CN116720784A - Intelligent judging method for ball event and electronic equipment - Google Patents
Intelligent judging method for ball event and electronic equipment Download PDFInfo
- Publication number
- CN116720784A CN116720784A CN202310753667.8A CN202310753667A CN116720784A CN 116720784 A CN116720784 A CN 116720784A CN 202310753667 A CN202310753667 A CN 202310753667A CN 116720784 A CN116720784 A CN 116720784A
- Authority
- CN
- China
- Prior art keywords
- model
- referee
- intelligent
- data
- score
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims description 56
- 238000012545 processing Methods 0.000 claims abstract description 31
- 238000012937 correction Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims description 50
- 238000007405 data analysis Methods 0.000 claims description 19
- 238000001514 detection method Methods 0.000 claims description 16
- 238000004891 communication Methods 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 8
- 230000006399 behavior Effects 0.000 claims description 7
- 230000010354 integration Effects 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000010801 machine learning Methods 0.000 abstract description 4
- 238000011156 evaluation Methods 0.000 abstract description 2
- 238000013527 convolutional neural network Methods 0.000 description 12
- 238000012706 support-vector machine Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000010191 image analysis Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000001788 irregular Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000037361 pathway Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06398—Performance of employee with respect to a job function
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Abstract
The application provides an intelligent judge evaluation method for ball events, which comprises the following steps: pre-constructing an intelligent referee prediction model; acquiring field competition condition data, and performing data processing on the field competition condition data to generate data to be judged; the on-site competition condition data comprise competition condition videos shot by a high-speed camera; inputting the data to be refereed into the intelligent referee prediction model to obtain a score discrimination result of the intelligent referee prediction model on the data to be refereed; judging whether the score judging result accords with the international judging rule of the ball event; if the score is not matched with the score, correcting the score judgment result through an error correction inlet to generate an error correction example; the error correction example is collected and used for optimizing the intelligent referee prediction model, and based on the electronic equipment provided by the error correction example, the intelligent referee of the ball game is realized by utilizing a machine learning technology, so that subjective factor interference caused by the artificial referee in the prior art can be reduced, and the misjudgment rate is reduced.
Description
Technical Field
The application relates to the field of intelligent management of sports activities, in particular to an intelligent judging method for ball events and electronic equipment.
Background
In recent years, the rapid development of artificial intelligence brings about technical revolution in many industries, which is more than capable of achieving technical effects exceeding those of human experts. Conventionally, judges of sports events are served as professionals. The expert as referee not only needs hard professional knowledge to remember the referee rule, but also has better eye and response agility. For highly stressed games, the game is slightly careless, and erroneous judgment, misjudgment and missed judgment can be caused; or the referee is unfair due to subjective factors. Based on the method and the system, how to construct a set of intelligent referee method and system reduces factors of manual participation in referees, thereby ensuring fairness and accuracy of referees and becoming a technical problem to be solved urgently in the field of intelligent sports.
Disclosure of Invention
In view of the above, the embodiments of the present application provide an intelligent referee method and electronic device for ball events, which performs referee scoring on ball events by means of artificial intelligence, so as to solve the problems of fairness and accuracy of ball event referees, where the technical scheme is as follows:
according to one aspect of the embodiment of the application, there is provided a method for evaluating an intelligent referee for a ball game, comprising: pre-constructing an intelligent referee prediction model; acquiring field competition condition data, and performing data processing on the field competition condition data to generate data to be judged; the on-site competition condition data comprise competition condition videos shot by a high-speed camera; inputting the data to be refereed into the intelligent referee prediction model to obtain a score discrimination result of the intelligent referee prediction model on the data to be refereed; judging whether the score judging result accords with the international judging rule of the ball event; if the score is not matched with the score, correcting the score judgment result through an error correction inlet to generate an error correction example; the error correction paradigm is collected for optimizing the intelligent referee prediction model.
In an exemplary embodiment, the method for constructing the intelligent referee prediction model includes: constructing an initial model of an intelligent referee prediction model; acquiring historical competition condition data of the ball competition, and performing data processing on the historical competition condition data to acquire historical data to be judged; acquiring international judge rules and expert experiences of the ball event, and constructing a scoring rule base; indexing the historical data to be refereed based on the scoring rule base to obtain a training data set; and inputting the training data set into the intelligent referee prediction model for training and debugging so as to obtain the intelligent referee prediction model.
