A standing long jump distance automatic measuring system based on deep learning
The standing long jump distance automatic measurement system based on deep learning solves the problems of human error and low efficiency in traditional measurement methods, and realizes efficient and accurate automatic measurement. It is adaptable to different environments and supports multimodal data fusion and anomaly detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- LUOYANG NORMAL UNIV
- Filing Date
- 2026-03-20
- Publication Date
- 2026-06-12
AI Technical Summary
Traditional methods for measuring the distance of the standing long jump rely on manual judgment, which is easily affected by subjective factors, leading to large measurement errors and low efficiency. Furthermore, there is a risk of human interference in large-scale testing scenarios.
An automatic measurement system for standing long jump distance based on deep learning is adopted, including modules for data acquisition, keyframe detection, image recognition, and score calculation. It uses a high-speed camera to collect video data, extracts special frame data of the target athlete through the keyframe detection module, accurately confirms the corner points and key points of the jump mat by combining the image recognition module, and performs high-precision calculations by the score calculation module.
It achieves automation and high precision in standing long jump distance measurement, reduces human identification errors, improves measurement efficiency and accuracy, adapts to different lighting environments, supports multimodal data fusion and anomaly detection, and ensures the reliability and stability of measurement results.
Smart Images
Figure FT_1 
Figure FT_2 
Figure QLYQS_1
Abstract
Description
Technical Field
[0001] This application relates to the field of monitoring and analysis technology, and in particular to an automatic measurement system for standing long jump distance based on deep learning. Background Technology
[0002] The standing long jump is a core item in physical education tests and college entrance examinations for primary and secondary schools. It not only accurately assesses students' physical fitness but is also a mandatory test due to its frequent testing and coverage of over 300 million students. A significant characteristic of this sport is its dynamic nature; athletes complete takeoff, flight, and landing in a very short time. Especially during the flight and landing phases, the athlete's body posture and position change drastically, posing a significant challenge to accurate measurement.
[0003] In related technologies, traditional methods for measuring the standing long jump distance typically rely on manual judgment, which is easily influenced by subjective factors, leading to significant measurement errors and controversies. Furthermore, manual measurement is inefficient in large-scale testing scenarios, and the probability of misjudgment increases significantly, posing a high risk of human interference, thus requiring improvement. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this application provides an automatic measurement system for standing long jump distance based on deep learning.
[0005] In a first aspect, this application provides an automatic measurement system for standing long jump distance based on deep learning, comprising:
[0006] The data acquisition module is used to collect long jump video data within a preset timestamp using a high-speed camera;
[0007] The keyframe detection module is used to detect long jump video data within a preset timestamp, and then extract the special frame data corresponding to the target athlete.
[0008] The image recognition module is used to analyze and process the special frame data corresponding to the target athlete, and to confirm the corner data of the jumping mat and the key point position data corresponding to the target athlete based on the analysis and processing results.
[0009] The performance calculation module is used to calculate the athletic performance of the target athlete based on the corner data of the jumping mat and the key point position data corresponding to the target athlete.
[0010] Preferably, the keyframe detection module includes a target detection unit, a jump peak frame detection unit, and a landing frame detection unit;
[0011] The target detection unit is used to identify the entry and exit frames of the target athlete based on the target detection algorithm and binary search method.
[0012] The jump peak frame detection unit is used to make a judgment by adopting a jump frame strategy and combining the minimum value of the vertical coordinate corresponding to the target athlete's heel, and to use a backward filtering mechanism to exclude incomplete human body frames.
[0013] The landing frame detection unit is used to analyze the position change threshold corresponding to the heel through a stability window mechanism.
[0014] Preferably, the keyframe detection module further includes:
[0015] The incomplete human frame filtering mechanism unit is used to detect changes in the height of the bounding box of subsequent frames through the look_ahead_box_size window. When the height of a subsequent frame exceeds 1.5 times the height of the current frame, the current frame is determined to be an incomplete frame.
