this level, respectively. The setting of these two values needs to consider multiple factors, including the competition level, the age of the rope jumper, and the current level of technical development of rope skipping. When setting these values, a large amount of historical data is usually analyzed to ensure that they can reasonably cover the performance range of the vast majority of rope jumpers.
[0139] The normalized V_norm value ranges from 0 to 1 (secondary scoring system). A value of 0 means that the average jumper's jump height is equal to or lower than V_min, and a value of 1 means that it is equal to or higher than V_max. If the actual performance of the jumper is outside this range, the system will limit it to between 0 and 1 to maintain consistency in scoring.
[0140] For horizontal displacement information, the system uses a similar but slightly different normalization method:
[0141] H_norm=1-(D_avg / D_max)
[0142] Here, D_avg represents the average horizontal displacement per jump of the rope jumper, which is calculated in step 110. D_max is the preset maximum allowable displacement, which represents the maximum horizontal movement distance allowed while maintaining good control. The setting of this value needs to take into account the characteristics of the rope skipping action and the limitations of the competition venue. The reason for using 1 minus the ratio is that in rope skipping, a smaller horizontal displacement generally indicates better control ability.
[0143] The value of H_norm is also between 0 and 1. The closer the value is to 1, the better the horizontal displacement control. If D_avg is greater than D_max, H_norm will be set to 0, indicating insufficient control capability.
[0144] Next, the system calculates the action difficulty coefficient D. This coefficient takes into account both vertical amplitude and horizontal control:
[0145] D=1+α*V_norm+β*H_norm
[0146] Among them, α and β are weight coefficients used to adjust the relative importance of vertical amplitude and horizontal control in difficulty assessment. The setting of these coefficients needs to be determined based on expert opinions and a large amount of competition data analysis. Usually, the value of α will be greater than β, because vertical amplitude often has a greater impact on the difficulty of the action.
[0147] This formula ensures that the value of the difficulty coefficient D is always greater than 1, even if the jumper performs poorly in both aspects (V_norm and H_norm are close to 0), the minimum value of D is 1. The purpose of this design is to ensure that even the most basic movements can be recognized as difficult.
[0148] Finally, the calculation formula for the fusion action score is:
[0149] FusionScore=BasicScore*D
[0150] Here, BasicScore is the equivalent action score calculated in step S130. The advantage of this calculation method is that it can flexibly balance basic skills and difficulty factors. For example, if two jumpers have the same BasicScore, the jumper who challenges a more difficult action (i.e., has a higher D value) will get a higher FusionScore. Conversely, if the difficulty coefficients of the two jumpers are similar, the jumper with a more solid basic technique (i.e., the jumper with a higher BasicScore) will get a higher final score.
[0151] Step S150, obtaining rope tension data according to the rope motion state data, and determining the current action type according to the rope tension data.
[0152] In order to distinguish different types of movements, the system sets two key thresholds, T1 (first key threshold) and T2 (second key threshold). These thresholds are set based on a large amount of historical data and expert analysis, and are dynamically adjusted according to different competition levels. Generally, T1 corresponds to the average tension of basic rope skipping movements, while T2 corresponds to the average tension of difficult movements.
[0153] In one example, the steps of obtaining rope tension data according to rope motion state data, and determining the current action type according to the rope tension data include the following steps: setting a first key threshold and a second key threshold based on historical data, expert analysis and competition level; when the rope tension data is less than or equal to the first key threshold, determining that the current action type is a non-critical action (this usually includes basic actions such as basic single swing jump, slow alternating foot jump and basic back swing jump); when the rope tension data is greater than the first key threshold and less than the second key threshold, determining that the current action type is a critical action (such as cross jump, double swing jump, high leg jump and side swing jump, etc., which require higher skills and coordination); when the rope tension data is greater than or equal to the second key threshold, determining that the current action type is a high-difficulty action (such as triple swing jump, back cross double swing jump, squat jump and inverted rope swing, etc., which require the jumper to have excellent technical level and body control ability).
[0154] In another example, the steps are also included: establishing a first buffer zone with a first critical threshold as the center and a first error value as the radius; establishing a second buffer zone with a second critical threshold as the center and a second error value as the radius; when the rope tension data is in the first buffer zone or the second buffer zone, judging the current action type based on the duration and context of the action corresponding to the current action type.
[0155] In order to handle the situation where the rope tension data fluctuates near the threshold, the system introduces a buffer mechanism. If the tension T fluctuates within the range of ±5% of T1 or T2, the system will consider the duration and context of the action to make a judgment. The design of this buffer can be expressed as:
[0156] Buffer lower limit = T*0.95
[0157] Buffer limit = T*1.05
[0158] For example, if T1 = 100 N, then its buffer range is 95 N to 105 N. When the tension falls within this range, the system analyzes a longer time series to determine the action type.
[0159] In order to improve the accuracy of judgment, the system uses sliding window technology. Usually a 200ms time window is taken, and the mean, standard deviation, peak and valley values of tension are analyzed within this window. These features are used to identify complex action patterns. The mathematical representation of the sliding window is as follows:
[0160] For the time series T={t1,t2,...,tn}, the window size is w, and the data set of the i-th window is:
[0161] Wi={ti,ti+1,...,ti+w-1}
[0162] Within each window, the system calculates the following statistics:
[0163] Mean: μi=(1 / w)*Σ(tj), j from i to i+w-1
[0164] Standard deviation: σi=sqrt((1 / w)*Σ(tj-μi)²), j from i to i+w-1
[0165] Peak value: Peaki=max(tj), j from i to i+w-1
[0166] Valley value: Valleyi=min(tj), j from i to i+w-1
[0167] These statistics help the system recognize different types of motion. For example, a single-swing jump usually shows relatively stable low tension with a small standard deviation. A crossover jump may show periodic moderate fluctuations in tension with a clear peak within each cycle. A double-swing jump may show higher tension values with two close peaks within each cycle.