In an exemplary embodiment, the acquiring the historical game condition data of the ball game event, and performing data processing on the historical game condition data to obtain historical judging data specifically includes: determining a target object to be referee of the ball game according to the international referee rule of the ball game; slicing the historical race condition data based on the motion trail of the target object to be judged; the slice comprises a plurality of frames of continuous video images; detecting whether each slice contains a referee key frame of the referee target object; extracting a target slice containing the referee key frame; performing frame-by-frame gray level conversion on the target slice to obtain a target slice after gray level processing; performing target detection and target tracking on the target slice subjected to the gray processing to obtain a motion track of a target object to be judged; and indexing the judge result of the target slice based on the motion trail of the target object to be judged and the international judge rule so as to obtain the historical data to be judged.
In an exemplary embodiment, the referee key frames of the referee target object include player foul key frames and sphere score key frames, the slicing the historical race data comprising: the historical race data is sliced based on the integrity of the player's foul behavior and the integrity of the sphere scoring process.
In an exemplary embodiment, the initial model for constructing the intelligent referee prediction model specifically includes: creating a race condition data analysis sub-model for carrying out target detection and target tracking on the data to be refereed; creating a score decision sub-model for carrying out score discrimination on the data to be refereed; and taking the race condition data analysis sub-model as a front model of the score decision sub-model, and carrying out model integration on the race condition data analysis sub-model and the score decision sub-model to generate an initial model of the intelligent referee prediction model.
In an exemplary embodiment, the method for constructing the race condition data analysis sub-model specifically includes: constructing a spatial information learning model based on a fast R-CNN model; constructing a time sequence information learning model based on an RNN model; the time sequence information learning model takes the space information learning model as a pre-model, and builds the race condition data analysis sub-model based on an integration algorithm.
In an exemplary embodiment, the method for constructing the spatial information learning model specifically includes: pre-training the fast R-CNN model based on an ImageNet data set to obtain a pre-trained fast R-CNN model; and taking the historical data to be refereed as a training set of the fast R-CNN model, and performing fine adjustment on the fast R-CNN model to obtain the spatial information learning model.
In an exemplary embodiment, the training method of the time sequence information learning model includes: and taking the historical race condition data as a training set of the time sequence information learning model, and training the time sequence information learning model.
In an exemplary embodiment, the method for constructing the score decision sub-model specifically includes: constructing an initialization model of the score decision sub-model based on the SVM model; and taking the training data set as a training sample of the score decision sub-model, and training and debugging the SVM model to obtain the score decision sub-model.
According to an aspect of an embodiment of the present application, there is provided an electronic apparatus including: at least one processor, at least one memory, and at least one communication bus, wherein the memory stores a computer program thereon, the processor reads the computer program in the memory through the communication bus; the computer program, when executed by the processor, implements an intelligent referee method for a ball game as described above.
The technical scheme provided by the embodiments of the application has the beneficial effects that:
1. the intelligent referee prediction model is constructed, the on-site video is shot by means of the high-speed camera, the video is transmitted to the rear-end intelligent referee prediction model, and the referee result prediction of the ball event is realized by a machine learning method, so that subjective factor interference caused by the artificial referee in the prior art can be reduced, and the misjudgment rate is reduced.
2. The scoring rule base is perfected by introducing international referee rules and expert experience so as to cope with various irregular referee scenes on the competition field, and the generalization capability of the model and the referee prediction accuracy are improved.
3. Slicing the historical race condition data based on the integrity of the player foul behaviors and the integrity of the sphere scoring process, screening out slices containing key frames, performing target detection and target tracking on the screened target slices, and marking the target slices to serve as a model training set, so that data with low information are discarded, the data processing amount and the data training amount are compressed, and the model training efficiency is improved; meanwhile, the overfitting of the model can be avoided, and the generalization capability of the model is improved.
4. By decomposing the intelligent referee prediction model into the race condition data analysis sub-model and the score decision sub-model, correspondingly, decomposing the referee task into two sub-tasks, namely a target detection task and a score decision task, compared with the single model, the method has the advantages that the integral referee task is realized, the construction difficulty of the intelligent referee prediction model is reduced, and the construction efficiency is improved for quickly constructing the intelligent referee prediction model.
