An athlete action data processing method and device based on image acquisition and a medium
By using image acquisition and feature vector analysis, the system distinguishes between accidental and non-accidental non-standard movements of athletes, solving the problem of lack of targeted feedback in existing technologies and achieving more accurate movement guidance.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TIANJIN JIANKUN RUNDE SPORTING GOODS CO LTD
- Filing Date
- 2025-11-21
- Publication Date
- 2026-06-05
AI Technical Summary
Existing technologies lack the ability to further discern the causes of abnormalities when detecting athletes' non-standard movements. They cannot distinguish between occasional deviations and non-occurring non-standard movements, resulting in abnormal feedback lacking specificity and reference value.
Athlete images are acquired through image acquisition, the position feature matrix of joint points is extracted and feature vectors are constructed, and the variance between the feature vectors and the center vector is used to determine whether the action is random. An abnormal prompt is sent when the action is not standard.
It effectively distinguishes between occasional and non-occurring non-standard movements, enhancing the pertinence and reference value of abnormal feedback and assisting athletes in adjusting their movements in a timely manner.
Smart Images
Figure CN121438403B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and in particular to a method, device, and medium for processing athlete motion data based on image acquisition. Background Technology
[0002] Monitoring athletes' movements during exercise to identify non-standard movements is an important means of preventing sports injuries and improving athletic performance. Current technology typically involves monitoring the execution of movements using images of the athlete captured during exercise, assessing the standardization of the movements, and issuing an alert when non-standard movements are detected, prompting the athlete to make timely adjustments.
[0003] However, the above method also has the following technical problems:
[0004] The above method can only trigger an error message when non-standard movements are detected, and lacks the ability to further distinguish the cause of the error; that is, it cannot distinguish whether the non-standard movement is an occasional deviation that occurs by chance, or a non-occasional non-standard movement caused by hidden injuries during the exercise or the athlete's low proficiency in the movement; thus, the error feedback lacks specificity and reference value. Summary of the Invention
[0005] To address the aforementioned technical problems, the technical solution adopted by this invention is as follows:
[0006] According to a first aspect of the present invention, a method for processing athlete motion data based on image acquisition is provided, the method comprising the following steps:
[0007] S1. When an athlete's non-standard movement is detected, the non-standard movement is identified as the first target movement Z, and the time point of occurrence of Z is taken as the target time point T.
[0008] S2. Obtain the time interval B during this exercise, in which the athlete performs the i-th second target action before T. i Corresponding athlete image group A i ; 1≤i≤n, where n is the number of second target actions performed by the athlete before T during this exercise; the action type of the second target action is the same as the action type of the first target action.
[0009] S3, Obtain A i The corresponding eigenvector E i E i By analyzing A i The corresponding position feature matrix D i Feature extraction was performed to obtain; D i Through A i D iLet p be a matrix with p rows and max(m(i)) columns; p is the number of joints of the athlete; m(i) is the number of joints of A. i The number of images of Chinese athletes; D i The positional characteristic of the x-th row and j-th column is C. ijx ; 1≤x≤p; C ijx According to A ij The extracted coordinates of the x-th joint of the athlete; if m(i) < max(m(i)), then D i All positional features in the (m(i)+1)th column are NULL; A ij For B i The j-th athlete image collected in the image; 1≤j≤m(i); the origin of the coordinate system corresponding to each position coordinate is a preset fixed point on the athlete; max() is the maximum value acquisition function.
[0010] S4. If F(H i If F < G, the target action is determined to be an occasional non-standard action, and no exception message is sent; otherwise, an exception message is sent; F() is the variance acquisition function; G is the preset variance; H i For E i With E1, E2, ..., E i , ..., E n The vector distance between the center vectors.
[0011] According to a second aspect of the present invention, a non-transitory computer-readable storage medium is provided, wherein a computer program is stored in the storage medium, and the computer program is loaded and executed by a processor to implement the aforementioned method.
[0012] According to a third aspect of the present invention, an electronic device is provided, comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned method.
