An intelligent guidance system for basketball skill grading

Abstract

The present invention discloses an intelligent guidance system for basketball skill grading, comprising a sound generator, an acquisition module, a motion analysis module, a learning guidance module and an injury assessment module. The sound generator is used to generate a starting whistle; the acquisition module is used to collect a motion video of a trainee performing a basketball skill grading training process; the motion analysis module is used to identify the trainee's basketball skill movements in the motion video, compare the movements with standard movements, and analyze the correctness of the basketball skill movements; the learning guidance module is used to identify the trainee's sports performance in the motion video, and provide learning guidance opinions in combination with expert knowledge and experience; the injury assessment module is used to identify the trainee's motion pattern in the motion video, and provide injury assessment results in combination with expert knowledge and experience; the system receives a working signal from the sound generator, controls other modules to start working, and generates an intelligent learning guidance report based on the motion recognition results.

Description

Technical Field

[0001] The present invention relates to the technical field of sports training, and in particular to an intelligent guidance system for basketball skill grading. Background Art

[0002] Basketball has a solid mass base and a mature competition system in my country. During its century-long development, it has also maintained a tradition of technological support. It is a unique application field for artificial intelligence. It uses AI's visual perception technology to capture images on the basketball court, realizes intelligent evaluation through robot deep learning, and realizes analysis of athletes' movements and behaviors, which is gradually becoming popular.

[0003] According to the "China Basketball Development Report", basketball is one of the most popular sports among young people in China. Many local education departments have included basketball as a compulsory subject in the physical education entrance examination for middle school students. Therefore, basketball skill training is an essential sports requirement for young people to pass the examination. However, since most people do not have a good understanding of their own physical conditions, it is easy for their fitness plans to be unsuitable, which greatly affects the effectiveness of sports training. Therefore, it is necessary to evaluate sports training.

[0004] Existing sports training assessments mostly adopt a model where the coach observes and guides on the spot. Human eye recognition is limited and may result in omissions, or cameras are used to collect training videos, and then the coach observes and guides after the training. Most people can only rely on their own experience and understanding, which makes the training efficiency extremely low and the movements non-standard. Intelligent assessment systems for the trainees' movement conditions during training mostly use three-dimensional data as the calculation basis. The data volume is huge, the calculation speed is relatively slow, and the real-time performance is poor. The collection of training movements is too broad and poorly targeted. At the same time, it only evaluates the training movements, without evaluating the trainees' athletic ability and potential injuries. It cannot help the trainees better understand their own athletic ability, thus affecting the training effect. Summary of the Invention

[0005] The present invention provides an intelligent coaching system for basketball skill grading, which solves the technical problems that the existing coaching mode mostly adopts on-the-spot observation and guidance by coaches, which is inefficient, and the intelligent monitoring device has poor real-time performance, and the training movement collection is too broad and poorly targeted. It only judges the correctness of the training movements and cannot provide training guidance or opinions.

[0006] The present invention can be achieved through the following technical solutions:

[0007] The beneficial technical effects of the present invention are:

[0008] 1. The sound generator's working signal is used as the starting signal for the entire system, providing the acquisition starting point for the acquisition module to avoid data loss and improve the energy utilization of the entire system. At the same time, the starting whistle comes from the sound generator rather than the coach. Even if the coach is not present, the trainee can still conduct self-training, which is very flexible.

[0009] 2. The motion analysis module analyzes and identifies the correctness of basic technical movements during training to meet the basic requirements of training guidance. It also identifies and analyzes movement parameters such as starting speed, displacement speed, running speed, and jump height during training, thereby improving the trainees' athletic ability and physical fitness and promoting better training results. At the same time, the injury assessment module can identify the hand reach, landing method, lower limb force line, and splayed feet during training to assess potential injuries, improve the safety performance of the training process, better protect the trainees' bodies, and be more practical and more conducive to promotion and application.