5. The historical race condition data is used as a training set of the time sequence information learning model, and the time sequence information learning model is trained, so that the fitting capacity of the time sequence information learning model to the complete time sequence of the ball race is improved, and the time sequence understanding of the slice to be tested after the time sequence information learning model is used as an integrated model is improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings that are required to be used in the description of the embodiments of the present application will be briefly described below.
FIG. 1 is a schematic view of an environment in which an intelligent referee method for ball events according to various exemplary embodiments of the present application is implemented;
FIG. 2 is a flow chart of an intelligent referee method for a ball game, according to an exemplary embodiment;
FIG. 3 is a schematic diagram of an intelligent referee evaluation system for a ball game, according to an exemplary embodiment;
FIG. 4 is a flowchart illustrating a method of constructing the intelligent referee prediction model according to an exemplary embodiment;
FIG. 5 is a flowchart illustrating a method for data processing of the historical race data, according to an exemplary embodiment;
fig. 6 is a schematic diagram showing a hardware structure of an electronic device according to an exemplary embodiment.
Detailed Description
Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative only and are not to be construed as limiting the application.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless expressly stated otherwise, as understood by those skilled in the art. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. The term "and/or" as used herein includes all or any element and all combination of one or more of the associated listed items.
As described above, how to construct a set of intelligent referee expert system to reduce the factors involved in referees manually, so as to ensure fairness and accuracy of referees, is a technical problem to be solved in the intelligent sports field.
For this purpose, according to an exemplary embodiment of the present application, the intelligent referee method for a ball game can intelligently referee the game data captured by the high-speed camera by means of the machine learning algorithm, and the implementation program of the intelligent referee method for a ball game can be deployed on an electronic device, where the electronic device may be a computer device configured with von neumann architecture, and the computer device may be a computer, a server, a virtual machine, or other devices with computing capabilities, for example.
For the purpose of making the objects, technical solutions and advantages of the present application more apparent, the embodiments of the present application will be described in further detail with reference to the accompanying drawings.
FIG. 1 is a schematic diagram of an implementation environment of an intelligent referee method for ball events. It should be noted that this implementation environment is only one example adapted to the embodiments of the present application, and should not be considered as providing any limitation to the scope of use of the present application.
The implementation environment includes a race condition collection end 110 and a server end 130 deployed with an intelligent referee prediction model.
Specifically, the race condition collection end 110 may be an electronic device or an embedded program with the capability of collecting at least one or more of text, audio and video, which are not limited herein.
A server 130, where the server 130 may be an electronic device or a virtual device such as a computer, a server, a virtual machine, and the like; it may also be a cluster of computer devices made up of multiple servers, or even a cloud computing center made up of multiple servers. The server 130 may be configured to provide services such as Al calculation, push messaging, etc., for example, background services including, but not limited to, providing referee prediction results for training, optimizing, and treating the predicted event data of the intelligent referee prediction model.
The server 130 and the race condition collection end 110 are pre-connected by wire or wireless, and data transmission between the server 130 and the race condition collection end 110 is realized through the network communication connection. The data transmitted includes, but is not limited to: historical race condition data, field race condition data, and the like.
Through the interaction between the race condition collection terminal 110 and the service terminal 130, the race condition collection terminal 110 collects historical race condition data and field race condition data from the front end and sends the historical race condition data and the field race condition data to the service terminal 130, the service terminal 130 processes the obtained historical race condition data and trains the obtained historical race condition data, so that the intelligent referee prediction model is obtained, and the field race condition data is refereed to by the intelligent referee prediction model.
The embodiment of the application provides an intelligent judging method for ball events. The method is suitable for deployment on an electronic device.
In the following method embodiments, for convenience of description, an execution body of each step of the method is taken as an example of a system for executing the method; in addition, in order to facilitate understanding and imagining the application scenario of the present solution by those skilled in the art, the following technical gist will be further described by taking badminton as an example, but this is not a specific limitation.