[0168] The system also analyzes the duration of the action. Basic single jumps usually last 0.5-0.7 seconds each time, while complex combination actions may last 1-2 seconds or longer. The system uses these time features to assist in determining the type of action, especially when distinguishing between continuous basic actions and single complex actions. The analysis of time features can be expressed as:
[0169] Action duration = t_end-t_start
[0170] Where t_start is the timestamp of the start of the action, and t_end is the timestamp of the end of the action. The system will compare this duration with the preset time range to assist in determining the action type.
[0171] The judgment process is continuous, and the system evaluates once every 10ms. In order to avoid misjudgment caused by instantaneous fluctuations, the system uses smoothing technology. It continuously evaluates the judgment results of 5 time points (i.e. 50ms). Only when a certain action type appears 3 or more times in these 5 points, it is confirmed as this type of action. This method can be expressed as:
[0172] For 5 consecutive judgment results {R1, R2, R3, R4, R5}, if:
[0173] count(Ri=X)≥3, where X is a certain action type
[0174] Then confirm that the current action is type X.
[0175] The advantage of this method of determining the type of movement is that by analyzing the changes in rope tension, the system can not only determine the basic type of movement, but also capture the subtle changes and difficulty level of the movement. For example, the system can distinguish the difference between a single swing jump, a double swing jump, and a triple swing jump, which may be visually similar but have obvious differences in tension patterns.
[0176] Step S160, comparing the posture stability index in the first data set with the expected posture model in the preset database, if the posture stability index meets the preset conditions, a judgment result of whether the rope jumper has made a movement error is obtained.
[0177] The preset condition is one or any combination of the following conditions: unstable posture condition, inconsistent rhythm condition or complete error condition.
[0178] In one example, whether the posture stability index satisfies the posture instability condition is determined based on the following steps: using a sliding time window technique, the standard deviation of the posture stability index is calculated within a preset window; the standard deviation of the posture stability index is compared with a preset standard deviation threshold of the expected posture model; if the standard deviation of the posture stability index is greater than the preset standard deviation threshold, the posture stability index satisfies the posture instability condition, otherwise it does not.
[0179] The core of judging the unstable posture condition is to analyze the change of the posture stability index S in step S110. The system adopts the sliding time window technology, usually selecting a time window of 0.5 seconds, and continuously calculates and analyzes the S value in this window. For each time window, the system calculates the standard deviation σ_S of the S value, and the formula is: σ_S=√(Σ(S_i-S_avg)² / N), where S_i represents the posture stability index of each frame in the window, S_avg is the average value of the S value in the window, and N is the total number of frames in the window. Assuming that the frame rate of the system is 60fps, then in a window of 0.5 seconds, N is usually equal to 30. The system compares the calculated σ_S with the preset threshold σ_threshold. If σ_S exceeds σ_threshold, it is determined that the rope jumper has an unstable posture in this time window. The value of σ_threshold needs to be adjusted according to different levels of competitions and the level of rope jumpers. For example, for high-level competitions, σ_threshold may be set to 0.05, while for beginner competitions, it may be set to 0.1 to allow for greater posture fluctuations. In order to improve the accuracy of judgment, the system does not rely solely on the results of a single time window. Instead, it continuously monitors multiple overlapping time windows. Only when multiple consecutive windows (for example, 3) show that the posture is unstable, the system will finally confirm the judgment of unstable posture. This method can effectively reduce misjudgments caused by instantaneous fluctuations.
[0180] In one example, whether the posture stability index satisfies the rhythm inconsistency condition is determined based on the following steps: processing a first data set to obtain the jumping height of each jump of the rope jumper; calculating the time intervals between adjacent jumping heights to form a time interval sequence; calculating the coefficient of variation based on the time interval sequence; comparing the coefficient of variation with a preset coefficient of variation threshold of the expected posture model, if the coefficient of variation is greater than the preset coefficient of variation threshold, the posture stability index satisfies the rhythm inconsistency condition, otherwise it does not.
[0181] The judgment of the rhythm inconsistency condition mainly focuses on the periodic change of the jump height H (calculated in step S110). The system first needs to identify the jump height of each jump, which is usually achieved by finding the local maximum value of the H value. Then, the system calculates the time interval between adjacent jump peaks to form a time interval sequence Δt. For this Δt sequence, the system calculates its coefficient of variation CV: CV=σ_Δt / μ_Δt, where σ_Δt is the standard deviation of the time interval and μ_Δt is the average time interval. The coefficient of variation CV reflects the degree of discreteness of the time interval, and the larger the coefficient, the more inconsistent the rhythm. The system compares the calculated CV with the preset threshold CV_threshold. If CV exceeds CV_threshold, it is determined that the skipping rhythm of the rope jumper is inconsistent. The setting of CV_threshold also needs to consider the competition level and action type. For example, for professional competitions that require extremely high rhythm stability, CV_threshold may be set to 0.1, while for creative competitions that allow more rhythm changes, it may be set to 0.2. In order to adapt to different types of rope skipping actions, the system dynamically adjusts CV_threshold according to the action type identified in step S150. For example, for complex actions that require changing rhythm, the system will appropriately increase CV_threshold to allow for greater rhythm changes.