[0016] The jump phase locking algorithm unit is used to determine the entry frame corresponding to the target athlete using a binary search method. and exit frame This limits the target detection range to ;
[0017] The dynamic threshold adaptive adjustment unit is used to automatically adjust the position change threshold of the heel according to the ambient light intensity, wherein the ambient light intensity is calculated through the luminance component of the HSV color space.
[0018] Preferably, the image recognition module includes the FastNetSeg semantic segmentation model, a jumping mat corner detection algorithm based on convex hull construction and polygon approximation, and a multi-stage heel point recognition algorithm.
[0019] Preferably, the FastNetSeg semantic segmentation model includes a multi-scale feature extraction and fusion unit, a channel mapping optimization unit, and a learning rate decay strategy unit;
[0020] The multi-scale feature extraction and fusion unit is used to perform feature fusion by combining global features of Transformer and local features of lightweight CNN using a multi-branch fusion structure.
[0021] The channel mapping optimization unit is used to employ dynamic channel pruning technology and automatically adjust the number of feature channels according to the complexity of the input image;
[0022] The learning rate decay strategy unit is used to decay the learning rate using a learning rate adjustment strategy, thereby optimizing the stability of training.
[0023] Preferably, the score calculation module includes a perspective transformation matrix optimization algorithm, an edge error compensation model, and a multi-frame verification mechanism;
[0024] The perspective transformation matrix optimization algorithm is used to map the pixel coordinates in the image to the physical plane coordinate system, thereby eliminating image distortion;
[0025] The edge error compensation model is used to compensate for the athletic performance of the target athlete through a preset multinomial regression model.
[0026] The multi-frame verification mechanism is used to perform Kalman filtering on five consecutive frames of measurement results, wherein the filtering gain matrix is adjusted in real time according to the measurement noise.
[0027] Preferred options also include:
[0028] The anomaly detection module is used to perform continuous analysis of the target athlete's movement trajectory to determine whether the target athlete has committed any violations.
[0029] The data security module is used to store hash values of long jump video data using blockchain technology.
[0030] The multimodal fusion interface is used to fuse multimodal data through a data fusion interface and supports multimodal data calibration.
[0031] Secondly, this application provides an automatic measurement method for standing long jump distance based on deep learning, comprising the following steps:
[0032] Long jump video data within a preset timestamp is collected using a high-speed camera;
[0033] The video data of long jump within a preset timestamp is detected, and then the special frame data corresponding to the target athlete is extracted;
[0034] The special frame data corresponding to the target athlete is analyzed and processed, and the corner point data of the jumping mat and the key point position data corresponding to the target athlete are confirmed based on the analysis and processing results.
[0035] The athletic performance of the target athlete is calculated based on the corner data of the jumping mat and the key point location data corresponding to the target athlete.
[0036] Thirdly, this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform any of the above-described deep learning-based automatic measurement system for standing long jump distance.
[0037] In summary, this application includes the following beneficial technical effects:
[0038] This application provides an automatic measurement system for standing long jump distance based on deep learning. It acquires long jump video data with preset timestamps using a high-speed camera, ensuring complete and traceable recording. The timestamps accurately pinpoint the sequence of actions, providing reliable raw material for subsequent analysis. A keyframe detection module filters the video data, extracting special frames of the target athlete to reduce data processing volume, focusing on key moments of action, improving system efficiency, and making subsequent analysis more targeted. An image recognition module deeply analyzes special frames, accurately confirming the corner points of the jumping mat and the athlete's key points, laying the foundation for accurate measurement. Its high-precision recognition capability effectively avoids human error. The performance calculation module calculates the athletic performance, integrating multiple algorithms to ensure accuracy and quickly outputting reliable results, achieving automated measurement. It is widely used in sports testing and training, providing strong support for evaluating athlete performance and improving training effectiveness, thereby effectively improving the efficiency of standing long jump distance measurement. Attached Figure Description
[0039] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the 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.
[0040] Figure 1 This is a schematic diagram of a system for automatic measurement of standing long jump distance based on deep learning, according to an embodiment of this application.
[0041] Figure 2 This is a flowchart of a method for automatically measuring the standing long jump distance based on deep learning, according to an embodiment of this application. Detailed Implementation
[0042] The following is in conjunction with the appendix Figure 1-2 This application will be described in further detail.