As shown in fig. 2 and 3, the method for evaluating an intelligent judge of a ball game according to an exemplary embodiment of the present application includes the following steps:
s1: pre-constructing an intelligent referee prediction model;
specifically, the intelligent referee prediction model is used for refereing the competition condition of the ball competition; the ball event includes: badminton games, basketball, table tennis games, volleyball games, etc.; specifically, which kind of ball event needs to be refereed, the intelligent referee prediction model needs to be trained in advance according to different ball events and different competition rules, so that the referee prediction model learns the referee rules of the ball events. For example. If the model is a badminton match, the intelligent referee prediction model needs to be trained or fine-tuned by a training set of the badminton match, referee rules of the badminton match and the like.
S2: acquiring field competition condition data, and performing data processing on the field competition condition data to generate data to be judged; the on-site competition condition data comprise competition condition videos shot by a high-speed camera;
specifically, after the intelligent referee prediction model is built, the video recorded by the ball game may be collected by the game condition collection end 110, so as to obtain on-site game condition data. In one possible implementation, the field competition condition data includes competition condition videos shot by high-speed cameras, the high-speed cameras on the competition field record the competition conditions in the whole course to generate video files, and the video files are stored in a memory, so that the field competition condition data are formed.
Carrying out data processing on the acquired field competition condition data; the data processing means is consistent with the later described historical race data processing logic and will be described only briefly herein, the specific method being described below. The data processing mode comprises the following steps: slicing the field competition condition data; and indexing the key frames, extracting slices containing the key frames based on the key frames, performing image processing, and storing the slices as data to be judged.
S3: inputting the data to be refereed into the intelligent referee prediction model to obtain a score discrimination result of the intelligent referee prediction model on the data to be refereed;
specifically, the processed data to be refereed can be transmitted to the intelligent referee prediction model in a calling interface mode, and model prediction is carried out on the data to be refereed through the intelligent referee prediction model, so that a score discrimination result of the data to be refereed is obtained; as described above, the data to be refereed is a set of target slices; therefore, the score discrimination result of each target slice is obtained by carrying out score discrimination result prediction on the race condition information of each target slice, and the score discrimination result of each target slice is accumulated to obtain the score discrimination result of the intelligent referee prediction model on the data to be refereed.
S4: judging whether the score judging result accords with the international judging rule of the ball event; if the score is not matched with the score, correcting the score judgment result through an error correction inlet to generate an error correction example;
specifically, the score discrimination result of each target slice is transmitted to an expert supervision module as a supervision value. And rechecking the score discrimination result of each target slice by a referee based on the international referee rule of the ball game, and if the score discrimination result does not accord with the international referee rule, correcting the score discrimination result by an error correction inlet so as to update the score discrimination result of the intelligent referee prediction model on the data to be refereed, and generating an error correction example. A paradigm of error correction paradigm may be stored as: [ object slice; original judging results; target discrimination result ].
And then, outputting and displaying the score discrimination result of the intelligent referee prediction model rechecked by the expert supervision module to the data to be refereed on an interface.
S5: the error correction paradigm is collected for optimizing the intelligent referee prediction model.
Specifically, the error correction example is used as a training set for further optimizing the intelligent referee prediction model of the ball game; and performing performance optimization on the intelligent referee prediction model.
Referring to fig. 4, in order to enable the intelligent referee prediction model to have referee capability, even to make fair and accurate referees by means of algorithm advantages, an efficient and feasible intelligent referee prediction model needs to be built according to the task complexity and the actual situation of the training set size; thus, in an exemplary embodiment, the method of constructing the intelligent referee prediction model includes:
s11: constructing an initial model of an intelligent referee prediction model;
specifically, since the event referee task can be decomposed into a video image analysis task and a score discrimination task according to international referee standards and expert experience, an initial model of the intelligent referee prediction model can also be constructed according to the above logic, that is, the method includes: creating a race condition data analysis sub-model for carrying out video image analysis on the data to be refereed so as to carry out target detection and target tracking; creating a score decision sub-model for performing score discrimination on the target detection and target tracking results of the data to be refereed; and taking the race condition data analysis sub-model as a front model of the score decision sub-model, and carrying out model integration on the race condition data analysis sub-model and the score decision sub-model to generate an initial model of the intelligent referee prediction model.