[0182] In one example, whether the posture stability index meets the complete error condition is determined based on the following steps: processing the first data set to obtain the jumping height of each jump of the rope jumper; comparing the jumping height with a preset height threshold of the expected posture model, if the jumping height is less than the preset height threshold, the posture stability index meets the complete error condition, otherwise it does not meet the condition.
[0183] The judgment of the complete failure condition mainly focuses on the sudden abnormal change of the jump height H. The system continuously monitors the value of H and compares it with the preset minimum effective jump height H_threshold. The judgment criterion is: H_i < H_threshold, where H_i is the jump height of the current frame. The setting of H_threshold needs to consider the characteristics of different types of rope skipping actions. For example, for ordinary single-rotation jumps, H_threshold may be set to 10 cm, while for some low-altitude technical actions, H_threshold may need to be reduced to 5 cm. To avoid misjudging normal low-altitude jumping actions, the system does not rely solely on single-frame data. Instead, it checks the H values of multiple consecutive frames (e.g., 5 frames). Only when the H values of multiple consecutive frames are all lower than H_threshold will the system determine it as a complete failure. The system will also dynamically adjust H_threshold in combination with the action type judgment result of step S150. For example, in the case of identifying a high-difficulty action, the system may slightly reduce H_threshold to allow for more action variations.
[0184] The judgments of these three failure conditions are not independent of each other. The system comprehensively considers these conditions to obtain the final failure judgment result. For example, if within the same time period, the system detects both unstable postures and inconsistent rhythms, then this may be judged as a more serious failure. The system also considers the duration and frequency of the failures. A short-term failure may be regarded as a minor error, while a continuous failure or a high-frequency failure may result in more severe deductions. For example, if more than 3 complete failures are detected within 10 seconds, the system may mark the entire time period as "serious failure". In addition, the system will also analyze the context in which the failure occurs. For example, if the failure occurs in the middle of a high-difficulty action sequence, the system may give a more lenient judgment because this may be a normal fluctuation when the rope skipper is attempting challenging actions. On the contrary, if the failure occurs in a basic action, it may be regarded as a more serious problem.
[0185] Step S170, according to the current action type, judgment result, and preset scoring strategy, obtain the scoring results of different time periods during the rope skipper's skipping process, and calculate the comprehensive rope skipping score based on the scoring results.
[0186] In one example, according to the current action type, the judgment result and the preset scoring strategy, the scoring results of the rope jumper at different time periods during the rope jumping process are obtained, and the steps of calculating the comprehensive rope jumping score according to the scoring results include the following steps: the preset scoring strategy includes a non-critical action scoring strategy, a critical action scoring strategy and a high-difficulty action scoring strategy; the video data is divided into N video sub-data; according to the current action type in the video sub-data, the corresponding scoring strategy is selected from the non-critical action scoring strategy, the critical action scoring strategy and the high-difficulty action scoring strategy; based on the corresponding scoring strategy and the judgment result in the video sub-data, the scoring result of the video sub-data is calculated; according to the scoring result and the duration of the video sub-data, the comprehensive rope jumping score is calculated.
[0187] a) Non-critical action scoring strategy:
[0188] For non-critical actions, the system focuses on two core indicators: rope skipping speed (V_normalized) and posture stability index (S). These two indicators come from the calculation results of step S120. Non-critical actions are usually basic actions in rope skipping performances. Although the technical difficulty is not high, they constitute the basis of the entire performance and reflect the basic technical level of the rope jumper.
[0189] The final score calculation formula for non-critical actions is:
[0190] FinalScore_Non_Critical=w1*Speed_Score+w2*Stability_Score
[0191] Among them, Speed_Score=10*V_normalized, Stability_Score=10*S, w1 and w2 are weight coefficients, usually set to w1=0.6, w2=0.4. This weight distribution reflects that speed is slightly more important than stability in basic movements. The 10 here is a normalization factor used to convert the original range of 0 to 1 indicators into a more intuitive 0 to 10 scale.
[0192] For non-critical movements, even slight errors will be directly reflected in the speed and stability indicators, so there is no need for additional error handling mechanisms. This method simplifies the scoring process while still accurately reflecting the basic technical level of the rope jumper.
[0193] b) Key action scoring strategy:
[0194] Key moves usually have higher technical difficulty and viewing value, so their scoring strategy is more complicated. For key moves, the system distinguishes between mistakes and non-mistakes, because such moves have a significant impact on the overall performance and require more detailed evaluation.
[0195] If no errors are detected, the system uses three indicators, namely, rope skipping speed, posture stability and spatial position, to calculate the score:
[0196] FinalScore_Critical=w1*Speed_Score+w2*Stability_Score+w3*Position_Score
[0197] Among them, w1 = 0.4, w2 = 0.3, w3 = 0.3 can be set. This weight distribution reflects that in key actions, speed, stability and spatial control are all important, but speed still has a slight advantage.