[0013] The present invention has at least the following beneficial effects:
[0014] This invention provides a method, device, and medium for processing athlete motion data based on image acquisition. In the method, when an athlete's non-standard movement is detected, the non-standard movement is identified as a first target movement, and the occurrence time of the first target movement is taken as the target time point. A group of athlete images is acquired, corresponding to the time periods during which the athlete performs each second target movement before the target time point. The position coordinates of each joint of the athlete are extracted from the athlete images, and a position feature matrix is constructed based on the position coordinates to obtain the feature vectors corresponding to the athlete image group. If the variance of the vector distance between the feature vector and the central vector of all feature vectors is less than a preset variance, the target movement is determined to be an occasional non-standard movement, and no abnormality prompt is sent; otherwise, an abnormality prompt is sent. It can be seen that this invention, by comparing the current non-standard movement with similar movements previously performed by the athlete during the current movement, and based on the variance of the vector distance between the feature vector and the central vector of all feature vectors, can effectively distinguish between occasional and non-occurring non-standard movements. When the target movement is determined to be an occasional non-standard movement, no abnormality prompt is sent; otherwise, an abnormality prompt is sent, significantly improving the targeting and reference significance of abnormal feedback. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 This is a flowchart illustrating an athlete motion data processing method based on image acquisition, provided as an embodiment of the present invention. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar tasks and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.
[0019] Embodiments of the present invention provide a method for processing athlete motion data based on image acquisition, the method comprising the following steps, such as... Figure 1 As shown:
[0020] S1. When an athlete's non-standard movement is detected, the non-standard movement is identified as the first target movement Z, and the time point of occurrence of Z is taken as the target time point T.
[0021] Specifically, during the athlete's movement, images of the athlete are captured by camera acquisition devices deployed within the athlete's movement area, and the captured images are monitored to determine whether the athlete's movements are standard. The athlete's images are images containing the athlete. Detecting non-standard movements of the athlete can be understood as determining that the athlete's movements are not standard. Those skilled in the art will know that any existing method for monitoring captured athlete images to determine whether the athlete's movements are standard falls within the protection scope of this invention, and will not be elaborated further here.
[0022] Specifically, the time point of occurrence of Z can be understood as the time point at which the athlete begins to execute Z.
[0023] S2. Obtain the time interval B during this exercise, in which the athlete performs the i-th second target action before T. i Corresponding athlete image group A i ; 1≤i≤n, where n is the number of second target actions performed by the athlete before T during this exercise; the action type of the second target action is the same as the action type of the first target action; for example: the first target action is a squat that is judged to be performed incorrectly, and the second target action is a squat that is not judged to be performed incorrectly, where squat is the action type of both the first and second target actions; the first target action is a shot that is judged to be performed incorrectly, and the second target action is a shot that is not judged to be performed incorrectly, where shot is the action type of both the first and second target actions.
[0024] Specifically, B i The starting time point is the time point in this movement when the athlete begins to perform the i-th second target action before T; B i The end time point is the time point in this exercise when the athlete completes the i-th second target action before T.
[0025] Specifically, when i ≠ n, B i Earlier than B i+1 B i+1 B represents the time interval during which the athlete performs the (i+1)th second target action before time T in this exercise. i The end time is earlier than B i+1 The starting time point.
[0026] S3, Obtain A i The corresponding eigenvector E i E i By analyzing A i The corresponding position feature matrix D i Feature extraction was performed to obtain; D i Through A i D i Let p be a matrix with p rows and max(m(i)) columns; p is the number of joints of the athlete; m(i) is the number of joints of A. i The number of images of Chinese athletes; D i The positional characteristic of the x-th row and j-th column is C. ijx ; 1≤x≤p; C ijx According to A ij The extracted coordinates of the x-th joint of the athlete; if m(i) < max(m(i)), then D i All positional features in the (m(i)+1)th column are NULL; A ij For B i The j-th athlete image collected in the image; 1≤j≤m(i); the origin of the coordinate system corresponding to each position coordinate is a preset fixed point on the athlete; max() is the maximum value acquisition function.
[0027] Specifically, a convolutional neural network is used to extract features from the position feature matrix to obtain feature vectors.
[0028] Specifically, max(m(i)) is the maximum value among m(1), m(2), ..., m(i), ..., m(n).
[0029] Specifically, D i The positional characteristic of the x-th row and j-th column can be understood as D. i The element in the x-th row and j-th column of D;i The positional features of the (m(i)+1)th column in D can be understood as D i All elements in the (m(i)+1)th column of the array.
[0030] In a specific implementation, if m(i) < max(m(i)), then D i All positional features in the (m(i)+1)th column are 0.
[0031] Specifically, the preset fixation point on the athlete is a fixation point determined in advance by those skilled in the art from the athlete's body, such as: the center point of the athlete's chest cavity, the center of mass of the athlete in a standing posture, or the center of the athlete's eyebrows, which will not be elaborated here.
[0032] Furthermore, any existing method for extracting the position coordinates of joint points from an image is within the scope of protection of this invention. For example, the OpenPose model is first used to extract the joint point coordinates from the image, and then the joint point coordinates are converted into relative coordinates with a preset fixed point on the athlete as the origin, and the relative coordinates are used as the position coordinates of the joint points; this will not be elaborated further here.
[0033] Specifically, A i Including A i1 A i2 A ij A im(i) When i ≠ 1 and i ≠ m(i), R ij+1 -R ij =R ij -R ij-1 ;R ij+1 For A ij+1 The corresponding data collection time point, R ij For A ij Corresponding data collection time point; R ij-1 For A ij-1 Corresponding data collection time point; A ij+1 For B i The (j+1)th athlete image collected in the image; A ij-1 For B i The j-1th athlete image collected in the process.
[0034] Specifically, A i1 A i2 A ij A im(i) They originated from the same camera acquisition device.
[0035] Specifically, R ij+1 Later than R ij R ij Later than R ij-1 .
[0036] In another embodiment, A ij For B i The sub-athlete image group acquired at the j-th image acquisition time; 1≤j≤m(i), m(i) is B i The number of image acquisition times in A ij This includes every camera capture device deployed within the athletes' activity area in B. i The athlete sub-image acquired at the j-th image acquisition time; B i The time interval between any two adjacent image acquisition times is consistent.
[0037] Specifically, E1, E2, ..., E i , ..., E n All vector dimensions are consistent.
[0038] Specifically, E i With E1, E2, ..., E i , ..., E n The cosine distance between the center vectors of E is taken as E i With E1, E2, ..., E i , ..., E n The vector distance between the center vectors.
[0039] Specifically, E1, E2, ..., E i , ..., E n The average vectors are E1, E2, ..., E i , ..., E n The center vector.
[0040] Furthermore, E1, E2, ..., E i , ..., E n The y-th component of the average vector is E 1y E 2y , ..., E iy , ..., E ny Average value; E iy For E i The y-th component, 1≤y≤q, where q is E i The vector dimension.
[0041] In this embodiment, each row of the position feature matrix corresponds to the position sequence of a joint during movement, reflecting the movement trajectory of that joint; different rows represent the relative positional relationships between the joints. Since all position coordinates are based on a preset fixed point on the athlete's body as the origin of the coordinate system, and this origin moves synchronously with the athlete's movement, the position feature matrix effectively eliminates the influence of global displacement in physical space. This allows the position feature matrix to focus on reflecting the relative movement patterns between the joints, rather than their absolute spatial positions. The feature vectors obtained from the position feature matrix can more accurately represent the movement patterns, improving the ability to capture the essential features of the movement and reducing the additional computational overhead caused by displacement compensation or coordinate alignment, thereby reducing the overall computational load.
[0042] S4. If F(H i If F < G, the target action is determined to be an occasional non-standard action, and no exception message is sent; otherwise, an exception message is sent; F() is the variance acquisition function; G is the preset variance; H i For E i With E1, E2, ..., E i , ..., E n The vector distance between the center vectors.
[0043] Specifically, the preset variance is determined by a person skilled in the art based on H1, H2, ..., H... i H n The pre-set variance is, for example: if n=3, H1, H2, and H3 are 0.1, 0.2, and 0.15 respectively, then the pre-set variance is 0.0025; if n=3, H1, H2, and H3 are 1, 2, and 1.5 respectively, then the pre-set variance is 0.25.
[0044] Specifically, F(H) i ) are H1, H2, ..., H i H n The variance.
[0045] Through the above steps, when an athlete's non-standard movement is detected, the non-standard movement is identified as the first target movement, and the time point of occurrence of the first target movement is taken as the target time point. A group of athlete images is obtained, corresponding to the time periods during which the athlete performs each second target movement before the target time point. The position coordinates of each joint of the athlete are extracted from the athlete images, and a position feature matrix is constructed based on these coordinates to obtain the feature vectors corresponding to the athlete image group. If the variance of the vector distance between the feature vector and the central vector of all feature vectors is less than a preset variance, the target movement is determined to be an occasional non-standard movement, and no abnormality prompt is sent; otherwise, an abnormality prompt is sent. By comparing the current non-standard movement with similar movements previously performed by the athlete during the current movement, and based on the variance of the vector distance between the feature vector and the central vector of all feature vectors, occasional and non-occurring non-standard movements can be effectively distinguished. When the target movement is determined to be an occasional non-standard movement, no abnormality prompt is sent; otherwise, an abnormality prompt is sent, significantly improving the targeting and reference value of the abnormality feedback.
[0046] Specifically, in step S4, when F(H) i When )≥G, if H1, H2, ..., H i H n If the values show a monotonically increasing or continuously rising trend, then the first type of prompt is considered an abnormal prompt; the first type of prompt is used to inform the user that the non-standard target action is caused by a hidden injury during the movement; if H1, H2, ..., H i H n If the trend is not monotonically increasing or continuously rising, the second type of prompt will be considered an abnormal prompt. The second type of prompt is used to remind the user that the non-standard target action is due to low proficiency in the action.
[0047] Specifically, when F(H) i When )≥G, if H1, H2, ..., H i H n If it does not show a monotonically increasing or continuously rising trend, then H1, H2, ..., H i H n It exhibits obvious fluctuation characteristics and has significant volatility.
[0048] In this embodiment, when F(H) i When )≥G, if H1, H2, ..., H i H nWhen the values show a monotonically increasing or continuously rising trend, it indicates that the athlete's movement pattern is gradually deviating from the standard state during the repetition of similar movements. This progressive movement deformation is often due to hidden losses incurred during exercise, such as minor strains. In this case, the first type of indicator is considered an abnormal indicator, which can help athletes detect potential health risks early, intervene in a timely manner, and avoid serious sports injuries. If H1, H2, ..., H i H n If the deviation does not show a monotonically increasing or continuously rising trend, it indicates that there is no obvious cumulative effect of the movement deviation. It is more likely that the athlete is not proficient in the key points of the movement, and the low level of proficiency in the movement leads to the non-standard performance of the first target movement. In this case, the second type of prompt is regarded as an abnormal prompt to help the athlete clarify the training focus. This is conducive to improving the pertinence and reference value of abnormal feedback.
[0049] In one specific embodiment, after step S2 and before step S3, the following is also included:
[0050] If n≥K, proceed to step S3; otherwise, end the process; K is a preset threshold for the number of image groups; the preset threshold for the number of image groups is set by those skilled in the art according to actual needs, for example: 8, 10, 15, which will not be elaborated here.
[0051] In this embodiment, subsequent analysis is only performed when the number of athlete image groups is less than a preset threshold for the number of image groups; otherwise, subsequent analysis is not performed to avoid inaccurate results due to insufficient data.
[0052] Embodiments of the present invention also provide a non-transitory computer-readable storage medium that can be disposed in an electronic device to store a computer program related to implementing a method in the method embodiments, the computer program being loaded and executed by the processor to implement the method provided in the above embodiments.
[0053] Embodiments of the present invention also provide an electronic device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method provided in the above embodiments.
[0054] Embodiments of the present invention also provide a computer program product including program code, which, when the program product is run on an electronic device, causes the electronic device to perform the steps of the methods described above in various exemplary embodiments of the present invention.
[0055] This invention provides a method, device, and medium for processing athlete motion data based on image acquisition. In the method, when an athlete's non-standard movement is detected, the non-standard movement is identified as a first target movement, and the occurrence time of the first target movement is taken as the target time point. A group of athlete images is acquired, corresponding to the time periods during which the athlete performs each second target movement before the target time point. The position coordinates of each joint of the athlete are extracted from the athlete images, and a position feature matrix is constructed based on the position coordinates to obtain the feature vectors corresponding to the athlete image group. If the variance of the vector distance between the feature vector and the central vector of all feature vectors is less than a preset variance, the target movement is determined to be an occasional non-standard movement, and no abnormality prompt is sent; otherwise, an abnormality prompt is sent. It can be seen that this invention, by comparing the current non-standard movement with similar movements previously performed by the athlete during the current movement, and based on the variance of the vector distance between the feature vector and the central vector of all feature vectors, can effectively distinguish between occasional and non-occurring non-standard movements. When the target movement is determined to be an occasional non-standard movement, no abnormality prompt is sent; otherwise, an abnormality prompt is sent, significantly improving the targeting and reference significance of abnormal feedback.
[0056] While specific embodiments of the invention have been described in detail by way of examples, those skilled in the art should understand that the examples are for illustrative purposes only and are not intended to limit the scope of the invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the invention.
Claims
1. A method for processing athlete motion data based on image acquisition, characterized in that, The method includes the following steps: S1. When an athlete's non-standard movement is detected, the non-standard movement is identified as the first target movement Z, and the time point of occurrence of Z is taken as the target time point T. S2. Obtain the time interval B during this exercise, in which the athlete performs the i-th second target action before T. i Corresponding athlete image group A i ; 1≤i≤n, where n is the number of second target actions performed by the athlete before T during this exercise; the action type of the second target action is the same as the action type of the first target action; S3, Obtain A i The corresponding eigenvector E i E i By analyzing A i The corresponding position feature matrix D i Feature extraction was performed to obtain the results; D i Through A i D i Let p be a matrix with p rows and max(m(i)) columns; p is the number of joints of the athlete; m(i) is the number of joints of A. i The number of images of Chinese athletes; D i The positional characteristic of the x-th row and j-th column is C. ijx ; 1≤x≤p; C ijx According to A ij The extracted coordinates of the x-th joint of the athlete; if m(i) < max(m(i)), then D i All positional features in the (m(i)+1)th column are NULL; A ij For B i The j-th athlete image collected in the image; 1≤j≤m(i); the origin of the coordinate system corresponding to each position coordinate is a preset fixed point on the athlete; max() is the maximum value acquisition function; S4. If F(H i If G < G, the target action is determined to be an occasional non-standard action, and no error message is sent. Otherwise, send an error message; F() is the variance acquisition function; G is the preset variance; H i For E i With E1, E2, ..., E i , ..., E n The vector distance between the center vectors.
2. The method for processing athlete motion data based on image acquisition according to claim 1, characterized in that, When i ≠ n, B i Earlier than B i+1 B i+1 Let T be the time interval during which the athlete performs the (i+1)th second target action before T during this exercise.
3. The method for processing athlete motion data based on image acquisition according to claim 1, characterized in that, In step S4, when F(H) i When )≥G, if H1, H2, ..., H i H n If the values show a monotonically increasing or continuously rising trend, then the first type of prompt is considered an abnormal prompt; the first type of prompt is used to inform the user that the non-standard target action is caused by a hidden injury during the movement; if H1, H2, ..., H i H n If the trend is not monotonically increasing or continuously rising, the second type of prompt will be considered an abnormal prompt. The second type of prompt is used to remind the user that the non-standard target action is due to low proficiency in the action.
4. The method for processing athlete motion data based on image acquisition according to claim 1, characterized in that, A i Including A i1 A i2 A ij A im(i) When i ≠ 1 and i ≠ m(i), R ij+1 -R ij =R ij -R ij-1 ;R ij+1 For A ij+1 The corresponding data collection time point, R ij For A ij Corresponding data collection time point; R ij-1 For A ij-1 The corresponding data collection time point.
5. The method for processing athlete motion data based on image acquisition according to claim 1, characterized in that, E1, E2, ..., E i , ..., E n All vector dimensions are consistent.
6. The method for processing athlete motion data based on image acquisition according to claim 1, characterized in that, E i With E1, E2, ..., E i , ..., E n The cosine distance between the center vectors of E is taken as E i With E1, E2, ..., E i , ..., E n The vector distance between the center vectors.
7. The method for processing athlete motion data based on image acquisition according to claim 1, characterized in that, Let E1, E2, ..., E i , ..., E n The average vectors are E1, E2, ..., E i , ..., E n The center vector.
8. The method for processing athlete motion data based on image acquisition according to claim 1, characterized in that, The steps following step S2 and before step S3 include: If n≥K, proceed to step S3; otherwise, end the process; K is a preset threshold for the number of image groups.
9. A non-transitory computer-readable storage medium, characterized in that, The storage medium stores a computer program, which is loaded and executed by a processor to implement the image acquisition-based athlete motion data processing method as described in any one of claims 1-8.
10. An electronic device, comprising: A processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the athlete motion data processing method based on image acquisition as described in any one of claims 1-8.