Because the task for which the score decision sub-model is responsible belongs to the category of classification tasks, an SVM (support vector machine) model can be selected as an initial model of the score decision model, and therefore, the construction method of the score decision sub-model specifically comprises the following steps: constructing an initial model of the score decision sub-model based on the SVM model; and taking the training data set as a training sample of the score decision sub-model, and training and debugging the SVM model to obtain the score decision sub-model.
Of course, another possible implementation is to choose KNN or a decision tree model as the initial model of the scoring decision model. In general, a model that is suitable for performing the classification task may be used as the initial model of the scoring decision model.
S12: acquiring historical competition condition data of the ball competition, and performing data processing on the historical competition condition data to acquire historical data to be judged;
s13: acquiring international judge rules and expert experiences of the ball event, and constructing a scoring rule base;
s14: indexing the historical data to be refereed based on the scoring rule base to obtain a training data set;
s15: and inputting the training data set into the intelligent referee prediction model for training and debugging so as to obtain the intelligent referee prediction model.
In addition, referring to fig. 5, the logic for performing data processing on the historical race condition data is consistent with the on-site data processing logic, so the step of performing data processing on the historical race condition data includes:
s121: determining a target object to be referee of the ball game according to the international referee rule of the ball game; for example, target objects to be refereed for badminton games include player actions, badminton drop points, tracks, etc.;
s122: slicing the historical race condition data based on the motion trail of the target object to be judged; the slice comprises a plurality of frames of continuous video images; each historical race condition data may be video data of a race; or may be part of video data of a game; in general, since the course is long, the amount of one piece of historical race data is large, in order to reduce the single data processing amount, each piece of historical race data needs to be sliced, so that a key frame is searched in each slice later, and a slice without the key frame is discarded to compress the data processing amount. Typically, each slice contains a plurality of consecutive frame images.
S123: and detecting whether each slice contains a referee key frame of the referee target object. If the detected referee key frame is marked;
specifically, in an exemplary embodiment, the referee key frames of the referee target object include player foul behavior key frames and sphere score key frames, and the slicing the historical race data includes: the historical race data is sliced based on the integrity of the player's foul behavior and the integrity of the sphere scoring process. For example, in a badminton game, the key frames for player foul actions are player touch screen key frames, ball touch body key frames, etc. The sphere scoring key frame comprises a key frame that the shuttlecock falls to the ground within the scoring boundary; basketball is a key frame that falls into the string bag.
S124: extracting a target slice containing the referee key frame;
and extracting the slice with the index based on the index of the judge key frame, and storing the slice as a target slice.
S125: performing frame-by-frame gray level conversion on the target slice to obtain a target slice after gray level processing; since the images of the target slice are all color images, if training is performed by directly using the color images, RGB multichannel processing is required, which tends to increase the calculation amount. Therefore, before target detection and target tracking are performed, the target slice is subjected to frame-by-frame gray level conversion so as to reduce the data processing amount and improve the calculation efficiency.
S126: performing target detection and target tracking on the target slice subjected to the gray processing to obtain a motion track of a target object to be judged; and indexing the judge result of the target slice based on the motion trail of the target object to be judged and the international judge rule so as to obtain the historical data to be judged.
Please refer to fig. 3; because the race condition data analysis submodel needs to complete tasks, the target slice is subjected to target detection and target tracking through video analysis; thus, in an exemplary embodiment, the method for constructing the race data analysis sub-model specifically includes: constructing a spatial information learning model based on a fast R-CNN model; constructing a time sequence information learning model based on an RNN model; the time sequence information learning model takes the space information learning model as a pre-model, and builds the race condition data analysis sub-model based on an integration algorithm. The spatial information learning model performs image segmentation and target detection on the target slice; the time sequence information learning model takes the output of the spatial information learning model as the input of the time sequence information learning model, so that the target object to be judged of the target slice and the motion trail of the target object to be judged are obtained.
Specifically, in an exemplary embodiment, the method for constructing the spatial information learning model includes: firstly, pre-training the Faster R-CNN model based on an ImageNet data set to obtain a pre-trained Faster R-CNN model; and taking the historical data to be refereed as a training set of the Faster R-CNN model, and performing fine adjustment on the Faster R-CNN model to obtain the spatial information learning model based on the training set of the ball game.