[0198] If an error is detected, the system will introduce the concept of reference score (Reference_Score). Reference_Score refers to the score of the jumper when he successfully completed the same type of action before the current scoring cycle. In the case of an error, the final score is calculated as follows:
[0199] FinalScore_Critical_Fault=α*Reference_Score+(1-α)*FusionScore
[0200] Among them, α is usually set to 0.7, and FusionScore comes from the calculation result of step S140. Reference_Score is a dynamically calculated reference score that reflects the overall performance level of the rope jumper before the current scoring cycle. Specifically, the calculation formula of Reference_Score is:
[0201] Reference_Score=Σ(Wi*Score_i) / Σ(Wi)
[0202] Among them, Score_i is the score of each action of the rope jumper in the previous period of time (usually 2-3 times the duration of the current scoring period), which may be one of the FinalScore_Non_Critical, FinalScore_Critical mentioned above, or FinalScore_High_Difficulty mentioned in the subsequent high-difficulty action scoring, depending on the type of the action. Wi is the corresponding weight, calculated as follows:
[0203] Wi=e^(-λt)*D
[0204] Where: λ is the time decay coefficient, which is usually set to 0.01. t is the number of seconds from the action to the current time. D is the difficulty coefficient of the corresponding action, which comes from the calculation result of step 140.
[0205] The time factor e^(-λt) introduces exponential decay, reflecting the greater impact of more recent actions on current performance. As time t increases, the value of e^(-λt) gradually decreases, causing the weight of earlier actions in calculating Reference_Score to gradually decrease. For example, for an action 10 seconds ago, the time factor is about 0.9048; for an action 100 seconds ago, the time factor drops to 0.3679. This design ensures that Reference_Score can more accurately reflect the recent state and performance trends of the jumper.
[0206] This method of calculating Reference_Score ensures that the system can take into account the recent performance and overall level of the jumper when evaluating the current mistake. Combined with FusionScore, the system can weigh the jumper's courage to challenge difficult movements and overall technical level while considering the current mistake. This method reflects the system's "fault tolerance" mechanism, that is, when evaluating the performance of the jumper, not only the current single performance is considered, but also their overall level. This strategy helps to reduce the impact of accidental factors on the scoring, making the scoring more fair and stable.
[0207] c) Scoring strategy for difficult moves:
[0208] It uses all four equivalent skipping rope indicators: skipping rope speed, posture stability, spatial position and movement fluency. This comprehensive evaluation method reflects the comprehensive requirements of high-difficulty movements on the various aspects of the skipper's ability.
[0209] The final score calculation formula for difficult moves is:
[0210] FinalScore_High_Difficulty=IntermediateScore+Bonus
[0211] The calculation formula of IntermediateScore is:
[0212] IntermediateScore=w1*Speed_Score+w2*Stability_Score+w3*Position_Score+w4*Fluency_Score
[0213] Here, w1=0.3, w2=0.25, w3=0.25, w4=0.2. Fluency_Score is calculated based on the action fluency index F in step 120: Fluency_Score=10*F.
[0214] The calculation formula for the Bonus item is:
[0215] Bonus=Max(Speed_Score,Stability_Score,Position_Score,Fluency_Score)*β
[0216] Among them, β is usually set to 0.2, which indicates the magnitude of the reward.
[0217] For difficult movements, the system does not introduce a separate error handling mechanism. This is because in such complex movements, minor errors are already reflected in the scores of various indicators. Introducing an additional error handling mechanism may lead to excessive punishment for errors. This design also encourages rope jumpers to try more challenging movements, even if there may be small errors, without being overly punished.
[0218] For each evaluated time segment, FinalScore_Non_Critical, FinalScore_Critical, or FinalScore_High_Difficulty is calculated, depending on the action type (non-critical, critical, high difficulty) and whether there are mistakes. For segments marked as "highly dynamic", the adjusted score of step S190 is used. Finally, the entire game time is divided into N consecutive time segments, and the score of each time segment is Si and the duration is Ti. The final score of the entire game is calculated as follows:
[0219] FinalScore=(S1T1+S2T2+...+SN*TN) / (T1+T2+...+TN)
[0220] This comprehensive scoring method not only takes into account the characteristics of different types of movements, but also reflects the overall performance of the entire competition in a time-weighted manner. It encourages rope jumpers to maintain a high level of performance throughout the competition, while also giving more weight to difficult movements and innovative performances. With this method, the system can comprehensively and fairly evaluate the overall performance of rope jumpers, providing a scientific and flexible scoring standard for rope skipping competitions.
[0221] In one embodiment, Figure 2 As shown, the intelligent rope skipping scoring method based on action recognition of the present application also includes the following steps: Step S180, obtaining the number of jumps of the rope jumper between two non-critical actions according to the equivalent action score.
[0222] In one example, the step of obtaining the number of jumps of a rope jumper between two non-critical actions based on the equivalent action score includes the steps of: setting a fusion action score threshold and a second action score threshold; when the equivalent action score rises from below the fusion action score threshold to above the second action score threshold, recording a forward jump; when the equivalent action score drops from above the second action score threshold to below the fusion action score threshold, recording a reverse jump; and counting the total number of forward jumps and reverse jumps within a set time period.
[0223] First, we need to clarify the definition of a jump: a jump refers to a significant switch in the equivalent action score between two preset thresholds. Specifically, when the equivalent action score rises from below the fusion action score threshold to above the second action score threshold, or from above the second action score threshold to below the fusion action score threshold, a jump is considered to have occurred. This definition allows the capture of important change points in the performance of rope jumpers, providing key information for subsequent coherence analysis.
[0224] The purpose of setting up jump judgment is to comprehensively evaluate the performance of rope jumpers. Imagine that rope skipping competition is like a wonderful concert, and jumps are like rapid key changes in music. Single movements (notes) are important, but what is more attractive is the changes and combinations (melodies) between movements. By tracking these "movement variations", the system can identify whether rope jumpers can smoothly perform movements of various difficulties, evaluate their ability to flexibly combine different techniques, and capture those wonderful clips that contain high-frequency skill switching. This method not only evaluates the quality of individual movements, but also encourages rope jumpers to show richer and more creative performances, making rope skipping competitions more exciting.
[0225] In order to capture significant changes in motion quality, the system sets two key thresholds: the fused motion score threshold and the second motion score threshold. The fused motion score threshold is usually set to 80% of the average score of all non-critical motions, while the second motion score threshold is set to 120% of the average score. This setting ensures that only truly significant changes in motion quality are recorded as jumps.
[0226] For example, suppose that in a high-level competition, the average score of non-critical movements is 8 points (out of 10 points). Then, the scoring threshold for the fusion movement is 6.4 points (8*80%), and the scoring threshold for the second movement is 9.6 points (8*120%). This means that only when the performance of the rope jumper improves from below 6.4 points to above 9.6 points, or drops from above 9.6 points to below 6.4 points, will the system consider a meaningful jump to have occurred.
[0227] When the score rises from below the fusion movement score threshold to above the second movement score threshold, the system records a positive transition. This usually indicates that the jumper is transitioning from a relatively simple movement to a more complex or high-quality movement. For example, the jumper may transition from a basic single swing jump to a more complex crossover jump or double swing jump. Conversely, when the score drops from above the second movement score threshold to below the fusion movement score threshold, the system records a negative transition. This may indicate that the jumper is transitioning from a difficult movement to a simple recovery movement, or it may indicate a significant decrease in movement quality, perhaps due to fatigue or error.
[0228] In order to ensure that what is captured is the real action change rather than the instantaneous score fluctuation, the system uses sliding window technology. The window size is set to 1 second, equivalent to 60 frames (assuming the video frame rate is 60fps). Within each window, the system calculates the average score and uses this average for jump judgment. This method effectively smooths short-term score fluctuations, allowing the system to focus on more lasting action changes.
[0229] For example, suppose a rope jumper performs a series of basic movements, and a small mistake in a moment causes the single-frame score to drop sharply, but it quickly returns to normal. If the sliding window technology is not used, the system may mistakenly judge this instantaneous fluctuation as a jump. However, by calculating the average score within 1 second, this short-term fluctuation will be smoothed out, and the system can more accurately reflect the overall performance of the rope jumper.
[0230] The system uses a counter to accumulate the number of transitions. Each time a transition occurs, the counter is incremented by 1. At the same time, the system records the specific time point and type (forward or reverse) of each transition. These data are stored in an array in the format of [time point, transition type], where the transition type is 1 for forward transition and -1 for reverse transition.
[0231] The set time period is 1 / 5 of the duration of the entire action sequence. For example, if a competition requires a jumper to perform for 3 minutes (180 seconds), the set time period is 36 seconds. The system uses a sliding window with a width equal to the set time period, moving across the entire score sequence and counting the number of jumps within each window.
[0232] The jump threshold is set based on the competition level and the expected movement complexity. For professional-level competition, the jump threshold may be set to 6 times in 36 seconds. This means that the system expects to see a high-level jumper show at least 6 significant movement changes in this time. This setting encourages jumpers to show a variety of techniques while maintaining the quality of the movement.
[0233] During the analysis, the system traverses the entire score sequence. For each frame, the system calculates the average score of the previous 60 frames (1 second) including the frame. Then, the system compares this average score with the previous state to determine whether a jump has occurred. If a jump has occurred, the system increases the jump count and records the time point and type of the jump.
[0234] After completing the jump statistics, the system begins to analyze the jump density using a 36-second sliding window. The window slides from the beginning to the end of the scoring sequence, and the number of jumps in the window is counted with each move. If the number of jumps exceeds the set threshold (for example, 6 times) in any 36-second window, the system will mark the jumper's performance as "highly dynamic" and trigger the execution of the next step.
[0235] If no window with the number of jumps exceeding the threshold is found during the entire analysis process, the system will maintain the scoring result of step S170. This indicates that the performance of the rope jumper is relatively stable, or mainly concentrated on a few types of movements. Although this performance may have advantages in stability and consistency, it may lack variation and challenge.
[0236] This method of analyzing the consistency of movements provides a powerful tool for evaluating the overall performance of a rope jumper. It not only considers the quality of individual movements (which has been evaluated in step S170), but also focuses on the transitions between movements and the complexity of the overall choreography. By accurately calculating and analyzing the number of jumps, the system can effectively distinguish between rope jumpers who are good at individual movements and rope jumpers who have truly mastered comprehensive rope jumping skills.
[0237] Step S190, if the number of jumps is less than a preset threshold within the set time period, the comprehensive rope skipping score is used as the final rope skipping score; if the number of jumps is greater than or equal to the preset threshold within the set time period, the comprehensive rope skipping score is adjusted to obtain the final rope skipping score.
[0238] In one example, if the number of jumps is greater than or equal to a preset threshold within a set time period, the comprehensive rope skipping score is adjusted, and the steps of obtaining the final rope skipping score include the following steps: obtaining a modified fusion score based on the action difficulty coefficient, the score result corresponding to the action difficulty coefficient, and the total number of actions in the high dynamic segment; the high dynamic segment is the video sub-data in which the number of jumps is greater than or equal to the preset threshold; dividing the video data into N video sub-data; weighting the score result corresponding to the action difficulty coefficient and the modified fusion score to obtain an adjusted score; adjusting the comprehensive rope skipping score based on the adjusted score to obtain the final rope skipping score.
[0239] The system first detects the number of jumps within a set time period (1 / 5 of the duration of the entire action sequence) in step S180. When the number of jumps exceeds the preset threshold, the system enters the score adjustment stage. This high frequency of jumps usually means that the rope jumper has demonstrated a rich and diverse combination of movements in a short period of time, or there are significant fluctuations in the difficulty and quality of the movements. This complex performance requires a more detailed and comprehensive scoring method to accurately reflect it.
[0240] During the score adjustment phase, the system first recalculates the ModifiedFusionScore, which is an indicator that comprehensively considers the quality and difficulty of all actions in the current high-dynamic segment. The calculation formula for ModifiedFusionScore is:
[0241] ModifiedFusionScore=(Σ(Di*Score_i)) / N
[0242] Wherein, Di is the difficulty coefficient of each action determined in step S140, Score_i is the score result of the corresponding type calculated in step S170 (which may be one of FinalScore_Non_Critical, FinalScore_Critical or FinalScore_High_Difficulty), and N is the total number of actions in the high dynamic segment.
[0243] This calculation method ensures that the ModifiedFusionScore accurately reflects the jumper's overall performance level in high-dynamic segments, taking into account the difficulty of the movements without neglecting the quality of execution. The difficulty coefficient Di ensures that more challenging movements are given higher weights, while Score_i reflects the quality of the jumper's execution of these movements. By multiplying these two factors and then dividing by the total number of movements, a score that fully reflects the jumper's overall performance in high-dynamic segments is obtained.
[0244] Next, calculate the adjusted score AdjustedScore:
[0245] AdjustedScore=α*Score_i+(1-α)*ModifiedFusionScore
[0246] Among them, Score_i is the score of the single action currently being evaluated, α is a weight coefficient used to balance the importance of the current action score and the overall performance, and (1-α) is the supplementary weight of ModifiedFusionScore.
[0247] The value of α is usually between 0.3 and 0.7, and the specific value needs to be determined according to the nature of the competition and the scoring strategy. A higher α value will place more emphasis on the performance of the current action, while a lower α value will consider the overall performance of the entire high-dynamic segment more. For example, if α is set to 0.5, the score of the current action and the performance of the entire high-dynamic segment will be given equal importance.
[0248] This adjustment formula retains the importance of the current action score, ensuring that excellent individual performances are appropriately recognized, while taking into account the performance of the entire high-dynamic segment to avoid a single action overly affecting the final score. By adjusting the α value, you can flexibly balance the importance of immediate performance and overall performance.
[0249] For example, if a rope jumper performs well in the current action (e.g. Score_i=9.5), but the overall performance is slightly below this level (e.g. ModifiedFusionScore=9.0), and α is set to 0.5, then:
[0250] AdjustedScore=0.5*9.5+0.5*9.0=9.25
[0251] This result not only reflects the jumper's excellent performance in the current action, but also takes into account the overall performance to give a balanced score.
[0252] The system will perform such adjustment calculations for each action in a high-dynamic section. This method can capture the outstanding performance of the rope jumper in high-intensity and high-difficulty sections, avoid a single outstanding action or a single mistake from having too much impact on the overall score, and provide a more comprehensive and balanced score that takes into account both immediate performance and the overall level. In addition, this adjustment mechanism can also handle possible mistakes in high-dynamic sections well. Since the ModifiedFusionScore takes into account the performance of the entire section, even if the rope jumper makes a mistake in a certain action resulting in a lower Score_i, the final AdjustedScore will not over-punish the mistake, but will evaluate it in the context of the overall performance.
[0253] It should be noted that AdjustedScore is not the final score of the entire game, but an adjusted score specifically for highly dynamic segments. During the entire scoring process, AdjustedScore will replace the FinalScore_Non_Critical, FinalScore_Critical or FinalScore_High_Difficulty of the segment originally calculated in step S170. This replacement ensures that the complexity and technical difficulty of highly dynamic segments are fully recognized and accurately evaluated.
[0254] When calculating the final score of the entire game, the system divides the game time (video data) into N consecutive time periods (video sub-data). For a segment identified as highly dynamic, the AdjustedScore is used as the score of the segment; for other segments, the corresponding score calculated in step S170 is used. The score of each time period is recorded as Si, and the duration is Ti. Finally, the score of the entire game (FinalScore) is calculated by time-weighted average:
[0255] FinalScore=(S1T1+S2T2+...+SN*TN) / (T1+T2+...+TN)
[0256] The present application comprises the following steps: obtaining video data of a rope jumper through a high-definition camera, analyzing the video data, obtaining a first data set, a second data set and rope motion state data; obtaining equivalent rope skipping indicators based on the first data set: obtaining equivalent motion scores based on rope skipping speed, posture stability, spatial position and motion fluency; obtaining fusion motion scores based on the second data set and equivalent motion scores; obtaining rope tension data based on rope motion state data, and judging the current motion type based on the rope tension data; comparing the posture stability index in the first data set with the expected posture model in the preset database, if the posture stability index meets the preset conditions, obtaining the judgment result of the rope jumper's motion error; obtaining the scoring results of the rope jumper at different time periods during the rope skipping process based on the current motion type, the judgment result and the preset scoring strategy, and calculating the comprehensive rope skipping score based on the scoring results. A unified scoring standard is achieved, the subjective factors of traditional manual scoring are avoided, the details of rope skipping movements can be captured, and a more refined scoring of rope skipping can be performed based on the quality and difficulty of rope skipping movements.
[0257] This application innovatively constructs a multi-dimensional dynamic evaluation system by analyzing the rope jumper's movement trajectory, posture changes, and rope dynamics in real time. The system can not only accurately capture technical details that are difficult to quantify in the past, such as posture stability and spatial control ability, but also dynamically adjust the scoring criteria to adapt to movements of different difficulty and style. This method significantly improves the accuracy and objectivity of the scoring.
[0258] This application introduces an intelligent error recognition mechanism and dynamic performance evaluation method based on multiple thresholds. By analyzing slight changes in jump height, rope tension, and body posture, the system can accurately distinguish between normal technical fluctuations and actual errors. At the same time, the system introduces an action continuity evaluation based on jump analysis, which identifies complex action combinations and style transitions by detecting significant changes in equivalent action scores. This method not only evaluates the quality of a single action, but also captures the ability of a rope jumper to demonstrate a variety of skills in a short period of time. For highly dynamic performances, the system uses a special scoring adjustment mechanism that takes into account historical performance, current execution, and action diversity, to achieve accurate evaluation of high-level and high-difficulty sections. This dynamic evaluation method greatly improves the fairness and comprehensiveness of scoring, while encouraging rope jumpers to demonstrate richer and more challenging action combinations.
[0259] This application proposes a method for evaluating movement coherence based on jump analysis. By detecting significant changes in equivalent movement scores, the system is able to identify complex movement combinations and style transitions. For highly dynamic performances, the system uses an innovative fusion scoring mechanism that comprehensively considers historical performance, current execution, and difficulty level to achieve accurate evaluation of high-level and high-difficulty segments. This approach not only encourages rope jumpers to demonstrate a diverse combination of skills, but also provides a more scientific and flexible tool for judging high-level competitions.
[0260] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.
[0261] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0262] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.
Claims
1. An intelligent rope skipping scoring method based on action recognition, characterized in that: The following steps are involved: Acquire video data of a rope jumper through a high-definition camera, analyze the video data, and obtain a first data set, a second data set, and rope motion state data; the first data set includes rope jumping speed information, posture stability index, motion trajectory data, and venue parameter information; the second data set includes the vertical motion amplitude and horizontal displacement information of the rope jumper; Based on the first data set, obtaining equivalent rope skipping indicators; the equivalent rope skipping indicators include rope skipping speed, posture stability, spatial position and movement fluency; Obtaining an equivalent action score according to the rope skipping speed, the posture stability, the spatial position and the action fluency; Based on the second data set and the equivalent action score, obtaining a fused action score; Acquiring rope tension data according to the rope motion state data, and determining the current action type according to the rope tension data; According to the comparison between the posture stability index in the first data set and the expected posture model in the preset database, if the posture stability index meets the preset condition, the judgment result of the rope jumper making a movement error is obtained; the preset condition is one or any combination of the following conditions: posture instability condition, rhythm inconsistency condition or complete error condition; According to the current action type, the judgment result and the preset scoring strategy, the scoring results of the rope jumper at different time periods during the rope skipping process are obtained, and a comprehensive rope skipping score is calculated according to the scoring results.
2. The intelligent rope skipping scoring method based on motion recognition according to claim 1 is characterized in that: The following steps are also included: According to the equivalent action score, obtaining the number of jumps of the rope jumper between two non-critical actions; If the number of jumps is less than a preset threshold within a set time period, the comprehensive rope skipping score is used as the final rope skipping score; If the number of jumps is greater than or equal to a preset threshold within the set time period, the comprehensive rope skipping score is adjusted to obtain a final rope skipping score.
3. The intelligent rope skipping scoring method based on motion recognition according to claim 1 is characterized in that: The step of obtaining an equivalent action score according to the rope skipping speed, the posture stability, the spatial position and the action fluency comprises the steps of: A unified scoring function is used to map the rope skipping speed, the posture stability, the spatial position and the movement fluency into the same first scoring system, respectively, to obtain a first mapping score corresponding to the rope skipping speed, a second mapping score corresponding to the posture stability, a third mapping score corresponding to the spatial position and a fourth mapping score corresponding to the movement fluency; Obtaining a sum of the first mapping score, the second mapping score, the third mapping score, and the fourth mapping score; Obtaining a maximum value and a minimum value among the first mapping score, the second mapping score, the third mapping score, and the fourth mapping score, and obtaining a realization difference factor according to the maximum value and the minimum value; Obtaining an average value of the first mapping score, the second mapping score, the third mapping score, and the fourth mapping score, obtaining a standard deviation of the first mapping score, the second mapping score, the third mapping score, and the fourth mapping score, and obtaining a consistency factor according to the average value and the standard deviation; The equivalent action score is obtained according to the sum value, the realization difference factor and the consistency factor.
4. The intelligent rope skipping scoring method based on motion recognition according to claim 1 is characterized in that: The step of obtaining a fused action score based on the second data set and the equivalent action score further includes the steps of: The vertical motion amplitude and the horizontal displacement information are standardized and mapped into a second scoring system to obtain a first standardized score corresponding to the vertical motion amplitude and a second standardized score corresponding to the horizontal displacement information; Performing weight processing on the first standardized score and the second standardized score to obtain a movement difficulty coefficient; The fusion action score is obtained according to the action difficulty coefficient and the equivalent action score.
5. The intelligent rope skipping scoring method based on motion recognition according to claim 1 is characterized in that: The step of obtaining rope tension data according to the rope motion state data, and determining the current action type according to the rope tension data comprises the following steps: Based on historical data, expert analysis and competition level, the first and second key thresholds are set; When the rope tension data is less than or equal to a first critical threshold, determining that the current action type is a non-critical action; When the rope tension data is greater than a first critical threshold and less than a second critical threshold, determining that the current action type is a critical action; When the rope tension data is greater than or equal to a second critical threshold, the current action type is determined to be a difficult action.
6. The intelligent rope skipping scoring method based on motion recognition according to claim 5 is characterized in that: Also includes the steps: A first buffer zone is established with the first critical threshold as the center and the first error value as the radius; A second buffer zone is established with the second critical threshold as the center and the second error value as the radius; When the rope tension data is in the first buffer or the second buffer, the current action type is determined based on the duration and context of the action corresponding to the current action type.
7. The intelligent rope skipping scoring method based on motion recognition according to claim 1 is characterized in that: According to the comparison between the posture stability index in the first data set and the expected posture model in the preset database, if the posture stability index meets the preset conditions, the step of obtaining the judgment result that the rope jumper has made a movement error includes the steps of: Determine whether the posture stability index satisfies the posture instability condition based on the following steps: Using a sliding time window technique, the standard deviation of the posture stability index is calculated within a preset window; Compare the standard deviation of the posture stability index with a preset standard deviation threshold of the expected posture model, if the standard deviation of the posture stability index is greater than the preset standard deviation threshold, then the posture stability index satisfies the posture instability condition, otherwise it does not satisfy; Determine whether the posture stability index satisfies the rhythm inconsistency condition based on the following steps: Processing the first data set to obtain the jumping height of each jump of the rope jumper; Calculating the time intervals between adjacent jump heights to form a time interval sequence; Calculating a coefficient of variation based on the time interval sequence; Comparing the coefficient of variation with a preset coefficient of variation threshold of the expected posture model, if the coefficient of variation is greater than the preset coefficient of variation threshold, the posture stability index satisfies the rhythm inconsistency condition, otherwise it does not satisfy; Determine whether the attitude stability index meets the complete failure condition based on the following steps: Processing the first data set to obtain the jumping height of each jump of the rope jumper; The jumping height is compared with a preset height threshold of the expected posture model. If the jumping height is less than the preset height threshold, the posture stability index satisfies a complete error condition, otherwise it does not satisfy the condition.
8. The intelligent rope skipping scoring method based on motion recognition according to claim 1 is characterized in that: The step of obtaining the score results of the rope jumper at different time periods during the rope skipping process according to the current action type, the judgment result and the preset scoring strategy, and calculating the comprehensive rope skipping score according to the score results includes the following steps: The preset scoring strategies include non-critical action scoring strategies, critical action scoring strategies and high-difficulty action scoring strategies; Dividing the video data into N video sub-data; According to the current action type in the video sub-data, selecting a corresponding scoring strategy from the non-critical action scoring strategy, the critical action scoring strategy and the difficult action scoring strategy; Calculating the score result of the video sub-data based on the corresponding scoring strategy and the judgment result in the video sub-data; The comprehensive rope skipping score is calculated according to the score result and the duration of the video sub-data.
9. The intelligent rope skipping scoring method based on motion recognition according to claim 2 is characterized in that: The step of obtaining the number of jumps of the rope jumper between two non-critical actions according to the equivalent action score comprises the steps of: Set the fusion action scoring threshold and the second action scoring threshold; When the equivalent action score rises from lower than the fused action score threshold to higher than the second action score threshold, a positive jump is recorded; When the equivalent action score drops from being higher than the second action score threshold to being lower than the fused action score threshold, a reverse jump is recorded; The total number of the forward jump and the total number of the reverse jump within the set time period are counted.
10. The intelligent rope skipping scoring method based on motion recognition according to claim 2 is characterized in that: If the number of jumps is greater than or equal to a preset threshold value within the set time period, the comprehensive rope skipping score is adjusted, and the step of obtaining the final rope skipping score includes the following steps: Obtaining a modified fusion score according to the action difficulty coefficient, the score result corresponding to the action difficulty coefficient, and the total number of actions in the high dynamic section; The high dynamic segment is the video sub-data with a jump number greater than or equal to a preset threshold; Dividing the video data into N video sub-data; Performing weighted processing on the score result corresponding to the action difficulty coefficient and the modified fusion score to obtain an adjusted score; The comprehensive rope skipping score is adjusted based on the adjusted score to obtain the final rope skipping score.