[0043] Example 1
[0044] This application discloses an automatic measurement system for standing long jump distance based on deep learning.
[0045] Reference Figure 1 An automatic measurement system for standing long jump distance based on deep learning, comprising:
[0046] The data acquisition module is used to collect long jump video data within a preset time stamp using a high-speed camera;
[0047] The keyframe detection module is used to detect long jump video data within a preset timestamp, and then extract the special frame data corresponding to the target athlete.
[0048] The image recognition module is used to analyze and process the special frame data corresponding to the target athlete, and to confirm the corner data of the jumping mat and the key point position data corresponding to the target athlete based on the analysis and processing results.
[0049] The performance calculation module is used to calculate the athletic performance of the target athlete based on the corner data of the jumping mat and the key point position data corresponding to the target athlete.
[0050] Furthermore, the keyframe detection module includes a target detection unit, a jump peak frame detection unit, and a landing frame detection unit;
[0051] The target detection unit is used to identify the entry and exit frames of the target athlete based on the target detection algorithm and binary search method.
[0052] Specifically, by using target detection algorithms and binary search to identify the entry and exit frames of the target athlete, the system can accurately identify athletes in the video. Binary search provides an efficient search strategy for determining these frames. The combination of these two methods significantly improves detection efficiency. In practical applications, traditional frame-by-frame detection is time-consuming when dealing with large amounts of long jump video data. However, using binary search, the search range can be halved with each search, quickly locating key frames where the athlete appears and disappears, greatly reducing processing time. Simultaneously, this method also improves detection accuracy. The precise identification by the target detection algorithm, combined with the ordered search of the binary search method, effectively avoids misjudgments caused by human judgment or other unstable factors, ensuring more reliable determination of entry and exit frames. Furthermore, this method facilitates subsequent accurate analysis of the athlete's jumping movements, providing strong support for the stable operation and measurement accuracy of the entire standing long jump automatic measurement system.
[0053] The jump peak frame detection unit is used to make a judgment by adopting a jump frame strategy and combining the minimum value of the vertical coordinate corresponding to the target athlete's heel, and to use a backward filtering mechanism to exclude incomplete human body frames.
[0054] Specifically, employing a frame-skipping strategy combined with determining the minimum vertical coordinate of the target athlete's heel, along with a backward filtering mechanism to eliminate incomplete human body frames, brings numerous benefits to keyframe detection in the standing long jump. The frame-skipping strategy significantly improves detection efficiency while maintaining accuracy. When processing high frame rate videos, analyzing frames at fixed intervals reduces data processing volume and computational resource consumption, enabling the system to quickly filter out potential keyframes from numerous frames. Taking a 200FPS video as an example, frame skipping can reduce the number of frames to be processed to a fraction of the original, significantly accelerating processing speed.
[0055] By combining the minimum vertical coordinate of the athlete's heel, the peak jump frame is accurately determined, as this minimum frame often corresponds to the moment the athlete reaches the highest point of the jump, providing a crucial time node for subsequent precise analysis of the jump motion. The backward filtering mechanism effectively eliminates interference from incomplete human body frames. In the initial stage of the jump, the athlete may not be fully within the frame, resulting in a smaller bounding box height and potential for misjudgment. The backward filtering mechanism compares the bounding box heights of subsequent frames to eliminate these incomplete frames, ensuring the accuracy of keyframe detection and preventing misjudgments from affecting the overall measurement system's precision, thus improving the system's stability and reliability.
[0056] The landing frame detection unit is used to analyze the position change threshold corresponding to the heel through a stability window mechanism.
[0057] Specifically, analyzing the threshold of heel position change through a stability window mechanism is of great significance to the standing long jump measurement system. In the standing long jump, the determination of the athlete's landing moment is crucial, and the stability window mechanism provides a reliable basis for this judgment. By observing the changes in heel position across a certain number of consecutive frames, it avoids misjudgments caused by fluctuations in single-frame data.
[0058] Specifically, a suitable stability window is set, and the maximum and minimum values of the heel position within the window are statistically analyzed. When the difference between the two values is less than a preset threshold, the athlete is considered to have landed. This method fully considers the dynamic changes during the athlete's landing process, and compared to judging based on a single frame, it can more accurately capture the landing moment. Moreover, the threshold setting can be adjusted according to actual conditions, enhancing the system's adaptability. It can flexibly determine the landing frame under different field conditions and individual athlete differences, effectively improving the accuracy and reliability of the measurement system, ensuring that the final measured long jump distance truly reflects the athlete's performance, and providing more accurate data support for sports testing and training.
[0059] It should be noted that the keyframe detection module specifically includes:
[0060] The incomplete human frame filtering mechanism unit is used to detect changes in the height of the bounding box of subsequent frames through the look_ahead_box_size window. When the height of a subsequent frame exceeds 1.5 times the height of the current frame, the current frame is determined to be an incomplete frame.
[0061] The jump phase locking algorithm unit is used to determine the entry frame corresponding to the target athlete using a binary search method. and exit frame This limits the target detection range to ;
[0062] The dynamic threshold adaptive adjustment unit is used to automatically adjust the position change threshold of the heel according to the ambient light intensity, wherein the ambient light intensity is calculated through the luminance component of the HSV color space.
[0063] Specifically, the incomplete human body frame filtering mechanism can effectively eliminate interference. In the early stages of an athlete's entry, the body may partially enter the frame, causing abnormal bounding box height. The look_ahead_box_size window is used to detect changes in the bounding box height of subsequent frames. When the height of a subsequent frame exceeds 1.5 times that of the current frame, the current frame is determined to be incomplete, which can avoid misjudging key frames and ensure detection accuracy.
[0064] The jump phase locking algorithm unit uses a binary search method to confirm the entry and exit frames, limiting the target detection range to the actual jump phase and reducing unnecessary computation. For example, when processing large amounts of video data, it can accurately locate the athlete's jump range, improve detection efficiency, allow the system to focus on key areas, and reduce computational resource consumption.
[0065] The dynamic threshold adaptive adjustment unit automatically adjusts the heel position change threshold based on ambient light intensity. Light intensity is calculated using the luminance component of the HSV color space, and the system dynamically adjusts the threshold under different lighting conditions. For example, the threshold is lowered under strong light and raised under weak light, enabling the system to accurately determine the landing frame even in complex lighting environments. This enhances the system's environmental adaptability and ensures the accuracy and reliability of standing long jump measurements.
[0066] Furthermore, the image recognition module includes the FastNetSeg semantic segmentation model, a corner detection algorithm for the jumping mat based on convex hull construction and polygon approximation, and a multi-stage heel point recognition algorithm.
[0067] It should be noted that the FastNetSeg semantic segmentation model includes a multi-scale feature extraction and fusion unit, a channel mapping optimization unit, and a learning rate decay strategy unit;
[0068] The multi-scale feature extraction and fusion unit is used to perform feature fusion by combining global features of Transformer and local features of lightweight CNN using a multi-branch fusion structure.
[0069] The channel mapping optimization unit is used to employ dynamic channel pruning technology and automatically adjust the number of feature channels according to the complexity of the input image;
[0070] The learning rate decay strategy unit is used to decay the learning rate using a learning rate adjustment strategy, thereby optimizing the stability of training.
[0071] Specifically, the various units in the FastNetSeg semantic segmentation model work together to bring significant advantages to the automatic measurement system for the standing long jump. The multi-scale feature extraction and fusion unit adopts a multi-branch fusion structure, combining the powerful global feature capture capability of the Transformer with the local feature extraction advantage of the lightweight CNN. This allows the model to grasp both the overall structure of the image and the detailed features of the jumping mat and the athlete, improving segmentation accuracy, accurately distinguishing the jumping mat from the athlete, and providing a reliable foundation for subsequent accurate calculation of scores.
[0072] The channel mapping optimization unit utilizes dynamic channel pruning technology to automatically adjust the number of feature channels based on the complexity of the input image. It reduces the number of channels when processing simple images to decrease computation, while increasing the number of channels for complex images to extract more features, avoiding wasted computational resources, improving model efficiency, and ensuring the system can quickly process image data in different scenarios.
[0073] The learning rate decay strategy unit employs a reasonable learning rate adjustment strategy for decay. In the early stages of training, the learning rate is relatively large to accelerate the model's convergence speed. As training progresses, the learning rate gradually decreases to prevent the model from oscillating when approaching the optimal solution, thus optimizing training stability. This allows the model parameters to more stably approach the optimal value, improving the model's generalization ability, reducing the risk of overfitting, and ensuring that the model performs well on different datasets.
[0074] Furthermore, the performance calculation module includes a perspective transformation matrix optimization algorithm, an edge error compensation model, and a multi-frame verification mechanism;
[0075] The perspective transformation matrix optimization algorithm is used to map the pixel coordinates in the image to the physical plane coordinate system, thereby eliminating image distortion;
[0076] The edge error compensation model is used to compensate for the athletic performance of the target athlete through a preset multinomial regression model.
[0077] The multi-frame verification mechanism is used to perform Kalman filtering on five consecutive frames of measurement results, wherein the filtering gain matrix is adjusted in real time according to the measurement noise.
[0078] Specifically, the perspective transformation matrix optimization algorithm can accurately map the pixel coordinates in an image to the physical plane coordinate system, effectively eliminating image distortion. In actual measurements, images may be distorted due to factors such as shooting angle and equipment. This algorithm can correct these distortions, making the measurement based on the true physical dimensions and greatly improving the basic accuracy of score calculation.
[0079] The edge error compensation model compensates for athletes' performance using a pre-defined multinomial regression model. During measurement, unavoidable errors may exist at image edges, such as those caused by lighting conditions or image resolution issues. This model can correct these errors based on multinomial relationships trained on a large amount of data, making the final score closer to the athlete's true performance.
[0080] The multi-frame verification mechanism uses Kalman filtering to process the measurement results of five consecutive frames, and the filter gain matrix is adjusted in real time according to the measurement noise. This effectively reduces errors caused by random factors, making the measurement results more stable and reliable. When the measurement noise is high, the dynamically adjusted filter gain matrix can better filter out noise interference, improve the accuracy and stability of the score calculation, and ensure that the measurement results truly reflect the athlete's standing long jump level.
[0081] It should be noted that this also includes:
[0082] The anomaly detection module is used to perform continuous analysis of the target athlete's movement trajectory to determine whether the target athlete has committed any violations.
[0083] The data security module is used to store hash values of long jump video data using blockchain technology.
[0084] The multimodal fusion interface is used to fuse multimodal data through a data fusion interface and supports multimodal data calibration.
[0085] Example 2
[0086] This application also discloses an automatic measurement method for standing long jump distance based on deep learning.
[0087] Reference Figure 2 An automatic measurement method for standing long jump distance based on deep learning includes the following steps:
[0088] Long jump video data within a preset timestamp is collected using a high-speed camera;
[0089] The video data of long jump within a preset timestamp is detected, and then the special frame data corresponding to the target athlete is extracted;
[0090] The special frame data corresponding to the target athlete is analyzed and processed, and the corner point data of the jumping mat and the key point position data corresponding to the target athlete are confirmed based on the analysis and processing results.
[0091] The athletic performance of the target athlete is calculated based on the corner data of the jumping mat and the key point location data corresponding to the target athlete.
[0092] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention, they should all fall within the protection scope of the present invention.
[0093] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0094] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.
Claims
1. An automatic measurement system for standing long jump distance based on deep learning, characterized in that, include: The data acquisition module is used to collect long jump video data within a preset timestamp using a high-speed camera; The keyframe detection module is used to detect long jump video data within a preset timestamp, and then extract the special frame data corresponding to the target athlete. The image recognition module is used to analyze and process the special frame data corresponding to the target athlete, and to confirm the corner data of the jumping mat and the key point position data corresponding to the target athlete based on the analysis and processing results. The performance calculation module is used to calculate the athletic performance of the target athlete based on the corner data of the jumping mat and the key point position data corresponding to the target athlete.
2. The automatic measurement system for standing long jump distance based on deep learning according to claim 1, characterized in that, The keyframe detection module includes a target detection unit, a jump peak frame detection unit, and a landing frame detection unit. The target detection unit is used to identify the entry and exit frames of the target athlete based on the target detection algorithm and binary search method. The jump peak frame detection unit is used to make a judgment by adopting a jump frame strategy and combining the minimum value of the vertical coordinate corresponding to the target athlete's heel, and to use a backward filtering mechanism to exclude incomplete human body frames. The landing frame detection unit is used to analyze the position change threshold corresponding to the heel through a stability window mechanism.
3. The automatic measurement system for standing long jump distance based on deep learning according to claim 2, characterized in that, The keyframe detection module further includes: The incomplete human frame filtering mechanism unit is used to detect changes in the height of the bounding box of subsequent frames through the look_ahead_box_size window. When the height of a subsequent frame exceeds 1.5 times the height of the current frame, the current frame is determined to be an incomplete frame. The jump phase locking algorithm unit is used to determine the entry frame corresponding to the target athlete using a binary search method. and exit frames This limits the target detection range to ; The dynamic threshold adaptive adjustment unit is used to automatically adjust the position change threshold of the heel according to the ambient light intensity, wherein the ambient light intensity is calculated through the luminance component of the HSV color space.
4. The automatic measurement system for standing long jump distance based on deep learning according to claim 1, characterized in that, The image recognition module includes the FastNetSeg semantic segmentation model, a corner detection algorithm for the jumping mat based on convex hull construction and polygon approximation, and a multi-stage heel point recognition algorithm.
5. The automatic measurement system for standing long jump distance based on deep learning according to claim 4, characterized in that, The FastNetSeg semantic segmentation model includes a multi-scale feature extraction and fusion unit, a channel mapping optimization unit, and a learning rate decay strategy unit. The multi-scale feature extraction and fusion unit is used to perform feature fusion by combining global features of Transformer and local features of lightweight CNN using a multi-branch fusion structure. The channel mapping optimization unit is used to employ dynamic channel pruning technology and automatically adjust the number of feature channels according to the complexity of the input image; The learning rate decay strategy unit is used to decay the learning rate using a learning rate adjustment strategy, thereby optimizing the stability of training.
6. The automatic measurement system for standing long jump distance based on deep learning according to claim 1, characterized in that, The performance calculation module includes a perspective transformation matrix optimization algorithm, an edge error compensation model, and a multi-frame verification mechanism. The perspective transformation matrix optimization algorithm is used to map the pixel coordinates in the image to the physical plane coordinate system, thereby eliminating image distortion; The edge error compensation model is used to compensate for the athletic performance of the target athlete through a preset multinomial regression model. The multi-frame verification mechanism is used to perform Kalman filtering on five consecutive frames of measurement results, wherein the filtering gain matrix is adjusted in real time according to the measurement noise.
7. The automatic measurement system for standing long jump distance based on deep learning according to claim 1, characterized in that, Also includes: The anomaly detection module is used to perform continuous analysis of the target athlete's movement trajectory to determine whether the target athlete has committed any violations. The data security module is used to store hash values of long jump video data using blockchain technology. The multimodal fusion interface is used to fuse multimodal data through a data fusion interface and supports multimodal data calibration.
8. A deep learning-based automatic measurement method for standing long jump distance, applied to the deep learning-based automatic measurement system for standing long jump distance described in any one of claims 1-7, characterized in that, Includes the following steps: Long jump video data within a preset timestamp is collected using a high-speed camera; The video data of long jump within a preset timestamp is detected, and then the special frame data corresponding to the target athlete is extracted; The special frame data corresponding to the target athlete is analyzed and processed, and the corner point data of the jumping mat and the key point position data corresponding to the target athlete are confirmed based on the analysis and processing results. The athletic performance of the target athlete is calculated based on the corner data of the jumping mat and the key point location data corresponding to the target athlete.
9. A computer-readable storage medium, characterized in that: The system stores instructions that, when executed on a computer, cause the computer to perform a deep learning-based automatic measurement system for standing long jump distance as described in any one of claims 1 to 7.