And then extracting the image characteristics of the object to be judged from the target slice through the Faster R-CNN convolutional neural network. These features are passed to a candidate region generation network (Region Proposal Network) which extracts candidate regions containing objects to be refereed in the frame image. Each candidate region is then cropped and scaled to convert them into a fixed-size feature map. These feature maps are then input to a classifier and regressor with a multi-layer perceptron to classify candidate regions and to a bounding box regression to detect target objects to be refereed.
Thereafter, target tracking is performed based on the detected target object to be refereed, and the detected target object to be refereed is tracked based on a motion information detection algorithm between adjacent frames. For example, a motion vector between a previous frame and a current frame is acquired using a light flow algorithm and matched to an object bounding box to be refereed. And tracking the target object to be refereed among different frames by utilizing the matching result, and predicting the position of the target object.
Then, taking the feature vectors of the target detection result and the target tracking result as the input of the time sequence information learning model, and predicting the target slice; in another exemplary embodiment, the training method of the time series information learning model includes: and training the time sequence information learning model by taking the historical race condition data as a training set of the time sequence information learning model so as to obtain the time sequence information learning model based on the ball race training set. Therefore, the fitting capability of the time sequence information learning model to the complete time sequence of the ball game is improved, and the time sequence understanding of the slice to be detected after the time sequence information learning model is used as an integrated model is improved.
Finally, after each initial sub-model is built, training the model using a training set and testing the referee effect of the model using a testing set is also included. In addition, a back propagation algorithm is adopted to optimize parameters of the model so as to extract characteristics of the moving object to the greatest extent; and finally, outputting the adjusted race condition data analysis sub-model.
In FIG. 6, the electronic device 4000 includes at least one processor 4001, at least one communication bus 4002, and at least one memory device 4003.
Wherein the processor 4001 is coupled to the memory 4003, such as via a communication bus 4002. Optionally, the electronic device 4000 may further comprise a transceiver 4004, the transceiver 4004 may be used for data interaction between the electronic device and other electronic devices, such as transmission of data and/or reception of data, etc. It should be noted that, in practical applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 is not limited to the embodiment of the present application.
The processor 4001 may be a CPU (Central Processing Unit ), general purpose processor, DSP (Digital Signal Processor, data signal processor), ASIC (Application Specific Integrated Circuit ), FPGA (Field Programmable Gate Array, field programmable gate array) or other programmable logic device, transistor logic device, hardware components, or any combination thereof. Which may implement or perform the various exemplary logic blocks, modules and circuits described in connection with this disclosure. The processor 4001 may also be a combination that implements computing functionality, e.g., comprising one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.
The communication bus 4002 may include a pathway to transfer information between the aforementioned components. The communication bus 4002 may be a PCI (Per ipheral Component Interconnect, peripheral component interconnect standard) bus or an EISA (Extended Industry Standard Architecture ) bus, or the like. The communication bus 4002 can be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, only one thick line is shown in fig. 6, but not only one bus or one type of bus.
The Memory 4003 may be, but is not limited to, ROM (Read 0nly Memory) or other type of static storage device that can store static information and instructions, RAM (Random Access Memory ) or other type of dynamic storage device that can store information and instructions, EEPROM (Electr ically Erasable Programmable Read Only Memory, electrically erasable programmable Read-only Memory), CD-ROM (Compact Disc Read Only Memory, compact disc Read-only Memory) or other optical disc storage, optical disc storage (including compact disc, laser disc, optical disc, digital versatile disc, blu-ray disc, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer.
The memory 4003 has stored thereon a computer program, and the processor 4001 reads the computer program stored in the memory 4003 through the communication bus 4002.
The computer program, when executed by the processor 4001, implements the sports automatic grouping method in the above embodiments.
