Scoring system, method and equipment for freestyle trolley match

Abstract

The invention relates to the technical field of computer application, and discloses a scoring system, method and equipment for a freestyle racing, and the system comprises an image collection module which is used for obtaining a video stream of a racing of a target object, analyzing one or more action units of the target object in the video stream, and obtaining a scoring result of the racing of the target object; acquiring a plurality of video frames in the action unit; the image analysis module is used for determining a plurality of action parameters in the action units based on the plurality of video frames of each action unit; the upper computer module is used for inputting the action units into the scoring architecture model, the scoring architecture model evaluates the action units according to the multiple action parameters of the action units, and a comprehensive score of the target object is obtained based on an evaluation result; and the display module is used for acquiring the comprehensive score of the target object and displaying the comprehensive score on a terminal screen. According to the technical scheme provided by the invention, the competition score of the freestyle sportsman can be quickly and accurately obtained in a complex environment.

Description

Technical Field

[0001] The present application relates to the field of computer application technology, and in particular to a scoring system, method and device for freestyle BMX competitions. Background Art

[0002] With the continuous development of the computer industry, the application of big data models in various fields is becoming more and more popular. In the field of competitive sports, freestyle BMX is an emerging sport, which was officially listed as an Olympic event in 2020.

[0003] However, freestyle BMX competitions require athletes to use various auxiliary terrains to complete a series of fancy moves in the air, and the complexity of their movements makes it difficult for referees to score. With the continuous development of freestyle BMX and the continuous improvement of athletes, there is an urgent need for a solution that can quickly and accurately score the competition in a complex environment. Summary of the invention

[0004] The present application provides a scoring system, method and device for freestyle BMX competition, which can quickly and accurately obtain the competition results of freestyle BMX athletes.

[0005] The first aspect of the present application provides a scoring system for freestyle BMX competitions, the system comprising an image acquisition module, an image analysis module, a host computer module, and a display module, wherein a scoring architecture model runs in the host computer module, wherein: the image acquisition module is used to obtain a video stream of a competition of a target object, and parse one or more action units of the target object in the video stream, and obtain multiple video frames in the action unit; the image analysis module is used to determine multiple action parameters in each action unit based on each action unit and multiple video frames of the action unit; the host computer module is used to obtain the one or more action units, and input the action units into a preset scoring architecture model, the scoring architecture model evaluates each action unit according to the multiple action parameters of the action units, and obtains a comprehensive score of the target object based on the evaluation result; the display module is used to obtain the comprehensive score of the target object in the host computer module, and display the comprehensive score on the terminal screen.

[0006] In one possible implementation, the image acquisition module includes an image acquisition unit and an image processing unit, wherein: the image acquisition unit is used to acquire a video stream of a game of a target object; the image processing unit is used to divide the target object into one or more action units according to the video stream, and acquire multiple video frames within an action unit, wherein the action unit is a process of the target object performing a vacant action.

[0007] In one possible implementation, the image analysis module includes a height detection unit, a time detection unit, a motion continuity detection unit, and a motion integrity detection unit, and the motion parameters include a flying height, a hovering time, a motion continuity parameter, and a motion integrity parameter, wherein: the height detection unit is used to obtain the first motion data of the current action unit, and determine the flying height of the target object according to the first motion data; the time detection unit is used to obtain the second motion data of the current action unit and the timestamp corresponding to each video frame, and determine the hovering time of the target object according to the second motion data and the timestamp corresponding to each video frame; the motion continuity detection unit is used to obtain multiple video frames of the target object in an action unit and the timestamps of the video frames, and determine the motion continuity parameters of the target object according to the video frames and the timestamps; the motion integrity detection unit is used to obtain a standard action module, which includes multiple specified standard actions, and compares each action unit with the corresponding specified standard action, and determines the action integrity parameter based on the comparison result.

[0008] In one possible implementation, the host computer module includes a basic scoring unit and a comprehensive scoring unit, and a scoring architecture model runs in the basic scoring unit, wherein: the basic scoring unit is used to obtain each action unit of the target object, input the video stream of each action unit into the scoring architecture model, the scoring architecture model evaluates each action unit according to a preset action standard, obtains a standard action code of each action unit based on the evaluation result, and determines a standard score of each action unit according to the standard action code of each action unit; the comprehensive scoring unit is used to obtain the standard score of each action unit, perform weighted calculation on the standard score corresponding to the current action unit according to the action parameters of each action unit, obtain the action score of each action unit based on the weighted calculation result, and obtain the comprehensive score of the target object based on the action score of each action unit.

[0009] A second aspect of the present application provides a scoring method for a freestyle BMX competition, the method comprising: obtaining a video stream of a competition of a target object through an image acquisition module, and parsing one or more action units of the target object in the video stream to obtain multiple video frames in each of the action units; based on the various action units and the multiple video frames of the action units, determining multiple action parameters of the target object in each action unit through an image analysis module, and transmitting the action parameters to a host computer module; the host computer module inputs the multiple action parameters into a preset scoring architecture model, the scoring architecture model evaluates the various action units according to the multiple action parameters of the various action units, obtains a comprehensive score of the target object based on the evaluation result, and transmits the comprehensive score to a display module; the display module receives the comprehensive score and displays the comprehensive score on a terminal screen.

[0010] A third aspect of the present application provides a computer device, which includes a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the above-mentioned freestyle BMX competition scoring method by executing the computer instructions.

[0011] The technical solution provided by this embodiment of the present application improves the scoring speed and accuracy of freestyle BMX competitions by acquiring the video stream of the competition of the target object and automatically scoring the competition process of the target object according to the video stream. Among them, an athlete is selected as the target object, the video stream of the athlete's competition is acquired through the image acquisition module, and the video stream is divided into different action units through the image analysis module. The upper computer module analyzes and evaluates each action unit, obtains multiple action parameters of each action unit and the standard score of each action unit, and then calculates the comprehensive score of the athlete in the video stream, and the comprehensive score is used as the competition result of the athlete.

[0012] It can be seen that through the technical solution provided in this embodiment of the present application, the competition results of freestyle BMX athletes can be obtained quickly and accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0014] Figure 1A schematic diagram of a scoring system for a freestyle BMX competition provided by an embodiment of the present application; Figure 2 A schematic diagram of the steps of a scoring method for a freestyle BMX competition provided by an embodiment of the present application; Figure 3 A schematic diagram of the structure of a computer device provided for one embodiment of the present application. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0016] In addition, the descriptions involving "first", "second", etc. in this application are for descriptive purposes only and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more. In addition, the use of "based on" or "according to" means openness and inclusiveness, because the process, steps, calculations or other actions "based on" or "according to" one or more of the conditions or values ​​can be based on additional conditions or beyond the described values ​​in practice.

[0017] With the rapid development of the computer industry, big data models are increasingly being used in various fields. As a new and challenging sport, freestyle BMX has received widespread attention since it officially became an Olympic event in 2020. Freestyle BMX athletes need to use various auxiliary terrains to complete a series of difficult aerial fancy moves in the competition. However, due to the complexity of the terrain and the diversity of the moves, it is easy for the referee to have a certain scoring error and take too long to consider the score due to the obstruction of the field of vision and the difficulty in seeing the details of the moves. Therefore, the referee often faces great challenges in the scoring process. In some application scenarios, there is an urgent need for a new solution that can quickly and accurately give fair scores to athletes' performances in complex competition environments.

[0018] In view of this, one or more embodiments of the present application provide a scoring system, method and device for freestyle BMX competitions, which can solve the above-mentioned problems, fully consider factors such as the difficulty and completion of the athletes' movements, and quickly and accurately derive the competition scores of freestyle BMX athletes.

[0019] See also Figure 1 One embodiment of the present application provides a scoring system for freestyle BMX competitions, the system comprising an image acquisition module, an image analysis module, a host computer module, and a display module, wherein a scoring architecture model is run in the host computer module, wherein: An image acquisition module, used to obtain a video stream of a game of a target object, parse one or more action units of the target object in the video stream, and obtain multiple video frames in the action units; An image analysis module, configured to determine a plurality of action parameters in each action unit based on each of the action units and a plurality of video frames of the action unit; The host computer module is used to obtain the one or more action units and input the action units into a preset scoring framework model, wherein the scoring framework model evaluates each action unit according to multiple action parameters of each action unit and obtains a comprehensive score of the target object based on the evaluation result; The display module is used to obtain the comprehensive score of the target object in the host computer module and display the comprehensive score on the terminal screen.

[0020] In this embodiment, the above-mentioned image acquisition module includes an image acquisition unit and an image processing unit, wherein the image acquisition unit is used to acquire a video stream of the target object's game; the image processing unit is used to divide the target object into one or more action units according to the video stream, and acquire multiple video frames within an action unit, wherein an action unit is a process of the target object performing a vacant action.

[0021] Specifically, the image acquisition unit collects the video stream of the target object. Taking the freestyle BMX competition as an example, a common webcam or a depth camera such as Kinect or Intel Realsense can be used to directly capture the video stream of the athlete's competition at the competition site, or a live video stream can be obtained through a network protocol or an API interface of a network platform.

[0022] Furthermore, after the image acquisition unit acquires the video stream of the game, the image processing unit extracts one or more action units of the target object in the video stream. In a freestyle BMX competition, the athlete is the target object for detection, and the process of performing a take-off action is an action unit. The target object in the video frame can be detected by a target detection algorithm, and the position of the target object in the video stream can be tracked by a target tracking algorithm. Once the start and end of the take-off action are identified, an action unit is identified, the start frame and end frame of each action unit are determined, and multiple video frames are extracted from one action unit.

[0023] In this embodiment, the image analysis module includes a height detection unit, a time detection unit, a motion continuity detection unit and a motion integrity detection unit, and the motion parameters include the flying height, the hovering time, the motion continuity parameter and the motion integrity parameter, wherein: a height detection unit, used to obtain first motion data of the current action unit, and determine the flying height of the target object according to the first motion data; a time detection unit, configured to obtain the second motion data of the current action unit and the timestamps corresponding to each video frame, and determine the hovering time of the target object according to the second motion data and the timestamps corresponding to each video frame; An action continuity detection unit, used to obtain multiple video frames of a target object in an action unit and timestamps of each video frame, and determine an action continuity parameter of the target object according to the video frames and the timestamps; The action completeness detection unit is used to obtain a standard action module, which includes multiple specified standard actions, and compare each action unit with the corresponding specified standard action, and determine the action completeness parameter based on the comparison result.

[0024] In one embodiment, the first motion data includes but is not limited to the position coordinates of the target object, and the flight height of the target object can be determined by identifying the position coordinates of the target object. In an action unit, a unified reference baseline is first determined, and the ground can be used as the reference baseline. The target detection algorithm is used to obtain the center point coordinates of the target object in each video frame, and the vertical distance between the center point of the target object and the reference baseline is calculated, and the distance is used as the relative height of the current video frame. The relative heights of each video frame are compared, and the relative height value farthest from the reference baseline is used as the flight height.

[0025] In one embodiment, the second motion data includes, but is not limited to, the vertical position coordinates, velocity, and acceleration of the target object, and the duration of the target object's hovering is determined by identifying the position coordinates, velocity, and acceleration of the target object. The target object is continuously tracked using a target detection algorithm and a target tracking algorithm to determine the time when the target object takes off and when it lands.

[0026] Optionally, the launch time and landing time are determined according to the speed and acceleration of the target object. 2 Transition to -9.8m / s 2 , and the vertical velocity of the target object is upward, it can be determined that this moment is the moment when the target object takes off; when the vertical acceleration of the target object changes from -9.8m / s 2 Convert to 0m / s 2, and the vertical speed of the target object is downward, it can be determined that this moment is the landing moment of the target object. Further, the video frames of the flying moment and the landing moment of the target object are obtained, and the time stamps of the video frames of the flying moment and the landing moment are extracted. The difference between the time stamps of the two moments is the duration of the target object's stay in the air.

[0027] Optionally, the take-off moment and the landing moment are determined according to the vertical position coordinates of the target object in each frame in an action unit. The vertical position coordinates of the wheels of the BMX in each video frame are extracted, and the difference in the vertical position coordinates of the wheels between adjacent video frames is calculated. When the difference changes from close to 0 to a positive number and the trend is continuously increasing, the current moment is determined to be the take-off moment; when the difference changes from a positive number to close to 0 and no longer changes thereafter, the current moment is determined to be the landing moment. Furthermore, the video frames of the take-off moment and the landing moment of the target object are obtained, and the respective timestamps of the video frames of the take-off moment and the landing moment are extracted. The duration of the target object's stay in the air can be calculated based on the timestamps of the two moments.

[0028] In one embodiment, action identifiers are aggregated into action sequences according to the timestamps of multiple video frames to determine action continuity parameters. First, all video frames with timestamps in an action unit are obtained, and the action identifiers of each video frame are identified in these video frames. The above action identifiers can be the distribution of the action of a specified part of the target object in space. All video frames are arranged in ascending order according to the timestamps to ensure that the aggregation of action identifiers is carried out in the time order of the video stream to form an action sequence. The action sequence is input into the continuity detection model for detection, and the continuity parameter is obtained based on the detection results.

[0029] Specifically, the continuity detection model can be trained based on action sequence samples. Collect action sequence samples of a series of action sequences, and the action sequence samples cover action sequences of various action types and various continuity parameters. Each action sequence sample is labeled by an expert or a trained annotator, and a continuity label is added. Select a suitable machine learning model such as a support vector machine model or a neural network model for sample training to form a continuity detection model. Use the labeled action sequence samples to train the continuity detection model so that it can predict the continuity of the action sequence. To evaluate the trained continuity model, cross-validation can be used to evaluate the performance of the model, and the model parameters can be adjusted according to the evaluation results to improve the accuracy of continuity recognition. Subsequently, if continuity detection is required, the action sequence to be detected only needs to be input into the continuity detection model to obtain the continuity parameters of the action sequence.

[0030] In an optional implementation, one-hot encoding can also be used to label the action sequence samples. Each action sequence sample is associated with a one-hot encoded label, and its encoding form is such as [0,0,1,0,…,0], that is, a unique label is assigned to each action sequence sample in order to predict its most likely classification. In TensorFlow (TensorFlow, a symbolic mathematical system based on data flow programming), tf.keras (tf.keras, a TensorFlow API interface) can be used to build a continuity detection model, and the model.fit function is used to train the continuity detection model. The continuity detection model performs a learning process according to a preset number of training rounds, and optimizes itself by continuously iterating the labeled training data, thereby improving the accuracy of the model prediction.

[0031] In one embodiment, the action completeness parameters of each action are determined by comparing the action identifier of each video frame with the specified standard action. First, a plurality of preset standard action modules are obtained, and a standard action module includes a plurality of specified standard actions arranged in chronological order. Then, the action identifier of each video frame in the above-mentioned action unit is obtained, and the action identifier is matched with the corresponding specified standard action in multiple dimensions. The first action feature group of the actual action unit can be compared with the second action feature group of the specified standard action, and the completeness of each action is determined based on the comparison result, wherein the first action feature group includes one or more first action features, and the second action feature group includes one or more second action features. The above-mentioned first action feature and second action feature types include but are not limited to posture features and amplitude features.

[0032] Specifically, standard action modules can be collected according to specific scenarios. In the field of freestyle BMX, professionals are required to define multiple representative standard action modules based on possible competition actions, and to provide detailed descriptions of multiple designated standard actions in each standard action module, including but not limited to action posture and action amplitude. According to the actual action sequence of freestyle BMX athletes, all designated standard actions in each standard action module are arranged in chronological order.

[0033] Specifically, the action identifiers of each video frame in an action unit are matched with the corresponding designated standard actions in multiple dimensions. The feature matching may include posture matching, amplitude matching, angle matching, etc. According to the results of the multi-dimensional matching, the action completeness parameter is calculated by weighted average. For example, if the posture matching accounts for 60% and the amplitude matching accounts for 40%, the posture matching is 80% and the amplitude matching is 70%, then the action completeness parameter is 0.76.

[0034] Exemplarily, in posture matching, the posture features in the athlete's competition movements are extracted as the first movement features, and the posture features in the specified standard movements are extracted as the second movement features. The above posture features include the relative positions of each posture key point of the athlete's body. The distance between the above first movement feature and the second movement feature in the posture feature space, such as the Euclidean distance, is calculated, and the matching degree of the posture features within each distance range is set. The smaller the distance, the higher the matching degree.

[0035] Exemplarily, in amplitude matching, the amplitude characteristics of the athlete's movements in competition are extracted as the first action features, and the amplitude characteristics of the movements in specified standard actions are extracted as the second action features. The above amplitude characteristics include the amplitude of the athlete's movements. For example, the distance between both hands and the handlebars is a movement amplitude. The amplitude difference between the first action feature and the second action feature is calculated, and the matching degree of the amplitude characteristics within each difference range is set. The smaller the distance, the higher the matching degree.

[0036] In this embodiment, the host computer module includes a basic scoring unit and a comprehensive scoring unit. A scoring architecture model runs in the basic scoring unit, wherein: A basic scoring unit is used to obtain each action unit of the target object, input the video stream of each action unit into the scoring framework model, the scoring framework model evaluates each action unit according to a preset action standard, obtains a standard action code of each action unit based on the evaluation result, and determines a standard score of each action unit according to the standard action code; A comprehensive scoring unit is used to obtain the standard scores of each action unit, perform weighted calculation on the standard scores corresponding to the current action unit according to the action parameters of each action unit, obtain the action scores of each action unit based on the weighted calculation results, and obtain the comprehensive score of the target object based on the action scores of each action unit.

[0037] In this embodiment, the standard action code of each action unit is first obtained. The video stream of each action unit is input into the scoring architecture model, and the scoring architecture model preprocesses and extracts features from the video stream of each action unit to obtain feature data of each action unit, and the standard action code of each action unit is determined based on the above feature data and preset action standards.

[0038] In an optional implementation, the video stream of each action unit is preprocessed and feature extracted. Since the color of the image does not affect the athlete's action and score in a freestyle BMX competition, the above action sequence sample can use a single-channel grayscale image to reduce the complexity of the data.

[0039] Specifically, the scoring architecture model can be trained using training samples, and the training samples are multiple video samples. TensorFlow is used to load the training samples so that the training samples are converted from video form to feature form. The training samples are converted into a three-dimensional data matrix, and the dimensions of the matrix are pixel height, width, and total number of frames. A 3×3×3 convolution kernel is used and a convolution operation is performed with a step size of 1, and a ReLu (Rectified Linear Unit) activation function is introduced to increase the nonlinearity of the conversion process. Then, a 2×2×2 maximum pooling is applied, and the largest value is selected from each 2×2×2 region as the representative value of the region, while retaining the most important features of the data, effectively reducing the amount of calculation in the subsequent training process and shortening the training time. Subsequently, the video data can be converted into a feature map by performing two convolution and pooling operations, and the feature data of each action unit is generated through a fully connected layer according to the feature map. The feature classification of different scoring groups is performed according to the feature data, and the scoring groups include action groups, separation of people and vehicles, number of athlete changes of direction, and number of BMX rotations. Among them, the action standards of different scoring groups are different classification characteristics, and the standard branches are the manifestations of different classification characteristics.

[0040] Optionally, the reliability of the scoring architecture model can be evaluated using a test sample, which is one or more video samples. Specifically, the test sample is divided into multiple time windows, each of which contains enough information to identify an action. The scoring architecture model is used to predict the action in each time window, and the np.argmax function is used to obtain the category index with the highest probability. The prediction results of all time windows are arranged to obtain the action sequence in the entire video sample. The predicted action sequence is matched with the scoring criteria, and the final result is given.

[0041] In another optional embodiment, feature extraction can extract features related to action units by extracting key frames to reduce the amount of data and retain key information of the action, for example, extract one frame every five frames, or extract key frames when the position or posture change of the action subject between adjacent frames exceeds a certain threshold. Then, extract the relevant features in the key frames and perform feature classification of different scoring groups, wherein the scoring groups include action groups and action features, and the action features include the separation of people and vehicles, the number of times the athlete changes direction, and the number of rotations of the BMX. Wherein, the above-mentioned action features are further characterizations of specific actions in the action groups, and the specific expressions of different scoring groups are different classification features, the group labels are the expression forms of the classification features of different action groups, and the feature labels are the expression forms of the classification features of different action features.

[0042] For example, in a freestyle BMX competition, the group marks and specific expressions of the preset action groups may be shown in Table 1, and the group marks and corresponding specific expressions of the preset action features may be shown in Tables 2, 3 and 4.

[0043] Table 1 Group markings of action groups and corresponding action standards Table 2 Characteristic marks of vehicle-passenger separation and corresponding specific manifestations Table 3 Characteristic marks of the number of direction changes of athletes and their corresponding specific expressions Table 4 Characteristic marks of the number of rotations of a bicycle and their corresponding specific expressions Among them, when determining the action group, the athlete's action characteristics are determined based on the spatial coordinate axis. According to the changes of the athlete's body relative to the spatial coordinate axis, the X axis is the direction of movement, the Y axis is the horizontal axis perpendicular to the direction of movement, and the Z axis is the plumb axis. Therefore, "no flip" means that the changes of the athlete's body relative to the X, Y, and Z axes are all less than 180°, "front flip" means that the athlete's body rotates clockwise around the Y axis by more than 180°, "back flip" means that the athlete's body rotates counterclockwise around the Y axis by more than 180°, "pivot" means that the athlete's body rotates around the Z axis by more than 180° and the rotation is not left or right, and "spiral" means that the athlete's body rotates around the X axis by more than 180° and the rotation is not left or right.

[0044] Specifically, the above standard action codes are output in the form of "number + additional characters". The number includes four digits, and the four digits represent the action group, the separation of the person and the vehicle, the number of changes of direction of the athlete, and the number of rotations of the BMX bike in order from front to back. The additional characters are used to represent some unconventional special actions, such as the reversal action, that is, the athlete controls the handlebars to rotate in one direction first, and then suddenly changes to another direction. This special action can be represented by a special character, for example: H- is defined as the reversal mark, that is, changing the direction of rotation. Other characters can also be used to represent some other special actions.

[0045] For example, in a possible action unit, the athlete walks straight forward, rotates 360° horizontally around the Z axis, and then lands, and completes two rotations of the handlebars with one hand while in the air, and turns during the rotation of the handlebars. After evaluation, the standard action code obtained for this action unit is "3124H".

[0046] For example, in a possible action unit, the athlete flips backward 180° around the Y axis and lands on the ground, and his feet are separated from the car during the flight, and the car body rotates one circle. The standard action code obtained after evaluation of the action unit is "2412".

[0047] For example, in one possible action unit, the athlete flips backward 180° around the Y axis and rotates horizontally 360° around the Z axis. ° Then it landed on the ground, and during the flight, its feet were separated from the car, and the car body rotated twice. After evaluation, the standard action code obtained by its action unit was "6434".

[0048] It should be noted that although the rotation axes of the athletes are different, the number of rotations around the rotation axes can be represented by the same number. For example, the first digit corresponding to a 360° rotation around the X-axis is "4", and the first digit corresponding to a 360° rotation around the Y-axis is "1" or "2", but in both cases the third digit can be represented by "2". Of course, because the difficulty of the athlete's rotation around the X-axis and around the Y-axis is different, even if the third digit is the same, the corresponding standard action code is different because it belongs to different action groups. Therefore, the difficulty of the action can be better distinguished and determined, thereby assigning different standard scores.

[0049] It should be noted that the third number represents the athlete's change of direction. For Group 1 and Group 2, it refers to the angle of the athlete's body changing around the Y axis during the entire process from the start of take-off to the completion of landing; for Group 3, it refers to the angle of the athlete's body changing around the Z axis during the entire process from the start of take-off to the completion of landing; for Group 4, it refers to the angle of the athlete's body changing around the X axis during the entire process from the start of take-off to the completion of landing; for Group 5 and Group 6, it refers to the sum of the angles of the athlete's body changing around the Y axis and around the Z axis during the entire process from the start of take-off to the completion of landing; for Group 7 and Group 8, it refers to the sum of the angles of the athlete's body changing around the Y axis and around the X axis during the entire process from the start of take-off to the completion of landing; for Group 9, it refers to the sum of the angles of the athlete's body changing around the X axis, Y axis and Z axis during the entire process from the start of take-off to the completion of landing.

[0050] It should be noted that when the second digit is 0, it means that the person and the vehicle are not separated. At this time, the number of circles turned by the athlete is equal to the number of circles turned by the BMX vehicle, that is, the third digit is the same as the fourth digit; when the second digit is 1 or 2, the fourth digit refers to the number of rotations of the handlebars; when the second digit is 3 or 4, the fourth digit refers to the number of rotations of the body of the vehicle; when the second digit is 5, the fourth digit refers to the sum of the number of rotations of the handlebars and the body of the vehicle.

[0051] It should be noted that the additional characters are set to represent some unconventional actions, and not every scoring unit will have additional characters. When an athlete changes the direction of the handlebar rotation in the air, it cannot be expressed by the four-digit standard action code, so the additional character "H" is added to indicate the handlebar reversal.

[0052] In one embodiment, a corresponding standard score is set for each standard action code in advance, and a corresponding standard score is assigned to each action unit according to the standard action code of each action unit. Since the action difficulty of each scoring unit is different, the corresponding standard score will also be different. The higher the difficulty coefficient of the action, the higher the preset standard score.

[0053] In one embodiment, the action parameters of each action unit are weighted and summed with the standard score, the action score of each action unit is calculated, and the action scores of each action unit are added to obtain the comprehensive score of the target object. The action parameters of each action unit are obtained, including the flight height, the duration of the air, the action continuity parameter and the action completeness parameter, and a weight is preset for each of the above action parameters. The standard score of the current action unit is weighted and calculated according to the above action parameters and the preset weight to obtain the action score of the current action unit. The action scores of all action units in the video stream of the target object's game are added to obtain the comprehensive score of the target object.

[0054] Exemplarily, the score of a possible action unit is 3124H, and its preset standard score is 11 points. Its various action parameters are: the take-off height is 110% of the preset standard height, the hovering time is 90% of the preset standard time, the action continuity parameter is 1, and the action completeness parameter is 0.76. The preset weights of its various action parameters are: the take-off height is 20%, the hovering time is 20%, the action continuity parameter is 30%, and the action completeness parameter is 30%. Then the action score of the current action unit is calculated as (1.1×20%+0.9×20%+1×30%+0.76×30%)×11, which is 10.208.

[0055] For example, in a possible freestyle BMX race video stream, the athlete performs three take-offs in sequence, and the standard action codes of the action units are 3124H, 2412, and 6434, and the corresponding preset standard scores are 11 points, 9 points, and 17 points, respectively. Then, according to the preset weights of 20% for take-off height, 20% for air time, 30% for action continuity parameter, and 30% for action completeness parameter, if the action scores of each action unit weighted by the action parameters are 10.208, 9.252, and 17.476, respectively, the athlete's final score is 36.936.

[0056] See also Figure 2 One embodiment of the present application provides a scoring method for a freestyle BMX competition, the method is applied to the above-mentioned freestyle BMX competition system, and the steps of the method are as follows: S1: acquiring a video stream of a game of a target object through an image acquisition module, parsing one or more action units of the target object in the video stream, and acquiring multiple video frames in each of the action units; S3: Based on the action units and the multiple video frames of the action units, determine multiple action parameters of the target object in each action unit through an image analysis module, and transmit the action parameters to a host computer module; S5: the host computer module inputs the multiple action parameters into a preset scoring framework model, the scoring framework model performs weighted calculation on the multiple action parameters, obtains a comprehensive score of each action unit based on the weighted calculation result, and transmits the comprehensive score to the display module; S7: The display module receives the comprehensive score and displays the comprehensive score on a display screen.

[0057] In this embodiment, firstly, a video stream of a target object's game is obtained. For example, a freestyle BMX game is used. Firstly, a video stream of a player's game is obtained and divided into multiple action units. The process of a player's vacating is an action unit. Then, each action parameter of the player in each action unit is obtained. The action parameters include vacating height, lingering time, action continuity parameter and action integrity parameter. The standard action code of each action unit is obtained, and the standard score of each action unit is determined according to the standard action code. Finally, the standard score of each action unit is weighted according to the action parameters of each action unit, and the action score of each action unit is obtained according to the weighted operation. The action scores of each unit are added together to obtain the comprehensive score of the player in the current game video stream. In addition, the comprehensive score can be displayed by a display device for reference by the players and referees.

[0058] For the specific implementation of the method embodiment of the present application, please refer to the description of the aforementioned system embodiment, and the method embodiment can also achieve the technical effect of the system embodiment, which will not be repeated here.

[0059] The technical solution provided by one or more embodiments of the present application improves the scoring speed and accuracy of freestyle BMX competitions by acquiring the video stream of the competition of the target object and automatically scoring the competition process of the target object according to the video stream. Among them, an athlete is selected as the target object, the video stream of the athlete's competition is acquired through the image acquisition module, and the video stream is divided into different action units through the image analysis module. The upper computer module analyzes and evaluates each action unit, obtains multiple action parameters of each action unit and the standard score of each action unit, and then calculates the comprehensive score of the athlete in the video stream, and the comprehensive score is used as the competition result of the athlete.

[0060] It can be seen that the technical solutions provided by one or more embodiments of the present application can quickly and accurately obtain the competition results of freestyle BMX athletes. Since the competition video of the athlete is obtained through the image acquisition module, the camera can detect the competition venue more comprehensively, and its field of view of the competition venue is less restricted. By processing and analyzing the video stream to identify the athlete's competition movements, it can avoid certain scoring errors and long scoring consideration time caused by the referee's blocked field of view and difficulty in seeing the details of the movement, thereby more quickly and accurately assigning points to freestyle BMX athletes.

[0061] The embodiments of the present application also provide a freestyle BMX competition scenario example. The scoring system, method and device for a freestyle BMX competition provided in the present application can be applied to the above-mentioned freestyle BMX competition scenario example.

[0062] In actual application, a freestyle BMX race usually lasts for 60 seconds. There are obstacles in the venue to assist athletes in performing aerial movements, including walls, flying platforms, and herringbone slopes. When the athlete starts to move, a 60-second countdown begins. Any aerial movement completed by the athlete will be scored. The standard for judging the end of the movement is that both wheels are in contact with the horizontal plane. One aerial movement is an action unit. Usually, an athlete can complete 5 to 6 aerial movements during a competition, which is 5 to 6 action units.

[0063] In the present embodiment, each action unit has its corresponding standard action module. In freestyle BMX competition, common action types have "dragon tail swing", "aerial level", "superman" etc., wherein, the action meaning of "dragon tail swing" is that the athlete keeps body and handlebar fixed in the air and rotates the car under the body around the front axle as the center and rides back to the seat to land, the action meaning of "aerial level" is that the athlete makes the body parallel to the ground in the air and then swings the body back to the ground, and the action meaning of "superman" is that the athlete leaves the car pedal with both legs in the air and makes the body and both legs straighten as much as possible behind the car. Each action type has a corresponding standard action module, and its standard action module includes a plurality of specified standard actions arranged in time.

[0064] In the present embodiment, first, the image acquisition module obtains the video stream of the athlete in the 60s game time, which can be collected in real time by the on-site camera, or the network video can be retrieved by the API interface. Then, the above-mentioned video stream is divided into different action units according to the athlete's action process by the image analysis module, and multiple action parameters of each action unit are obtained. Different action units are carried out target detection and target tracking by the host computer module, the action type of the athlete in an action unit is determined, and the standard action code is performed to the action type according to the preset standard, for example, "Shenlong Wagtail" determines that the standard score of the action unit is 6 points according to the standard action code, and according to the action parameters and the standard score of the action unit, the action score of the action unit is obtained by weighted calculation. According to the above method, the action score of all action units in the 60s game time is calculated, and all action scores are added to obtain the comprehensive score of the athlete, which is the score of the athlete's round of competition.

[0065] The above description is only an example scenario provided in the specification and is not intended to limit the present invention. Any modifications, equivalent substitutions, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0066] A scoring system for a freestyle BMX competition in an embodiment of the present application is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, or other devices that can provide the above functions.

[0067] See also Figure 3 , Figure 3 is a schematic diagram of the structure of a computer device provided in an embodiment of the present application, such as Figure 3As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 3 A processor 10 is taken as an example.

[0068] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0069] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0070] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0071] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0072] The computer device further includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means, and FIG. X takes the bus connection as an example.

[0073] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device may be a touch screen.

[0074] The embodiment of the present application also provides a computer-readable storage medium. The above method according to the embodiment of the present application can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0075] The systems, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0076] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0077] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0078] The present application is described with reference to the flowcharts and / or block diagrams of the methods, systems, and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0079] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0080] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0081] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0082] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0083] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

[0084] Although the embodiments of the present application have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A scoring system for freestyle BMX competitions, characterized in that: The system includes an image acquisition module, an image analysis module, a host computer module, and a display module. A scoring architecture model runs in the host computer module, wherein: The image acquisition module is used to obtain a video stream of a game of a target object, parse one or more action units of the target object in the video stream, and obtain multiple video frames in the action units; The image analysis module is used to determine a plurality of action parameters in each action unit based on each action unit and a plurality of video frames of the action unit; The host computer module is used to obtain the one or more action units and input the action units into a preset scoring framework model, wherein the scoring framework model evaluates each action unit according to multiple action parameters of each action unit and obtains a comprehensive score of the target object based on the evaluation result; The display module is used to obtain the comprehensive score of the target object in the host computer module and display the comprehensive score on the terminal screen.

2. The system according to claim 1, characterized in that The image acquisition module includes an image acquisition unit and an image processing unit, wherein: The image acquisition unit is used to acquire a video stream of a game of a target object; The image processing unit is used to divide the target object into one or more action units according to the video stream, and obtain multiple video frames in one action unit, wherein the one action unit is a process of the target object performing a vacant action.

3. The system according to claim 1, characterized in that The image analysis module includes a height detection unit, a time detection unit, a motion continuity detection unit and a motion integrity detection unit, and the motion parameters include a flying height, a hovering time, a motion continuity parameter and a motion integrity parameter, wherein: The height detection unit is used to obtain first motion data of the current action unit and determine the flying height of the target object according to the first motion data; The time detection unit is used to obtain the second motion data of the current action unit and the timestamps corresponding to each video frame, and determine the hovering time of the target object according to the second motion data and the timestamps corresponding to each video frame; The action continuity detection unit is used to obtain multiple video frames of the target object in an action unit and the timestamps of the respective video frames, and determine the action continuity parameters of the target object according to the video frames and the timestamps; The action completeness detection unit is used to obtain a standard action module, which includes multiple specified standard actions, and compare each action unit with the corresponding specified standard action, and determine the action completeness parameter based on the comparison result.

4. The system according to claim 3, characterized in that The height detection unit is used to obtain first motion data of the current action unit, and determine the flying height of the target object according to the first motion data, specifically including: Determine a unified reference baseline in the video frame of the current action unit, detect first motion data of the target object, and determine the relative height of the target object in each video frame based on the reference baseline and the first motion data, wherein the first motion data includes the position coordinates of the target object; Based on the relative heights of the target objects in the respective video frames, a relative height farthest from the reference baseline is determined as the flight height.

5. The system according to claim 3, characterized in that The time detection unit is used to obtain the second motion data of the current action unit and the timestamps corresponding to the multiple video frames, and determine the hovering time of the target object according to the second motion data and the timestamps corresponding to the respective video frames, specifically including: Acquire second motion data of the current action unit, and determine the video frames of the launch moment and the landing moment of the current action unit according to the second motion data, wherein the second motion data includes the vertical position coordinates, speed and acceleration of the target object; The time stamps of the video frame at the time of taking off and the video frame at the time of landing are obtained, and the air duration of the target object is calculated according to the time stamps.

6. The system according to claim 3, characterized in that The action continuity detection unit is specifically used for: Acquire multiple video frames of the target object, and identify an action identifier of the target object in each of the video frames; aggregating the identified action identifiers into an action sequence according to the timestamps of the multiple video frames; The action sequence is input into a continuity detection model for detection, and action continuity parameters are obtained based on the detection results, wherein the continuity model is generated through training based on action sequence samples.

7. The system according to claim 6, characterized in that The continuity model is generated through training according to the action sequence samples, specifically including: Acquire an action sequence sample, wherein the action sequence sample includes an action sequence set of multiple action types, and the action sequence set includes multiple action sequences with different continuities; Inputting the above action sequence samples into the continuity model to train the continuity model; The trained continuity model is evaluated, and the continuity model is adjusted based on the evaluation result.

8. The system according to claim 3, characterized in that The action completeness detection unit is specifically used for: Acquire one or more standard action modules, wherein the standard action module includes a plurality of specified standard actions arranged in time; A plurality of video frames of each action unit are obtained, an action identifier of each video frame is identified, and feature matching is performed between the action identifier and the specified standard action, and an action completeness parameter of each action unit is determined based on the feature matching result.

9. The system according to claim 8, characterized in that Performing feature matching between the action identifier and the specified standard action, and determining the completeness parameter of each action unit based on the feature matching result includes: Extracting a first action feature group of the action identifier and a second action feature group of the specified standard action, and performing multi-dimensional feature matching on the first action feature group and the second action feature group, wherein the first action feature group includes one or more first action features, and the second action feature group includes one or more second action features; According to the multi-dimensional feature matching results, the action completeness parameter of each action unit is obtained through weighted calculation.

10. The system according to claim 1, characterized in that The host computer module includes a basic scoring unit and a comprehensive scoring unit. A scoring architecture model runs in the basic scoring unit, wherein: The basic scoring unit is used to obtain each action unit of the target object, input the video stream of each action unit into the scoring framework model, the scoring framework model evaluates each action unit according to the preset action standard, obtains the standard action code of each action unit based on the evaluation result, and determines the standard score of each action unit according to the standard action code of each action unit; The comprehensive scoring unit is used to obtain the standard scores of each action unit, perform weighted calculation on the standard scores corresponding to the current action unit according to the action parameters of each action unit, obtain the action scores of each action unit based on the weighted calculation results, and obtain the comprehensive score of the target object based on the action scores of each action unit.

11. A scoring method for a freestyle BMX competition, characterized in that: The method comprises: Acquire a video stream of the game of the target object through an image acquisition module, and parse one or more action units of the target object in the video stream to acquire multiple video frames in each of the action units; Based on the action units and the multiple video frames of the action units, determine multiple action parameters of the target object in each action unit through an image analysis module, and transmit the action parameters to a host computer module; The host computer module inputs the multiple action parameters into a preset scoring framework model, and the scoring framework model evaluates each action unit according to the multiple action parameters of each action unit, obtains a comprehensive score of the target object based on the evaluation result, and transmits the comprehensive score to the display module; The display module receives the comprehensive score and displays the comprehensive score on a terminal screen.

12. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the scoring method for freestyle BMX competition described in claim 11 by executing the computer instructions.

Images