[0010] 3. The present invention adopts a skeleton joint target tracking algorithm to realize target tracking of the trainee, and adopts a template matching algorithm to realize target tracking of the basketball, and then obtains the movement conditions of the trainee and the basketball respectively, as well as the relative movement conditions between the two, thereby completing the real-time recognition of the trainee's skill movements, sports performance and movement patterns, realizing all-round training guidance, making up for the omissions of human eye recognition, helping the trainee to master the skill movements required for grading faster, improving training efficiency, and at the same time greatly reducing the workload of the coach, with a higher level of intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is the overall circuit connection block diagram of the present invention;

[0012] Figure 2 Schematic diagram of the positions of 25 joint points in the skeleton joint point target tracking algorithm of the present invention;

[0013] Figure 3 This is a schematic diagram of the front neutral line of the present invention. DETAILED DESCRIPTION

[0014] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0015] like Figure 1As shown, the present invention provides an intelligent guidance system for basketball skill grading, including a sound generator, an acquisition module, a motion analysis module, a learning guidance module and an injury assessment module, all of which are connected to a processor. The sound generator is used to generate a starting whistle; the acquisition module is used to collect a motion video of a trainee performing a basketball skill grading training process; the motion analysis module is used to identify the trainee's basketball skill movements in the motion video, compare them with standard movements, and analyze the correctness of the basketball skill movements; the learning guidance module is used to identify the trainee's sports performance in the motion video, and provide learning guidance opinions based on expert knowledge and experience; the injury assessment module is used to identify the trainee's movement pattern in the motion video, and provide injury assessment results based on expert knowledge and experience; the processor receives a working signal from the sound generator, controls the acquisition module, motion analysis module, learning guidance module, and injury assessment module to start working, and generates and displays an intelligent learning guidance report based on the recognition results of the motion analysis module, learning guidance module, and injury assessment module. In this way, the working signal of the sound generator is used as the starting signal of the entire system, providing the collection starting time point for the collection module, avoiding data loss, and improving the energy utilization rate of the entire system. At the same time, the starting whistle comes from the sound generator rather than the coach, so even if the coach is not there, the trainee can still conduct self-training; the intelligent guidance system of the present invention can not only judge the correctness of basketball skill movements, but also evaluate the trainee's sports performance and movement pattern, so as to guide the trainee's overall sports ability and evaluate potential injuries, better protect the trainee's body, further improve the training effect, better practicality, and more conducive to promotion and application.

[0016] The details are as follows:

[0017] In order to ensure real-time performance, the motion analysis module, learning guidance module and injury assessment module of the present invention all use the skeleton joint target tracking algorithm to track the target, obtain the position information of each joint point of the trainee in each frame image, and realize the target tracking of the trainee. Figure 2 As shown, there are 25 joints, including 0 nose, 1 neck, 2 right shoulder, 3 right elbow, 4 right wrist, 5 left shoulder, 6 left elbow, 7 left wrist, 8 hip center, 9 right hip, 10 right knee, 11 right ankle, 12 left hip, 13 left knee, 14 left ankle, 15 right eye, 16 left eye, 17 right ear, 18 left ear, 19 left big toe, 20 left little toe, 21 left heel, 22 right big toe, 23 right little toe, 24 right heel;

[0018] At the same time, a template matching algorithm is used to identify the position information of the center of the basketball in each frame image to achieve target tracking of the basketball; then, based on the changing trends of the position information of each joint point and the position information of the center of the basketball and the relative position relationship between the two, the trainee's basketball skills, sports performance and movement patterns are identified.

[0019] We can build a three-dimensional model of the basketball court and mark the position information of the center line, basket, etc. of the basketball court to provide a position basis for subsequent calculations of the motion analysis module, learning guidance module, and injury assessment module; considering the shape and surface texture characteristics of the basketball itself, multiple template graphics can be evenly spaced on its surface so that the basketball can be identified from basketball images at different angles and its position information can be obtained; considering that the OpenPose algorithm is an open source real-time system for multi-person 2D posture detection, it is an open source library developed by Carnegie Mellon University (CMU) in the United States based on convolutional neural networks and supervised learning and using Caffe as the framework. It can realize accurate recognition of human skeletal joints in conjunction with the COCO dataset, and can detect information of multiple skeletal key points including the body, feet, hands and face. Therefore, the present invention adopts the OpenPose algorithm to perform skeleton joint target tracking of the trainee.

[0020] The acquisition module includes multiple groups of cameras arranged at intervals on the basketball court, all of which are set along the trainee's running trajectory, with at least one group of cameras corresponding to the ball-holding starting position, the dribbling position while moving, and the shooting position while moving. They respectively capture the trainee's movement status from opposite sides, thereby avoiding as much as possible the unclear relative positions of the trainee, basketball, court rim, and court center line in the collected motion images due to body occlusion, which makes subsequent calculation and recognition impossible.

[0021] The sound generator can be pre-recorded with the basketball whistle sound used by the coach, or a basketball-specific electronic whistle, etc. It is controlled by a remote control switch. When training starts, the whistle can be sounded by simply pressing the remote control switch.

[0022] Based on the movement requirements of the grading skills, and to facilitate analysis and calculation of sports videos, we divide the sports videos into three stages: starting with the ball, dribbling on the move, and shooting on the move. Based on the teaching experience of coaches and experts, we select the basic key movements or patterns corresponding to each stage. Taking the right-handed shooting on the move as an example, the types of basketball skills, sports performances, and movement patterns are as follows:

[0023]

[0024] According to the needs of subsequent calculations, we can pre-set the holding distance threshold between the center of the basketball and the trainee's wrist joint when holding the ball, the landing distance threshold between the center of the basketball and the ground when landing, and record the distance between the heel joint and toe joint and the ground in the first few frames of the sports video at the beginning of the ball holding, which is the ground distance threshold, for subsequent ball holding and ball landing calculations.

[0025] For the ball-holding start phase

[0026] The motion analysis module determines whether the ball-holding start state is correct based on the relative positions of the basketball and the trainee's hands, the consistency of the positions of the hands and feet on the same side, and the timing relationship between the first basketball landing and the back foot.

[0027] Since the right hand is directly above the ball when holding the ball, the left hand supports the ball, and the right foot is in front and the left foot is behind, the ball holding distance threshold can be used to determine whether the trainee is holding the ball with both hands. Then, the relative position information of the center of the basketball and the wrist joints of both hands is compared with the standard relative position. At the same time, based on the right wrist joint point holding the ball above, it is determined whether the right ankle joint point on the same side is in front to determine whether the ball holding action is correct.

[0028] Then, based on the timing relationship between the first time the basketball lands and the first time the left heel joint behind the left ankle joint leaves the ground, determine whether the starting action is correct; for the first time the basketball lands, as long as you find the first time the distance between the center of the basketball and the ground is less than the landing distance threshold, you can determine whether the basketball has landed, and record the corresponding time as time T2; for the first time the left heel joint leaves the ground, you can first record the time when the ball starts to be held, that is, the distance between the left heel joint and the ground in the first few frames of the sports video, which is the ground distance threshold, and then record the time when the subsequent distance starts to be greater than the ground distance threshold, which is time T4. By comparing time T2 and time T4, you can determine whether the starting action is correct.

[0029] Considering that starting speed has a significant impact on the test standard and can also reflect the trainee's reaction speed and movement speed, this learning guidance module provides starting training guidance based on the first basketball landing time and the corresponding back foot off-the-ground time, i.e., starting speed;

[0030] The detection time when the sound generator starts working is time T1, the detection time when the landing distance between the center of the basketball and the ground is less than the landing distance threshold for the first time is time T2, and the difference between time T1 and time T2 is calculated as the first landing time of the basketball; the detection time when the ground distance between the left heel joint behind the left ankle joint and the ground begins to increase for the first time is time T3, and the difference between time T1 and time T3 is calculated as the first time the back foot leaves the ground, that is, the starting speed, and then compare the first basketball landing time and the first back foot leaving the ground with the average time of big data to give learning guidance. For example, if the starting speed is relatively low, subsequent training tends to develop lower limb explosive power to improve the working efficiency of the action chain and promote the coordination of completing technical movements.

[0031] For the dribbling phase

[0032] The motion analysis module determines whether the dribbling state is correct based on the relative position of the dribbling height and the trainee's head and shoulders, as well as the relative position of the hands and the basketball;

[0033] It can detect whether the center of the basketball is higher than the trainee's nose joint, left shoulder joint, and right shoulder joint during dribbling, so as to judge whether the dribbling height is correct;

[0034] Since the relative distance between the wrist joint point and the ball center must be within the holding distance threshold when dribbling with two hands or switching hands, illegal dribbling can be determined by detecting whether the distances between the left wrist joint point and the right wrist joint point and the ball center are simultaneously within the holding distance threshold; and switching hands can be determined by detecting whether the distances between the left wrist joint point and the right wrist joint point and the ball center are alternately within the holding distance threshold.

[0035] Since the relative positional relationship between the center of the ball and the wrist joint changes when dribbling with a wrist flip, the present invention can detect whether the left wrist joint or the right wrist joint is above the center of the basketball during dribbling to determine whether the dribbling is performed with a wrist flip.

[0036] The learning guidance module provides dribbling guidance based on the changing trends of the trainee's stride length and speed, as well as the changes in the rebound height of the basketball and the distance between landing points.

[0037] Since there is an inseparable relationship between walking speed and stride length, we can detect the distance between the left and right heel joints corresponding to each step during walking, which is the stride length f and duration t. f And the number of steps n from starting with the ball to starting to shoot, grabbing the rebound and dribbling to the center line, obtain the change in stride length and speed during movement, that is, the change in displacement speed. Based on the above-mentioned ground distance threshold, determine whether the distance between the left and right heel joints and the ground is within the ground distance threshold, determine whether the left and right feet have landed, and then find the corresponding position information and time information of each adjacent left and right foot landing to obtain the corresponding stride length f and duration t f , and record the number of steps n from starting with the ball to starting to shoot, grabbing the rebound and dribbling to the center line, and then calculate the speed of each step = stride f / time t f , we can get the displacement speed change during the movement. Considering that the displacement speed needs to be reduced when performing technical actions to ensure the stability and accuracy of technical actions, the displacement speed change should be acceleration-deceleration-acceleration-deceleration during the entire running basketball test.

[0038] Then detect the position of the center of the ball each time the basketball bounces to the highest point, that is, the rebound height h, and detect whether the distance between the center of the basketball and the ground is less than the landing distance threshold, then determine that the basketball has landed, and then detect the distance d between each two adjacent basketball landing points to obtain the changes in the basketball movement during movement, and then compare the changes in stride length, displacement speed and basketball movement during movement with the average duration of big data to provide guidance on dribbling ability during movement. For example, if the stride length changes greatly, the corresponding basketball movement changes will also be relatively large, indicating that the trainee lacks ball control ability. Therefore, during training, the difficulty of dribbling can be appropriately increased, and various exercises can be used to improve ball sense and ball control ability, such as transitioning from dribbling in place to dribbling on the move, from normal force dribbling to high-power dribbling, from single ball to double ball, from uniform speed dribbling to variable frequency dribbling.

[0039] For the moving shooting phase

[0040] The motion analysis module detects the relative position of the feet when holding the ball, and the relative position of the basketball and the trainee's shoulders when shooting, to determine whether the shooting state is correct while moving;

[0041] When holding the ball and shooting, first close the ball and step with the right foot at the same time, and then push the ball out with the right hand after three steps. We can determine whether to close the ball by detecting whether the distances between the left wrist joint point and the right wrist joint point and the center of the basketball are both within the holding distance threshold. Then, we can detect whether the right heel joint point is in front of the left heel joint point at this time to determine whether to step with the right foot first, so as to complete the correct action at the beginning of the shot. Then, we can detect in each frame image over a subsequent period of time that the position of the center of the basketball is higher than the left shoulder joint point or the right shoulder joint point, and whether the distances between the left wrist joint point and the right wrist joint point and the center of the basketball are within the holding distance threshold to determine whether the basketball is higher than the shoulder when shooting. If so, it means that the basketball is raised over the shoulder with both hands during the shooting process.

[0042] The learning guidance module provides shooting guidance on the move based on the angle between the trainee's body and the ground when the ball is combined, the jump height and the distance between the take-off point and the landing point, as well as the time from the start of the ball combination to the ball leaving the hand (running speed) and the time from the ball leaving the hand to the rebound (rebounding speed).

[0043] When korfballing, the trainee's body should be upright on the ground. The angle between the front neutral line and the ground can be used to determine whether the body is upright. A tolerance space can be set, such as 90°+ / -5°, where Figure 3 As shown, the frontal neutral line can be obtained by fitting the nose, neck, hip center, the middle point of the right eye and the left eye, the middle point of the right ear and the left ear, the middle point of the right shoulder and the left shoulder, the middle point of the right hip and the left hip, the middle point of the right knee and the left knee, and the middle point of the right ankle and the left ankle.

[0044] The position information S1 of the heel joint at the beginning of the jump, the maximum height h of the heel joint from the ground after the jump m And the position information S2 of the heel joint after landing, and calculate the distance between the position information S1 and S2, that is, the landing distance. Under ideal conditions, the landing distance is zero, of course, deviations are also allowed, so a landing distance threshold can be set;

[0045] Since when shooting, the right hand pushes the ball hard to force it to leave the basket and enter the basket, the speed of running the basket can be detected from the moment of ball contact to the moment when the distance between the right wrist joint and the center of the basketball begins to exceed the ball holding distance threshold. At the same time, the speed of grabbing the rebound can be detected from the moment of ball contact to the moment when both hands touch the ball. The judgment method of both hands touching the ball and the ball contact is the same. Then the angle information and the maximum height h are combined. m , take-off and landing distance, running speed, rebounding speed and the average time of big data are compared to give shooting guidance during movement. According to the take-off height, that is, the maximum height h m , you can give tips on lower limb explosive power training. If the height is not enough, you need to improve the explosive power of the lower limbs, such as the proficiency of single-leg jumping technique and the basic strength of single foot; according to the take-off and landing distance, you can give suggestions on jumping technique and core strength training of the body.

[0046] The injury assessment module determines whether the trainee has bowlegs based on the speed and toe direction of the trainee's movement. It also determines whether the landing method is correct based on the order in which the forefoot and heel of the landing foot land after the shooting jump, thereby reducing the trainee's knee and spinal injuries. It also determines whether the trainee reaches for the ball based on the direction of the basketball's fall and the movement of the trainee's hands after the shot to avoid finger contusions. It also determines the possible injury risk level based on the deviation of the lower limb force line when holding the ball and shooting.

[0047] Among them, running with a figure-eight gait increases the wear and tear of the ankle joint, and the hip external rotator muscles are tense, which may cause the femoral head to move forward. The adductor muscles partially assume the function of hip flexion, which can easily cause symptoms such as groin pain and low back pain. At the same time, if the trainee suffers from figure-eight feet, the speed direction and the direction of the toes during running are likely to be different, and compared with normal trainees, the deviation will be greater. Therefore, the direction of the toes can be used as the direction of the line connecting the toe joint point of the landing foot and the corresponding ankle joint point during walking, and then the angle between the speed direction of each step and the direction of the toes during walking can be used to judge whether the trainee has figure-eight feet. Of course, the angle comparison can set the angle threshold based on expert experience;

[0048] During running, the body is subjected to an impact force approximately three times its own body weight. This force is even greater when jumping and landing. Without the cushioning of the arch of the foot, the impact force will inevitably be directly transmitted to the knee joints, spine, and even the brain. Therefore, when jumping and landing, it is best to land on the forefoot first to reduce the impact force on the body. After jumping, the distance between the toe joint and the heel joint of the landing foot and the ground can be detected to determine whether it has landed. The landing time can then be recorded to determine which part of the foot landed first and whether the landing method is correct.

[0049] Whether catching the ball or rebounding, if you don't reach out to meet the ball, it is very easy to cause finger contusions. Therefore, by detecting the displacement direction D1 of the basketball center falling from the highest point to the second highest point after the shot, and the displacement direction D2 of the wrist joint rising from the lowest point to the highest point at this time, it can be determined whether the two displacement directions are opposite, so as to determine whether the trainee has executed the action of reaching out to meet the ball;

[0050] During the process of shooting or starting with the ball, the direction of the knee joint is offset from the direction of the toes, indicating that there is a deviation in the force line of the lower limb. The larger the deviation angle, the higher the risk of injury. Under normal circumstances, when starting to shoot with the ball, the trainee is in a half-squat state. At this time, the knee joint and toes should be in the same straight line and perpendicular to the ground. When the force line of the lower limb is offset, the projection points of the knee joint and toes on the ground must not coincide. Therefore, the angle between the line connecting the knee joint point and the toe joint point on the same side when preparing to shoot and starting with the ball can be detected to determine the degree of deviation of the force line of the lower limb.

[0051] In addition, the guidance method for left-handed pitching training while moving is similar to that for the right hand. Just swap the corresponding positions of the right and left hands according to the relevant skill requirements.

[0052] The intelligent guidance system of the present invention can also have an input module and a display module. The input module is used to collect the trainee's personal information, including name, age, class, gender, test level, etc., so as to select the corresponding assessment content in a targeted manner, especially the judgment criteria and evaluation basis in the injury assessment module and the learning guidance module. The display module is used to display the intelligent guidance report, including exercise videos corresponding to incorrect movements, learning guidance opinions, exercise pattern evaluation results, etc., which are convenient for trainees to view at any time.

[0053] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent coaching system for basketball skill grading, characterized by: It includes a sound generator, an acquisition module, a motion analysis module, a learning guidance module and an injury assessment module, all of which are connected to the processor. The sound generator is used to generate a start whistle; The acquisition module is used to collect sports videos of trainees performing basketball skill grading training; The action analysis module is used to identify the trainee's basketball skill movements in the sports video, compare them with the standard movements, and analyze the correctness of the basketball skill movements; The learning guidance module is used to identify the exercise performance of the trainee in the exercise video and provide learning guidance based on expert knowledge and experience; The injury assessment module is used to identify the movement patterns of the trainees in the sports video and provide injury assessment results based on expert knowledge and experience; The processor receives the working signal of the sound generator, controls the acquisition module, the motion analysis module, the learning guidance module, and the injury assessment module to start working, and generates an intelligent learning guidance report based on the recognition results of the motion analysis module, the learning guidance module, and the injury assessment module; The motion analysis module, learning guidance module, and injury assessment module all use a skeleton joint target tracking algorithm to obtain the position information of each joint of the trainee in each frame image to achieve target tracking. There are 25 joints set. A template matching algorithm is used to identify the position information of the center of the basketball in each frame image. Then, based on the change trends of the position information of each joint and the position information of the basketball center and the relative position relationship between the two, the trainee's basketball skills, sports performance, and movement patterns are identified. The sports video is divided into three stages: starting with the ball, dribbling while moving, and shooting while moving. For the ball-holding start phase The motion analysis module determines whether the ball holding start state is correct based on the relative positions of the basketball and the trainer's hands, the consistency of the positions of the hands and feet on the same side, and the timing relationship between the first landing of the basketball and the back foot; The learning guidance module provides starting training guidance based on the first landing time of the basketball and the corresponding back foot off-ground time, i.e., the starting speed; For the dribbling phase The motion analysis module determines whether the dribbling state is correct based on the relative positions of the dribbling height and the trainee's head and shoulders, as well as the relative positions of the hands and the basketball; The learning guidance module provides dribbling guidance based on the changing trends of the trainee's stride and speed, as well as the changes in the rebound height of the basketball and the changes in the distance between landing points. For the moving shooting phase The motion analysis module determines whether the shooting state is correct while moving based on the relative position relationship between the two feet when holding the ball and the relative position of the basketball and the trainee's shoulders when shooting; The learning guidance module provides shooting guidance on the move based on the angle between the trainee's body and the ground, the jump height and the distance between the jump point and the landing point, the time from the start of the ball to the ball leaving the hand (i.e., the running speed), and the time from the ball leaving the hand to the rebound being grabbed (i.e., the rebounding speed); The injury assessment module determines whether the trainee has a splay foot based on the speed and toe direction of the trainee during movement; and determines whether the landing method is correct based on the order in which the forefoot and the heel of the landing foot land after the shooting jump. Based on the falling direction of the basketball after shooting and the movement direction of the trainee's hands, determine whether the trainee should reach out to meet the ball to avoid finger contusion; based on the deviation of the lower limb force line when starting with the ball and shooting with the ball, determine the possible injury risk level.

2. The intelligent coaching system for basketball skill grading according to claim 1, characterized in that: Set the holding distance threshold between the wrist joint and the center of the ball when holding the ball, and the landing distance threshold between the center of the basketball and the ground when landing. For the ball-holding start phase The motion analysis module compares the relative position information of the center of the basketball and the wrist joints of both hands with the standard relative position, and at the same time determines whether the right ankle joint point on the same side is in front based on the right wrist joint point holding the ball, so as to determine whether the ball holding action is correct; and then determines whether the starting action is correct based on the timing relationship between the first landing of the basketball and the first lifting of the left heel joint point behind the left ankle joint point. The learning guidance module detects the moment when the sound generator starts working, which is time T1; detects the moment when the landing distance between the center of the basketball and the ground is less than the landing distance threshold for the first time, which is time T2; calculates the difference between time T1 and time T2 as the first basketball landing time; detects the moment when the ground distance between the left heel joint behind the left ankle joint and the ground starts to increase for the first time, which is time T3; calculates the difference between time T1 and time T3 as the first back foot off-the-ground time, that is, the starting speed; then compares the first basketball landing time and the first back foot off-the-ground time with the average time of the big data, and provides learning guidance; For the dribbling phase The motion analysis module detects whether the center of the basketball is higher than the trainee's nose joint, left shoulder joint, and right shoulder joint during dribbling to determine whether the dribbling height is correct; and detects whether the distances between the left wrist joint and the right wrist joint and the center of the basketball are both within the ball holding distance threshold to determine whether the dribbling is illegal. Detect whether the distances between the left and right wrist joints and the center of the ball are alternately within the ball holding distance threshold during dribbling to determine whether to switch hands; Detect whether the left or right wrist joint is above the center of the basketball during movement to determine whether to dribble with a wrist flip; The learning guidance module detects the distance between the left and right heel joints corresponding to each step during the march, which is the stride length f and the duration t. f The number of steps n from starting with the ball to starting to shoot, grabbing the rebound and dribbling to the center line is used to obtain the changes in stride length and speed during movement, that is, the changes in displacement speed. The center position of the ball when the basketball bounces to the highest point each time, that is, the rebound height h, is detected. The distance between the center of the basketball and the ground is detected to be less than the landing distance threshold, and then it is determined that the basketball has landed. The distance d between each two adjacent basketball landing points is then detected to obtain the changes in basketball movement during movement. The changes in stride length, speed and basketball movement during movement are then compared with the average duration of big data to provide guidance on dribbling ability during movement. For the moving shooting phase The motion analysis module detects whether the distances between the left wrist joint and the right wrist joint and the center of the basketball are both within the holding distance threshold to determine whether the ball is in contact, and then detects whether the right heel joint is in front of the left heel joint to determine whether the right foot is stepped on first. Then, when the position of the center of the basketball is higher than the left shoulder joint or the right shoulder joint, the distances between the left wrist joint and the right wrist joint and the center of the basketball are within the holding distance threshold to determine whether the basketball is higher than the shoulder when shooting; The learning guidance module fits the nose, neck, hip center, the midpoint of the right eye and the left eye, the midpoint of the right ear and the left ear, the midpoint of the right shoulder and the left shoulder, the midpoint of the right hip and the left hip, the midpoint of the right knee and the left knee, and the midpoint of the right ankle and the left ankle in the joint points to establish the front neutral line; then detects the angle information between the front neutral line and the ground at the start of the jump, as well as the position information S1 of the heel joint point at the start of the jump, and the maximum height h of the heel joint point from the ground after the jump. m And the position information S2 of the heel joint after landing, and calculate the distance between the position information S1 and S2, that is, the take-off and landing distance; Detect the speed of running from the moment of holding the ball to the moment when the distance between the right wrist joint and the center of the basketball begins to exceed the ball holding distance threshold, and detect the speed of grabbing the rebound from the moment when the ball leaves the hand to the moment when both hands touch the ball. Then use the angle information and the maximum height h m , take-off and landing distance, running speed, rebounding speed and the average time of big data are compared to provide shooting guidance during movement; The injury assessment module determines whether the trainee has a bowed foot based on the speed direction of each step during movement and the angle between the toe joint of the landing foot and the corresponding ankle joint; detects the landing moment of the toe joint and the heel joint of the landing foot after taking off to determine whether the landing method is correct; detects the displacement direction D1 of the center of the basketball falling from the highest point to the second highest point after shooting, and the displacement direction D2 of the wrist joint rising from the low point to the high point at this time, and determines whether the two displacement directions are opposite to determine whether the trainee has performed the action of reaching out to meet the ball; detects the angle between the line connecting the knee joint and the toe joint on the same side and the ground when preparing to shoot and starting with the ball to determine the degree of deviation of the lower limb force line.

3. The intelligent guidance system for basketball skill grading according to claim 2, characterized in that: According to the speed of each step = stride f / time t f , obtain the displacement speed change during the movement. During the entire basketball running test process, the displacement speed change is set to acceleration-deceleration-acceleration-deceleration.

4. The intelligent coaching system for basketball skill grading according to claim 1, characterized in that: The acquisition module includes multiple groups of cameras arranged at intervals on the basketball court, all of which are set along the trainee's running trajectory, with at least one group of cameras corresponding to the ball holding starting position, the dribbling position while moving, and the shooting position while moving. They respectively capture the trainee's movement status from opposite sides.

Images