In summary, the technical solutions provided in the embodiments of the present application construct an intelligent referee prediction model, capture a field video with the help of a high-speed camera, and transmit the video to a rear-end intelligent referee prediction model, so as to implement referee result prediction on ball events by a machine learning method, thereby reducing subjective factor interference caused by artificial referees in the prior art, and thus reducing the false judgement rate. The scoring rule base is perfected by introducing international referee rules and expert experience so as to cope with various irregular referee scenes on the competition field, and the generalization capability of the model and the referee prediction accuracy are improved. Slicing the historical race condition data based on the integrity of the player foul behaviors and the integrity of the sphere scoring process, screening out slices containing key frames, performing target detection and target tracking on the screened target slices, and marking the target slices to serve as a model training set, so that data with low information are discarded, the data processing amount and the data training amount are compressed, and the model training efficiency is improved; meanwhile, the overfitting of the model can be avoided, and the generalization capability of the model is improved. By decomposing the intelligent referee prediction model into the race condition data analysis sub-model and the score decision sub-model, correspondingly, decomposing the referee task into two sub-tasks, namely a target detection task and a score decision task, compared with the single model, the method has the advantages that the integral referee task is realized, the construction difficulty of the intelligent referee prediction model is reduced, and the construction efficiency is improved for quickly constructing the intelligent referee prediction model. The historical race condition data is used as a training set of the time sequence information learning model, and the time sequence information learning model is trained, so that the fitting capacity of the time sequence information learning model to the complete time sequence of the ball race is improved, and the time sequence understanding of the slice to be tested after the time sequence information learning model is used as an integrated model is improved.
It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited in order and may be performed in other orders, unless explicitly stated herein. Moreover, at least some of the steps in the flowcharts of the figures may include a plurality of sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, the order of their execution not necessarily being sequential, but may be performed in turn or alternately with other steps or at least a portion of the other steps or stages.
The foregoing is only a partial embodiment of the present application, and it should be noted that it will be apparent to those skilled in the art that modifications and adaptations can be made without departing from the principles of the present application, and such modifications and adaptations are intended to be comprehended within the scope of the present application.
Claims (10)
1. An intelligent referee method for ball events, comprising:
pre-constructing an intelligent referee prediction model;
acquiring field competition condition data, and performing data processing on the field competition condition data to generate data to be judged; the on-site competition condition data comprise competition condition videos shot by a high-speed camera;
inputting the data to be refereed into the intelligent referee prediction model to obtain a score discrimination result of the intelligent referee prediction model on the data to be refereed;
judging whether the score judging result accords with the international judging rule of the ball event;
if the score is not matched with the score, correcting the score judgment result through an error correction inlet to generate an error correction example;
and collecting the error correction examples for optimizing the intelligent referee prediction model.
2. The intelligent referee method for a ball game of claim 1, wherein: the construction method of the intelligent referee prediction model comprises the following steps:
constructing an initial model of an intelligent referee prediction model;
acquiring historical competition condition data of the ball competition, and performing data processing on the historical competition condition data to acquire historical data to be judged;
acquiring international judge rules and expert experiences of the ball events, and constructing a scoring rule base;
indexing the historical data to be refereed based on the scoring rule base to obtain a training data set;
and inputting the training data set into the intelligent referee prediction model for training and debugging so as to obtain the intelligent referee prediction model.
3. The intelligent referee method for ball game according to claim 2, wherein said obtaining historical game condition data of said ball game, and performing data processing on said historical game condition data to obtain historical referee data, comprises:
determining a target object to be referee of the ball game according to the international referee rule of the ball game;
slicing the historical race condition data based on the motion trail of the target object to be refereed; the slice comprises a plurality of frames of consecutive video images;
detecting whether each slice contains a referee key frame of the referee target object;
extracting a target slice containing the referee key frame;
performing frame-by-frame gray level conversion on the target slice to obtain a target slice after gray level processing;
performing target detection and target tracking on the target slice subjected to the gray processing to obtain a motion track of a target object to be judged;
and indexing the judge result of the target slice based on the motion trail of the target object to be judged and the international judge rule so as to obtain the historical data to be judged.
4. The intelligent referee method for ball game according to claim 3, wherein the referee key frames of the referee target object include player foul behavior key frames and sphere score key frames, said slicing the historical game data comprising: the historical race data is sliced based on the integrity of player foul behaviors and the integrity of the sphere scoring process.
5. The intelligent referee method for a ball game of claim 2, wherein: the construction of the initial model of the intelligent referee prediction model specifically comprises the following steps:
creating a race condition data analysis sub-model for carrying out target detection and target tracking on the data to be refereed; the method comprises the steps of,
creating a score decision sub-model for scoring and judging the data to be judged;
and taking the race condition data analysis sub-model as a front model of the score decision sub-model, and carrying out model integration on the race condition data analysis sub-model and the score decision sub-model to generate an initial model of an intelligent referee prediction model.
6. The intelligent referee method for ball game according to claim 5, wherein the construction method for the game condition data analysis sub-model comprises the following steps:
constructing a spatial information learning model based on a fast R-CNN model; the method comprises the steps of,
constructing a time sequence information learning model based on an RNN model;
and the time sequence information learning model takes the space information learning model as a pre-model, and builds the race condition data analysis sub-model based on an integration algorithm.
7. The intelligent referee method for ball game according to claim 6, wherein the method for constructing the spatial information learning model comprises the following steps:
pre-training the fast R-CNN model based on an ImageNet data set to obtain a pre-trained fast R-CNN model;
and taking the historical data to be refereed as a training set of the Faster R-CNN model, and performing fine adjustment on the Faster R-CNN model to obtain the spatial information learning model.
8. The method of intelligent referee to a ball game of claim 6, wherein the training method of the time series learning model comprises:
and taking the historical race condition data as a training set of the time sequence information learning model, and training the time sequence information learning model.
9. The method for intelligently judging a ball game according to claim 5, wherein the method for constructing the molecular model comprises the following steps:
constructing an initialization model of the score decision sub-model based on an SVM model;
and training and debugging the SVM model by taking the training data set as a training sample of the score decision sub-model so as to obtain the score decision sub-model.
10. An electronic device, comprising: at least one processor, at least one memory, and at least one communication bus, wherein the memory has a computer program stored thereon, the processor reading the computer program in the memory through the communication bus; the computer program, when executed by the processor, implements a method of intelligent referee to a ball game as claimed in any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310753667.8A CN116720784A (en) | 2023-06-26 | 2023-06-26 | Intelligent judging method for ball event and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310753667.8A CN116720784A (en) | 2023-06-26 | 2023-06-26 | Intelligent judging method for ball event and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116720784A true CN116720784A (en) | 2023-09-08 |
Family
ID=87871346
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310753667.8A Pending CN116720784A (en) | 2023-06-26 | 2023-06-26 | Intelligent judging method for ball event and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116720784A (en) |
-
2023
- 2023-06-26 CN CN202310753667.8A patent/CN116720784A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11544928B2 (en) | Athlete style recognition system and method | |
| Huang et al. | Tracknet: A deep learning network for tracking high-speed and tiny objects in sports applications | |
| Yuan et al. | Temporal action localization by structured maximal sums | |
| CN110267119B (en) | Video precision and chroma evaluation method and related equipment | |
| CN104679818B (en) | A kind of video key frame extracting method and system | |
| CN110858394A (en) | Image quality evaluation method and device, electronic equipment and computer readable storage medium | |
| CN112132119A (en) | Passenger flow statistical method and device, electronic equipment and storage medium | |
| Ghosh et al. | Towards structured analysis of broadcast badminton videos | |
| CN108280421B (en) | Human behavior recognition method based on multi-feature depth motion map | |
| CN104065872B (en) | Moving image extraction element, moving image extracting method and recording medium | |
| KR101844874B1 (en) | System for predicting decision using big data based on data mining techniques and method therefor | |
| Das et al. | Deep transfer learning-based foot no-ball detection in live cricket match | |
| CN111340019A (en) | Grain bin pest detection method based on Faster R-CNN | |
| CN111986163A (en) | Face image selection method and device | |
| Zhang et al. | Efficient golf ball detection and tracking based on convolutional neural networks and kalman filter | |
| Iyer et al. | Automated third umpire decision making in cricket using machine learning techniques | |
| US11908190B2 (en) | Game monitoring | |
| CN113688804B (en) | Multi-angle video-based action identification method and related equipment | |
| CN116720784A (en) | Intelligent judging method for ball event and electronic equipment | |
| CN116186561A (en) | Running gesture recognition and correction method and system based on high-dimensional time sequence diagram network | |
| CN116012417A (en) | Track determination method and device of target object and electronic equipment | |
| CN113537168A (en) | Basketball goal detection method and system for rebroadcasting and court monitoring scene | |
| Patton et al. | Predicting nba talent from enormous amounts of college basketball tracking data | |
| Zhang et al. | Application of optimized BP neural network based on genetic algorithm in rugby tackle action recognition | |
| CN110490064B (en) | Sports video data processing method and device, computer equipment and computer storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |