KR102450432B1 - A method for detecting sports events and system performing the same - Google Patents

Abstract

The present invention relates to a sports event detection method and a system for performing the same, comprising the steps of: acquiring image data of a specific sports game image; Acquiring a plurality of online text data related to the specific sports game image, wherein the online text data includes a writing time that is information about a time when the online text data is generated and a writing content related to content information of the online text data ; Using an artificial neural network trained using a running set in which the analysis result of the writing time and writing content included in the online text data is labeled in image data related to a plurality of sports events acquired in advance, the plurality of It relates to a method of generating a sports highlight image, including; extracting sports event data related to at least one sports event reflected by the text data of .

Description

A METHOD FOR DETECTING SPORTS EVENTS AND SYSTEM PERFORMING THE SAME

The present invention relates to a method for detecting a sports event and a system for performing the same, and more particularly, to a method for generating sports highlight data using data obtained from a sports field and data obtained indirectly from outside the sports field, and a method for performing the same It's about the system.

In the existing sports industry, sports highlight videos are directly generated by humans at broadcasters, resulting in serious waste of manpower, and highlight scenes are selected based on the subjective criteria of the highlight creators, which does not satisfy the various needs of viewers. did.

Therefore, in the modern sports industry, there is an increasing demand for a technology for generating a sports highlight image including only major sports scenes that occurred within a corresponding game using an automated highlight generation method, as well as a technology that responds to the individual tastes of viewers. Demand for generating highlight images is also increasing.

One problem to be solved by the present invention is to provide a system for detecting a sporting event for generating a highlight image including only a main scene occurring in a sporting event,

SUMMARY OF THE INVENTION An object of the present invention is to increase the efficiency of sports highlight generation by generating sports highlights using information on Internet posts that can be easily obtained.

According to an aspect of the present invention, there is provided a method for generating a sports highlight image, the method comprising: acquiring image data in which a specific sports game image is captured; Acquiring a plurality of online text data related to the specific sports game image, wherein the online text data includes a writing time that is information about a time when the online text data is generated and a writing content related to content information of the online text data ; Using an artificial neural network trained using a running set in which the analysis result of the writing time and writing content included in the online text data is labeled in image data related to a plurality of sports events acquired in advance, the plurality of Extracting sports event data related to at least one or more sports events reflected by the text data of ;

The method further includes: analyzing the written content included in the plurality of online text data to calculate a sentiment index for the plurality of online texts; A method of generating a sports highlight image that is learned using the correlation between data and the emotional index may be provided.

Further, the method further comprises; determining the amount of the online text data written for a predetermined time based on the writing time, wherein the artificial neural network includes image data related to the plurality of pre-obtained sports events and the online text data. A method of generating a sports highlight image, which is learned based on a positive correlation of , may be provided.

The method further includes: based on the writing time, calculating a time difference between the occurrence time of the sports event and the writing time of the online text; A method of generating a sports highlight image that is learned based on a correlation between data and the time difference information may be provided.

In addition, the sports event data is tagged with a characteristic index reflecting the importance of the sports event for the sports event, and based on the characteristic index, at least one or more for generating a highlight image from the plurality of sports event data. A method of generating a sports highlight image further comprising; extracting sports event data may be provided.

In addition, the image data includes a first image photographed at a first angle and a second image photographed at a second angle different from the first angle, wherein the sports event data reflects the type of the sports event, The method of generating a sports highlight image further comprising the step of selecting at least one of the first image and the second image to reflect the information on the sports event in consideration of the type of event may be provided.

According to another aspect of the present invention, there is provided a method for generating a sports highlight image, the method comprising: acquiring base sports data including at least one or more image data in which an image of a target sports game is captured; obtaining at least one or more reference sports data related to the target sports event, the reference sports data including acoustic data related to the target sports event; Normalizing the intensity of the sound signal included in the sound data in consideration of the target characteristics reflecting the characteristics of the target sports game; detecting a sports event by analyzing the reference sports data including the normalized acoustic data; and generating highlight data by extracting an image corresponding to the sporting event from the base sports data.

Also, the method for generating a sports highlight image may be provided, wherein the target characteristic is a home team or an away team of a sports team playing the target sports game.

In addition, the method for generating a sports highlight image may be provided, wherein the target characteristic is a change in sound signal strength according to a competition time of the target sports competition.

Also, the reference sports data may be provided with a method of generating a sports highlight image including at least one of location data, dynamic data, and online text data.

In addition, the base sports data further includes at least one of location data, dynamic data, and online text data, and the highlight data is at least any one of the location data, dynamic data, and online text data extracted to correspond to the sports event. A method of generating a sports highlight image further including one part may be provided.

According to the present invention, by generating a highlight image based on a feature score for a sporting event occurring in a sporting event, it is possible to generate a highlighting image including only major scenes occurring in the sporting event.

According to the present invention, by detecting a sports event by using the text of the Internet posting as input data of the artificial neural network, it is possible to detect the sports event from the easily obtained Internet posting.

1 schematically shows an embodiment of a data analysis system according to an embodiment.
2 is a block diagram schematically illustrating a configuration of a data acquisition system according to an embodiment.
3 is a block diagram schematically illustrating a configuration of a data analysis system according to an embodiment.
4 is a block diagram schematically illustrating an operation of generating highlight data of a data analysis system according to an exemplary embodiment.
5 is a flowchart illustrating an operation of generating highlight data of a data analysis system according to an exemplary embodiment.
6 shows an example of sports data in an image data format.
7 shows an example of sports data in an acoustic data format.
8 shows an example of sports data in a heat map data format.
9 shows an example of sports data in a dynamic data format.
10 shows an example of sports data obtained from an online web server.
11 is a flowchart schematically illustrating an operation of directly analyzing sports data by a data analysis system according to an exemplary embodiment.
12 is a diagram illustrating an example of a correspondence relationship between a sound level included in sports data and a sports event according to an exemplary embodiment.
13 is a diagram illustrating another example of a correspondence between a sound level included in sports data and a sports event according to an exemplary embodiment.
14 illustrates another example of a correspondence between a sound level included in sports data and a sports event according to an exemplary embodiment.
15 is a diagram illustrating an example of a correspondence relationship between a sound factor related to a type of sound and a sport event according to an exemplary embodiment.
16 illustrates an example of a correspondence relationship between an image factor included in image data and a sports event according to an exemplary embodiment.
17 illustrates an example of a correspondence relationship between a heat map factor included in heat map data and a sports event according to an exemplary embodiment.
18 illustrates an example of a correspondence relationship between factors included in dynamic data and a sports event according to an embodiment.
19 is a diagram illustrating an example in which a data analysis system directly uses sports data to calculate a feature score for a sports event according to an embodiment.
20 is a schematic flowchart of an indirect sports data analysis operation according to an embodiment.
21 is a schematic flowchart of an operation of the data analysis system filtering text included in indirect sports data.
22 shows an example of indirect sports data processing related to text filtering and calculation of a sentiment index.
23 is a schematic flowchart of an operation in which the data analysis system detects an increase moment of text included in indirect sports data.
24 shows an example of indirect sports data processing related to text increase index calculation.
25 is a flowchart of an operation of the data analysis system processing indirect sports data through text clustering.
26 illustrates an example of text clustering according to an embodiment.
27 is an exemplary diagram of an artificial neural network according to an embodiment.
28 is an example of a sports event detection output result using an artificial neural network according to an embodiment.
29 is a schematic flowchart of an operation of generating highlight data of a data analysis system according to an exemplary embodiment.
30 illustrates an example of highlight data according to an embodiment.
31 illustrates another example of highlight data according to an embodiment.

The above-mentioned objects, features and advantages of the present invention will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. However, since the present invention may have various changes and may have various embodiments, specific embodiments will be exemplified in the drawings and described in detail below.

In the drawings, the thicknesses of layers and regions are exaggerated for clarity, and also that an element or layer is “on” or “on” another component or layer. What is referred to includes all cases in which another layer or other component is interposed in the middle as well as directly on top of another component or layer. Throughout the specification, like reference numerals refer to like elements in principle. In addition, components having the same function within the scope of the same idea shown in the drawings of each embodiment will be described using the same reference numerals.

If it is determined that a detailed description of a known function or configuration related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted. In addition, numbers (eg, first, second, etc.) used in the description process of the present specification are merely identification symbols for distinguishing one component from other components.

In addition, the suffixes "module" and "part" for the components used in the following description are given or mixed in consideration of only the ease of writing the specification, and do not have distinct meanings or roles by themselves.

The present specification relates to a method for detecting a sports event by analyzing at least one or more sports data, and a system for performing the same.

Here, sports may be variously interpreted as all sports in which an object capable of acquiring data and an event that can be detected by analyzing the acquired data exist. That is, sports in the present specification includes, for example, both individual sports and team sports, and includes both ball sports and non-ball sports, sports using sports instruments and non-use sports, professional sports and amateur sports. It means the sport of the human concept. However, in the present specification, for convenience of explanation, the description will be focused on soccer, but the idea of the present specification is not limited to soccer.

In the following drawings, a data analysis system according to an embodiment will be schematically described, and terms to be used in the following description of the present specification will be summarized with reference to the drawings. However, the terms used in this specification are only for convenience of description, and the spirit of the present specification is not limited by these terms.

1 schematically shows an embodiment of a data analysis system according to an embodiment.

Referring to FIG. 1 , a data analysis system according to an embodiment may acquire sports data related to sports and analyze the acquired data to detect a sports event.

In this specification, sports data should be interpreted in a comprehensive sense including all types of data related to sports. That is, sports data includes not only sports data directly related to sports such as image data of sports events, acoustic data recording the sounds of sports events, location data of sports participants, and dynamic data of sports participants, but also game logs of Internet sports relay broadcasting. , texts, and data that can be indirectly obtained online, such as sports-related SNS texts, may also be included. A detailed example of the aforementioned sports data will be described later in detail.

Referring back to FIG. 1 , the data analysis system 2000 according to an embodiment acquires sports data from at least one of the data acquisition system 1000 and the online server 3000 and analyzes it to detect a sports event. can

In this specification, the data acquisition system 1000 and/or the online web server 3000 may form a single system with the data analysis system 2000 . Alternatively, the data acquisition system 1000 and/or the online web server 3000 may be implemented as a separate system from the data analysis system 2000 . In the following description of the present specification, examples in which the data acquisition system 1000, the data analysis system 2000, and the online web server 3000 are implemented as separate systems will be mainly described, but this is for convenience of description. , the spirit of the present specification is not limited thereto.

Here, the data acquisition system 1000 may directly acquire sports data in a field where sports are performed. Here, the field in which sports are performed is a concept including all sports fields such as professional sports events, amateur sports competitions, professional sports training, and amateur sports training. explained as However, it will be understood that the description herein is not limited to professional sports events.

As an example, the data acquisition system 1000 may be implemented through an Electrical Performance Tracking System (EPTS). Specifically, the data acquisition system 1000 may acquire sports data through a sensor installed in a field where sports are performed. More specifically, the data acquisition system 1000 may acquire sports data including an image of sports through at least one image sensor 1030 installed in a field where sports are performed. In addition, the data acquisition system 1000 may acquire sports data including the sports sound through at least one or more acoustic sensors 1040 installed at the site where sports are performed.

Also, the data acquisition system 1000 may acquire sports data directly related to a sports participant. Specifically, the data acquisition system 1000 may acquire sports data through a sensor attached to a sports participant, It is also possible to obtain sports data. For example, the data acquisition system 1000 may acquire location data of a sports participant using the GPS 10 . Specifically, location data of the sports participant may be acquired through the GPS sensor 1010 attached to the sports participant. In addition, the data acquisition system 1000 may acquire the location data of the sports participant through the near field communication (UWB) system 12 . Also, the data acquisition system 1000 may acquire dynamic data of the sports participant through the dynamic sensor 1050 attached to the sports participant. Here, the dynamic sensor may be implemented based on an IMU sensor such as an accelerometer or a gyrometer, and the dynamic data may include velocity, acceleration, angular velocity, angular acceleration, and the like.

That is, the data acquisition system 1000 may acquire sports data in various ways in the field where sports are played, and may transmit the acquired sports data to the data analysis system 2000 .

Referring back to FIG. 1 , the online web server 3000 may acquire sports data outside the field where sports are played. The online web server 3000 may acquire content related to sports written by a plurality of users online, which is used by a plurality of users. For example, the online web server 3000 may acquire texts or posts written by viewers who watch an Internet sports broadcast in which sports data transmitted through the data acquisition system 1000 are broadcast. As another example, the online web server 3000 may acquire sports related posts or texts on a social network service (SNS).

That is, the online web server 3000 may acquire sports data from a plurality of users online in various ways, and transmit the acquired sports data to the data analysis system 2000 .

The data analysis system 2000 may analyze sports data acquired through the data acquisition system 1000 and/or the online web server 3000 and detect a sports event from the sports data.

Hereinafter, schematic configurations of the data acquisition system 1000 and the data analysis system 2000 according to an embodiment of the present specification will be described with reference to the drawings.

2 is a block diagram schematically illustrating a configuration of a data acquisition system according to an embodiment.

Referring to FIG. 2 , the data acquisition system 1000 according to an embodiment includes a position sensor 1010 , an image sensor 1030 , an acoustic sensor 1040 , a dynamic sensor 1050 , a first communication module 1400 , It may include a first memory 1200 and a first controller 1002 .

According to an embodiment, the data acquisition system 1000 may acquire sports data through at least one sensor and transmit the acquired sports data to the data analysis system 2000 . Specifically, the first controller 1002 transmits sports data obtained through the position sensor 1010 , the image sensor 1030 , the acoustic sensor 1040 and/or the dynamic sensor 1050 to the first communication module 1400 . through the data analysis system 2000 .

Hereinafter, each configuration of the data acquisition system 1000 according to an embodiment will be described.

According to an embodiment, the image sensor 1030 may acquire image data related to a sports game. Here, at least one image sensor 1030 may be installed in a stadium where a sports game is played, and through this, image data taken from various angles for the same scene may be acquired. The image sensor 1030 may acquire image data related to a sports event and transmit the acquired image data to the first controller 1002 .

According to an embodiment, the location sensor 1010 may measure location data of a sports participant. Specifically, the position sensor 1010 may measure the position data of the sports participant and transmit the measured position data to the first controller 1002 . Here, the position sensor 1010 may be implemented in various forms. For example, the position sensor 1010 may be implemented as a GNSS antenna to detect a GNSS signal of the player. As another example, the position sensor 1010 may be implemented as a UWB receiving module to detect a UWB signal of a sports participant.

According to an embodiment, the dynamic sensor 1050 may measure dynamic data of a sports participant. Specifically, the dynamic sensor 1050 may measure dynamic data of a sports participant and transmit the measured dynamic data to the first controller 1002 . Here, the dynamic data may be data related to the sports participant's motion, such as speed, acceleration, jerk, and the like of the sports participant. For example, the dynamic sensor 1050 may be implemented as an accelerometer to measure the acceleration of a sports participant. As another example, the dynamic sensor 1050 may be implemented as a gyrometer. As another example, the dynamic sensor 1050 may be implemented as a magnetometer.

The first communication module 1400 according to an embodiment may communicate with an external system, an external device, or an external server. The data acquisition system 1000 may perform data communication between the first communication module 1400 and the data analysis system 2000 . For example, the data acquisition system 1000 may transmit sports data acquired from a plurality of sensor modules to the data analysis system 2000 or an external device by using the first communication module 1400 . Here, the data acquisition system 1000 may transmit sports data in real time or may transmit sports data after a preset time has elapsed. Also, the data acquisition system 1000 may transmit at least one piece of sports data individually or collectively.

Also, the first controller 1002 may acquire dynamic data of the sports player measured by the dynamic sensor 1050 and transmit the dynamic data acquired through the first communication module 1400 to the data analysis system 2000 .

Also, the first controller 1002 may obtain position data of the sports player measured through the position sensor 1220 , and process the obtained position data to generate dynamic data.

The first controller 1002 may control the overall operation of the data acquisition system 1000 . For example, the first controller 1002 may receive position data from the position sensor 1220 and generate a control signal to transmit it to the data analysis system 2000 through the first communication module 1400 . As another example, the first controller 1002 may receive dynamic data from the dynamic sensor 1050 and generate a control signal to transmit it to the data analysis system 2000 through the first communication module 1400 .

The data acquisition system 1000 may have a separate power supply unit, or may receive power from the outside by wire or wirelessly, and may have a switch for controlling the power supply unit.

3 is a block diagram schematically illustrating a configuration of a data analysis system according to an embodiment.

Referring to FIG. 3 , the data analysis system 2000 may include a second controller 2002 , a second memory 2200 , and a second communication module 2400 .

According to an embodiment, the second controller 2002 acquires sports data from the data acquisition system 1000 through the second communication module 2400 and uses the analysis algorithm stored in the memory 2200 to transmit the sports data. It can be analyzed to detect sporting events. Here, the data analysis system 2000 may acquire sports data from the data acquisition system 1000 in real time and detect a sports event in real time.

Hereinafter, each configuration of the data analysis system 2000 according to an embodiment will be described.

The second memory 2200 may store various types of information of the data analysis system 2000 .

In the second memory 2200 , various data necessary for the operation of the data analysis system 2000 , including an operation program for driving the data analysis system 2000 or a program for operating each component of the data analysis system 2000 , are stored in the second memory 2200 . can be saved. For example, the memory 2200 may store a machine learning algorithm for detecting a sports event from sports data as will be described later.

The second communication module 2400 may communicate with an external device or an external server. The data acquisition apparatus 2000 may perform data communication with the data acquisition system 1000 or an external server using the second communication module 2800 . For example, the data analysis system 2000 may acquire data necessary for evaluating a sports player from the data acquisition system 1000 using the second communication module 2400 .

The second controller 2002 may control the overall operation of the data analysis system 2000 . For example, the second controller 2002 loads a program for analyzing sports data from the memory 2200, processes and analyzes data obtained from the data acquisition system 1000, and transmits the result to the second communication module 2400 ) to generate a control signal to be provided to an external device or an external server. A detailed sports event detection operation performed by the data analysis system 2000 will be described later in detail.

Also, the data analysis system 2000 may further include a separate output unit for outputting a sports event detection result. Here, the data second controller 2002 may process and analyze the sports data acquired from the data acquisition system 1000 and generate a control signal for providing the result through an output unit.

Also, the data analysis system 2000 may be implemented as a server. However, for convenience of explanation, in the following description of the present specification, the data analysis system 2000 will be mainly described, but it will be apparent that it may be implemented as a data analysis server.

In the above, the configuration of the data analysis system 2000 according to an embodiment has been described. Hereinafter, a sports event detection operation implemented in the data analysis system 2000 according to an embodiment will be described with reference to the drawings.

4 is a block diagram schematically illustrating an operation of generating highlight data of a data analysis system according to an exemplary embodiment.

Referring to FIG. 4 , the data analysis system 2000 according to an exemplary embodiment may acquire at least one or more sports data to detect a sports event, and output highlight data based on the detected result. Here, sports data including information on at least one specific sport event among a plurality of sport events detected by the data analysis system 2000 may be referred to as highlight data, and in the following description of the present specification, for convenience of explanation For this purpose, use the term 'highlight data'. That is, the highlight data may be an output result of the data analysis system 2000 according to an embodiment of the present specification.

In addition, according to an exemplary embodiment, at least one or more sports data transmitted to the data analysis system 2000 may be classified according to a purpose used in the data analysis system 2000 . Specifically, it may be divided into sports data that is a base for extracting highlight data to be output from the data analysis system 2000 and sports data that is an analysis target for sports event detection. Illustratively, the image data obtained by the data analysis system 2000 from the data acquisition system 1000 is used as data as a base of the highlight data, and the SNS obtained by the data analysis system 2000 from the online web server 3000 When the text is used as sports data to be analyzed for the detection of a sports event, the data analysis system 2000 may generate a sports highlight image extracted from the image data based on the analysis result of the SNS text. will be. In the following description of the present specification, for convenience, sports data used as a base of highlight data will be referred to as 'base sports data', and sports data to be analyzed for detection of a sports event will be referred to as 'reference sports data'. shall be referred to as

In addition, it should be noted that although data obtained from the data acquisition system 1000 are used as base sports data and data obtained from the online web server 3000 are used as reference sports data in the drawings, this is only an example. That is, the base sports data and the reference sports data are classified according to the purpose used in the data analysis system 2000 , and do not depend on the subject that acquires the sports data. For example, image data acquired in the data acquisition system 1000 may be used as base sports data, and sound data acquired in the data acquisition system 1000 may also be used as reference sports data.

Also, the data analysis system 2000 may use the same sports data as both the base sports data and the reference sports data. For example, the data analysis system 2000 may use the image data as both the base sports data and the reference sports data.

According to an embodiment, rather than detecting a sports event using only one sports data, by analyzing at least one or more reference sports data in which a specific content about a sports event to be extracted is well expressed, it is more accurate from the base sports data In addition, it is possible to generate highlight data that reflects the sporting event required by the viewer well.

5 is a flowchart illustrating an operation of generating highlight data of a data analysis system according to an exemplary embodiment.

Referring to FIG. 5 , the sports event detection operation according to an exemplary embodiment includes acquiring at least one or more sports data (S1000), analyzing the acquired sports data (S1200), and generating highlight data based on the analysis result. It may include a step (S1400) of doing.

First, in the sports data acquisition step ( S1000 ), the data analysis system 2000 may acquire sports data. Specifically, the data analysis system 2000 may acquire at least one or more sports data from the data acquisition system 1000 and/or the online web server 3000 . Specific examples of at least one or more sports data will be described later in detail.

After the sports data acquisition step S1000, the data analysis system 2000 may analyze the acquired sports data to detect a sports event from the sports data. Specifically, the second controller 2002 may analyze at least one or more sports data using an algorithm for data analysis stored in the second memory 2200 and detect at least one or more sports events from the sports data. More specifically, the second controller 2002 may analyze the reference sports data and detect at least one or more sports events from the reference sports data. More details regarding the sports data analysis step ( S1200 ) will be described later in detail.

After the sports data analysis step ( S1200 ), the data analysis system 2000 may acquire highlight data from the sports data based on the analysis result ( S1400 ). Specifically, the second controller 2002 , the base sports data Highlight data may be obtained by extracting data related to at least one sports event detected in the sports data analysis step ( S1200 ) from the . Details of the highlight sports data acquisition step will be described later in detail.

In the above, a schematic sports event detection operation performed in the data analysis system 2000 has been described.

Hereinafter, an example of sports data used for sports event detection in the present specification will be first looked at through FIGS. 6 to 10 , and then, an operation of the data analysis system 2000 detecting a sports event from sports data will be described in detail. let's take a look

6 to 10 show various examples of sports data according to an embodiment.

Specifically, FIG. 6 shows an example of sports data in an image data format, FIG. 7 shows an example of sports data in an acoustic data format, and FIG. 8 shows an example of sports data in a heat map data format. , FIG. 9 is an example of sports data in a dynamic sensor data format. Also, FIG. 10 shows an example of sports data implemented as an Internet sports relay screen.

Sports data according to an embodiment may also be distinguished according to an acquisition subject. Specifically, sports data may be divided into direct sports data directly obtained from a sports event and indirect sports data that is not directly related to sports data, but includes content related to sports indirectly.

That is, examples of sports data shown in FIGS. 6 to 9 are examples of direct sports data, and FIG. 10 may be expressed as showing an example of indirect sports data.

6 shows an example of sports data in an image data format.

Referring to FIG. 6 , the data analysis system 2000 according to an embodiment may acquire highlight data using image data. Specifically, the second controller 2002 may acquire the highlight data using the image data as the base sports data. Also, the second controller 2002 may acquire highlight data by using the image data as reference sports data.

Also, the data analysis system 2000 may acquire image data through the data acquisition system 1000 . Also, the data analysis system 2000 may acquire image data through the online web server 3000 . Specifically, image data acquired through the data acquisition system 1000 may be first transmitted to the online web server 3000 , and image data transmitted to the online web server 3000 may be transmitted to the data analysis system 2000 . .

In addition, the image data may include content related to a sports game photographed from various angles. Specifically, according to the number of cameras included in the data acquisition system 1000 that has acquired the image data or the installation composition of the cameras, the image data may include images related to sports events captured from various angles.

7 shows an example of sports data in an acoustic data format.

Referring to FIG. 7 , the data analysis system 2000 according to an embodiment may acquire highlight data using sound data. Since the operation of the data analysis system 2000 to obtain highlight data using the sound data may be performed similarly to the operation of the data analysis system 2000 to obtain the highlight data using the image data, a detailed description will be omitted. let it do

Also, like image data, the sound data may include sound data related to sports events recorded at various places in the stadium according to the number or installation composition of the sound sensors installed in the stadium. In addition, the sound data may be included in the image data or may be extracted from the video data.

8 shows an example of sports data in a heat map data format.

Referring to FIG. 8 , the data analysis system 2000 according to an embodiment may acquire highlight data using heat map data. Specifically, the second controller 2002 may acquire the highlight data using the heat map data as the base sports data. Also, the second controller 2002 may acquire highlight data by using the heat map data as reference sports data.

Also, the data analysis system 2000 may obtain heat map data by processing sports data (particularly, location data) of a sports participant obtained through the data acquisition system 1000 . Also, the data analysis system 2000 may acquire heat map data through the online web server 3000 . For example, the data analysis system 2000 may acquire heat map data through an Internet sports broadcast relayed through the online web server 3000 . In addition, the heat map data may include heat map data for a plurality of participants who performed the sports game.

9 shows an example of sports data in a dynamic data format.

Referring to FIG. 9 , the data analysis system 2000 according to an embodiment may acquire highlight data using dynamic data. Specifically, the second controller 2002 may acquire the highlight data using the dynamic data as the base sports data. Also, the second controller 2002 may acquire highlight data by using the dynamic data as reference sports data.

Also, the data analysis system 2000 may acquire dynamic data from the data acquisition system 1000 . Specifically, the data analysis system 2000 may acquire dynamic data through dynamic data acquired from a dynamic sensor attached to a sports participant. In addition, the data analysis system 2000 may obtain dynamic data by processing sports data (particularly, location data) of a sports participant obtained through the data acquisition system 1000 . Specifically, the second controller 2002 may obtain dynamic data such as speed data by differentiating the position data. In addition, the dynamic data may include dynamic data about a plurality of participants who performed the sporting event.

10 shows an example of sports data obtained from an online web server.

Referring to FIG. 10 , the data analysis system 2000 according to an embodiment may acquire indirect sports data through the online web server 3000 .

The data analysis system 2000 according to an embodiment may acquire highlight data using indirect sports data. Specifically, the second controller 2002 may acquire the highlight data by using the indirect sports data as the base sports data. Also, the second controller 2002 may acquire highlight data by using the indirect sports data as reference sports data.

Here, the indirect sports data may include a game log in which sports events are annotated for each game running time.

In addition, the indirect sports data may include Internet texts or posts written by viewers who watch Internet sports broadcasts. Here, the Internet text or post may include information about a specific team or player supported by the viewer. In addition, the indirect sports data may include opinions of viewers reflecting the contents of the sports game.

In addition, the indirect sports data may include an absolute time at which the sports data was written.

In the above, direct sports data and indirect sports data were examined as examples. Indirect sports data has an advantage in that it is relatively easy to obtain compared to direct sports data, and direct sports data has an advantage in that it can include more accurate information related to sports events compared to indirect sports data.

Also, since the information included in the direct sports data and the indirect sports data is different from each other, an analysis method for detecting a sports event may be different.

Hereinafter, an analysis operation of direct sports data and indirect sports data performed by the data analysis system 2000 will be described in detail with reference to the drawings.

First, an operation in which the data analysis system 2000 directly analyzes sports data will be described in detail with reference to FIGS. 11 to 19 . First, with reference to FIG. 11 , an operation in which the data analysis system 2000 directly analyzes sports data will be generally described.

11 is a flowchart schematically illustrating an operation of directly analyzing sports data by a data analysis system according to an exemplary embodiment.

11 , the direct sports data analysis operation according to an embodiment includes the steps of directly acquiring sports data (S2000), extracting a sports event by analyzing factors included in the acquired direct sports data (S2200), and It may include calculating a feature score for each extracted event (S2400).

First, the data analysis system 2000 according to an embodiment may directly acquire sports data (S2000). Since the operation of acquiring sports data is the same as described above, a detailed description thereof will be omitted.

Thereafter, the data analysis system 2000 may extract a sports event in consideration of factors included in the obtained direct sports data ( S2200 ). Specifically, the second controller 2002 may directly analyze various factors included in the sports data and determine a sports event corresponding to the analyzed factors.

In this specification, the meaning of factors directly included in sports data should be interpreted as a concept meaning all information that is included in sports data and can reflect the sports event itself or the characteristics of the sports event.

Illustratively, the factors in the sound data may be distinguished by the size of the sound, the type of the sound, the location of the sound, etc. In the image data, the shape of the image in the image may be a factor for judging a sports event. . Specific examples of factors in sports data reflecting the characteristics of a sports event will be described later in detail.

After the sports event extraction step (S2200), the data analysis system 2000 may calculate a feature score for the extracted sports event (S2400). Specifically, the second controller 2002 may calculate a feature score corresponding to the highlight data to be obtained for each of the extracted sports events. Calculation of feature scores of direct sports data will also be described later.

Here, the feature score may be used as a criterion for generating highlight data. The content related to the feature score will be described later in detail.

In the above, the analysis operation of direct sports data performed by the data analysis system 2000 has been overviewed. Hereinafter, examples of an operation in which the data analysis system 2000 directly analyzes factors included in sports data will be described with reference to the drawings.

The data analysis system 2000 according to an embodiment may directly detect a sports event in consideration of various factors included in sports data. Specifically, the second controller 2002 may directly detect various factors included in sports data and determine a sports event corresponding to the detected factors.

12 to 15 show examples of a correspondence relationship between an acoustic factor considered when the data analysis system 2000 analyzes an acoustic factor included in acoustic data and a sporting event.

Specifically, FIGS. 12 to 14 relate to examples showing the relationship between a sound factor related to a sound level and a sporting event, and FIG. 15 is an example showing the relationship between an acoustic factor related to a type of sound and a sporting event. it's about

12 is a diagram illustrating an example of a correspondence relationship between a sound level included in sports data and a sports event according to an exemplary embodiment.

According to an embodiment, the data analysis system 2000 may detect a sports event in consideration of acoustic factors included in the acoustic data. Specifically, the second controller 2002 may determine the sporting event corresponding to the acoustic factors by considering the acoustic factors related to the sound level.

Here, the sound factor related to the sound level may reflect the characteristics of spectators for a general team or player playing a sporting event.

Also, the general characteristics of a team or player playing a sporting event may vary. For example, the general characteristic may be a home team/away team.

A specific example will be described with reference to the drawings.

Referring to FIG. 12 , acoustic data when the same or similar sporting event is performed by a home team or an away team is illustrated. Although it is the same sporting event, it is shown that the intensity of the acoustic data measured by the home team was higher than that performed by the away team. For example, in a soccer game, it is common that the sound intensity when the home team scores a goal is higher than when the away team scores. According to an embodiment, the second controller 2002 may detect a sports event from the acoustic sports data in consideration of these acoustic factors.

In addition, the data analysis system 2000 may directly normalize sports data in consideration of the characteristics of spectators for a general team or player playing a sports game. can be used to normalize the acoustic data.

That is, the data analysis system 2000 may directly normalize the sports data in consideration of whether the home team or the away team is present. Specifically, when normalizing data that can be determined as a sports event among information directly included in the sports data, the second controller 2002 may consider whether the home team/the away team is present. In other words, it can be expressed that the second controller 2002 can directly normalize the sports data in consideration of the reaction of the spectators to the team or player playing the sports event (specifically, a specific sports event).

Even though it is sound data generated from the same sporting event, information included in the sound data may vary depending on team characteristics, such as whether a home team or an away team is present. can cause confusion when detecting Accordingly, the data analysis system 2000 may more accurately detect a sports event by removing these environmental factors.

13 is a diagram illustrating another example of a correspondence between a sound level included in sports data and a sports event according to an exemplary embodiment.

According to an embodiment, the sound factor related to the sound level may reflect a unique characteristic of a specific team or a specific player.

Here, the intrinsic tendencies of a specific team or specific players may vary. For example, the intrinsic tendency may be a support tendency of a fan who supports a professional team/professional player, and may be a support tendency of a fan who supports a popular player/a less popular player. Specifically, when a team belonging to a hometown having a noisy cheering culture plays a sports event, the signal of the acoustic data will be stronger than when a team belonging to a hometown other than the hometown performs the same sports event. As another example, when a team having a quiet cheering culture plays a sports event, the signal of the acoustic data may be weakly measured compared to when a team having a non-supporting culture plays a sporting event.

The data analysis system 2000 may directly normalize sports data in consideration of the intrinsic tendencies of a specific team or players. Here, the intrinsic tendency of a specific team or players may be a cheering tendency of a fan who supports a professional team/pro player.

A specific example will be described with reference to the drawings.

Referring to FIG. 13 , acoustic data when a specific team or player having different tendencies performs the same or similar sporting event is shown. Although it is the same sporting event, when the team having the first tendency performed, the intensity of the acoustic data was measured to be high, and when the team having the second tendency was performed, it was shown that the intensity was measured low. As such, the second controller 2002 may detect a sporting event in consideration of the propensity of a specific team or a specific player to play a sporting event.

Also, as described above in FIG. 12 , the data analysis system 2000 may directly pre-process sports data in consideration of the unique characteristics of a specific team or a specific player. That is, the second controller 2002 may normalize the sound data by using sound factors reflecting the unique characteristics of a specific team or a specific player.

14 illustrates another example of a correspondence between a sound level included in sports data and a sports event according to an exemplary embodiment.

According to an embodiment, the sound factor related to the sound level may reflect a characteristic according to the duration of the sports game. Here, the characteristics according to the running time of the sports game may vary. For example, as the running time of the sports game elapses, the cheers of the spectators may increase.

A specific example will be described with reference to the drawings.

Referring to FIG. 14 , acoustic data when the same sporting event occurs at the beginning and at the end of a sporting event, respectively, is shown. Although it is the same sporting event, it is shown that the intensity of the acoustic data for the sports event that occurred at the beginning of the game was measured to be low, and the intensity of the acoustic data for the sport event that occurred at the end of the game was measured to be high.

In addition, in the case of sports in which a specific session is distinguished, such as tennis or martial arts, or a sport in which the intensity of an acoustic signal increases when a specific sports event (for example, a punch hit or a down in martial arts) occurs, acoustic data according to time Detecting a sporting event according to the change of .

As such, the second controller 2002 may detect a sports event from the acoustic sports data in consideration of acoustic factors reflecting characteristics according to the duration of the sports competition.

Also, as described above with reference to FIG. 12 , the data analysis system 2000 may directly pre-process sports data in consideration of the running time of a sports game. That is, the second controller 2002 may normalize the acoustic data by using the acoustic factors reflecting the characteristics according to the running time of the sports game.

In the above, examples of sound factors related to sound levels that can be considered when the data analysis system 2000 detects a sporting event using sound sport data have been described.

Hereinafter, examples of acoustic factors related to sound types that can be considered when the data analysis system 2000 according to an embodiment extracts a sports event from acoustic sports data will be described with reference to the drawings.

15 illustrates an example of a correspondence relationship between a sound factor related to a type of sound and a sport event according to an embodiment.

Referring to FIG. 15 , the data analysis system 2000 may detect various factors included in sound data. Specifically, the second controller 2002 may distinguish a plurality of acoustic factors included in the acoustic data and detect each of the distinguished acoustic factors.

In particular, the second controller 2002 may analyze sound factors related to the type of sound and detect a sport event corresponding to the sound factor.

Here, the sound factor related to the sound type may reflect the characteristics of various sports events. For example, a sound factor related to a type of sound may reflect a situation that may occur according to a rule of a specific sport when playing a sport.

Specifically, for example, when the target sport for which the sound data is obtained is soccer, the sound factor is according to the soccer rules when playing soccer, such as a scoring situation, a shooting situation, a tackle situation, a foul situation, a dribble breakthrough, a chance close to a goal, etc. It can reflect any situation that may arise.

Illustratively, in a scoring situation, the acoustic data is "Wow!", "Wow!" It may include acoustic factors such as a shout sound, etc., and in a sports event such as a dribble breakthrough, acoustic data may include acoustic factors such as exclamations such as "Oh!" and "Wow".

Also, in sports such as tennis, it goes without saying that the acoustic factor may reflect characteristics of sports events such as breakpoints, sub-aces, winning shots, and errors.

As another example, the characteristic of the sporting event reflected by the acoustic factor may be the behavioral characteristic of a sports participant who plays the sport. Specifically, when a positive sporting event such as manner play or fair play of a sports participant occurs, the sound data may include sound factors reflecting the characteristics of the sporting event. Illustratively, in a manner play situation of a sports participant, the acoustic data may include acoustic factors such as a soft applause. Alternatively, of course, other acoustic factors may be detected in a negative event such as a dirty play of a sports participant.

In the above, an example of a correspondence relationship between an acoustic factor included in acoustic data and a sports event has been described. Hereinafter, an example of a correspondence relationship between an image factor included in image data and a sports event will be described with reference to the drawings.

16 illustrates an example of a correspondence relationship between an image factor included in image data and a sports event according to an exemplary embodiment.

Referring to FIG. 16 , the data analysis system 2000 may detect a sports event using image data. Specifically, the second controller 2002 may detect image factors included in the image data and extract the sports event by analyzing the detected image factors.

The image factor may reflect characteristics related to various sporting events.

As an example, the image factor may reflect a sports event related to a motion played by a sports participant. Here, the motion played by the sports participant may mean any motion that can be played in the corresponding sport.

For example, the motion of the sports participant may mean a shooting motion, a passing motion, a dribbling motion, a running motion, a tackle motion, a swing motion, a smash motion, and the like. The second controller 2002 may analyze the image factor reflecting the motion of the sports participant to detect a sports event related to the motion. As a specific example, the second controller 2002 may detect a shooting event by analyzing an image factor in which a shooting motion included in image data related to soccer or basketball is reflected, and a smash motion included in image data related to tennis is reflected. A smash event can be detected by analyzing the image factor.

As another example, the motion of the sports participant may mean a gesture. For example, it may mean an action such as hugging, shaking hands, or applause of sports participants after a sports game is finished. The second controller 2002 may analyze the image factor to which the gesture of the sports participant is reflected, and detect a sports event related to the gesture.

In addition, the image factor may reflect a sports event related to a sports situation that may occur other than the motion of the sports participant. Here, the sports situation may be an event that occurs according to the rules of the sport, such as a foul situation, a scoring situation, etc., or an event related to an unexpected event such as intrusion of spectators or interruption of a match.

That is, the image factor may reflect information on all events that can be photographed as image data in a sports game. In other words, the second controller 2002 may detect a sports event by analyzing an image factor that can reflect various events that can be photographed as image data in a sports game.

In the above, an example of a correspondence relationship between an image factor included in image data and a sports event has been examined. Hereinafter, an example of a correspondence between a heat map factor included in the heat map data and a sports event will be described with reference to the drawings.

17 illustrates an example of a correspondence relationship between a heat map factor included in heat map data and a sports event according to an exemplary embodiment.

Referring to FIG. 17 , the data analysis system 2000 may detect a sports event using location data. Specifically, the second controller 2002 may detect heat map factors included in the heat map data related to the location data, and extract the sports event by analyzing the detected heat map factors.

Here, the heat map data may be obtained by processing the location data as described above. In other words, the heat map data may be expressed as a set of location data of sports participants for a predetermined time. Therefore, in the following description of the present specification, the term heat map data is used for convenience of description, but it will be understood that this is a concept including location data of course, and the terms location data and heat map data may be used interchangeably. have.

The heat map data may also reflect a sports event performed by a sports participant. For example, the heat map data may reflect a sports event related to a tactical play performed by a sports participant.

As another example, the heat map data may reflect a sporting event related to various play situations in a sporting event.

Specifically, the heat map data may reflect a corner kick situation. As shown in the figure, the heat map data may include information on the location of the corner kick kicker on the stadium corner flag in a corner kick situation, which means that a sports event related to a corner kick is reflected.

As another specific example, the heat map data may reflect a free kick situation. As shown in the figure, the heat map data may include location information of players who built a wall in a free kick situation or the remaining players preparing for a free kick situation, which means that a sports event related to a free kick is reflected in the heat map data do .

In addition, the heat map data may reflect location information of a sports participant in a specific sports event occurrence situation. That is, as will be described later in detail, when the data analysis system 2000 uses the heat map data as the base sports data, the data analysis system 2000 performs sports in the sports event situation detected from the reference sports data from the heat map data. Participant location information can be extracted.

As a specific example, as illustrated in the drawing, when a sports event in a scoring situation is detected as a result of analysis of the reference sports data, location information of a specific player in the corresponding scoring situation may be extracted from the heat map data.

In the above, an example of a correspondence relationship between a heat map factor included in the heat map data and a sports event has been described. Hereinafter, an example of a correspondence between a dynamic factor included in dynamic data and a sport event will be described with reference to the drawings.

18 illustrates an example of a correspondence relationship between factors included in dynamic data and a sports event according to an embodiment.

Referring to FIG. 18 , the data analysis system 2000 may detect a sports event using dynamic data. Specifically, the second controller 2002 may detect dynamic factors included in the dynamic data and extract the sports event by analyzing the detected dynamic factors.

Here, the dynamic data may be obtained by processing the position data as described above, or may be obtained through a dynamic sensor attached to a sports participant. In addition, dynamic data should be interpreted as meaning all data related to the motion of an object, such as velocity, acceleration, direction, jerk, angular velocity, and angular acceleration. Therefore, in the following description of the present specification, the term dynamic data will be mainly used for convenience of explanation, but it is understood that this is a concept that includes various dynamic data such as speed, acceleration, angular acceleration, jerk, etc., which are of course sub-concepts. will be, and the terms of these sub-concepts of dynamic data and the terms of dynamic data may be used interchangeably.

Dynamic data may also reflect sporting events performed by sporting participants. As an example, the dynamic data may reflect a sports event related to a tactical play performed by a sports participant. Primarily, the dynamic data will reflect information about sporting events related to individual tactical plays performed by sporting participants.

In addition, the dynamic data may reflect a sporting event of various objects used in a sporting event. For example, the dynamic data may reflect information about a sport event related to a ball used in a sporting event or a racquet.

Also, similar to the heat map data described above, the dynamic data may reflect dynamic information of a sports participant in a specific sports event occurrence situation. That is, as will be described later in detail, when the data analysis system 2000 uses dynamic data as the base sports data, the data analysis system 2000 determines the sports participants in the sports event situation detected from the reference sports data from the dynamic data. Dynamic information can be extracted.

As a specific example, as shown in the drawing, when a sports event in a scoring situation is detected as a result of analysis of the reference sports data, dynamic information of a specific player in the corresponding scoring situation may be extracted from the dynamic data.

As another example, when a sub-ace in tennis is detected as a result of analyzing the reference sports data, speed information of the sports object in the corresponding sub-ace situation may be extracted from the dynamic data.

As another example, as shown in the drawing, the data analysis system 2000 may determine that a sports event has occurred when dynamic data greater than or equal to a predetermined dynamic data value is detected.

In the above, examples of sports factors included in various sports data considered when the data analysis system 2000 detects a sports event have been described.

Hereinafter, an example of calculating a feature score for a detected event will be described with reference to the drawings.

19 is a diagram illustrating an example in which a data analysis system directly uses sports data to calculate a feature score for a sports event according to an embodiment.

Referring to FIG. 19 , the data analysis system 2000 according to an embodiment may calculate a feature score for a detected sports event. Specifically, the second controller 2002 may calculate a feature score for the detected sport event by using the checca sports data.

Here, the feature score may mean a degree of importance of a specific sporting event within a corresponding sport. For example, when a plurality of sporting events related to shooting are detected from the reference sport data, it is necessary to select a major sporting event to be extracted as highlight data from among the plurality of sporting events related to the shooting, and An index serving as a criterion for selection may be expressed as a feature score.

Also, here, the feature score may be calculated in consideration of various factors included in the aforementioned direct sports data.

For example, when the detected sporting event is a shooting, the higher the feature score is, the higher the feature score may be for a sporting event including an acoustic factor in which the shout or exclamation of the audience is largely reflected. As another example, when the detected sporting event is a foul, the higher the feature score is, the higher the feature score may be calculated as the sporting event includes an acoustic factor that greatly reflects spectators' boos or sighs.

The content related to the generation of highlight data using the feature score will be described later in detail.

Also, referring back to FIG. 19 , the data analysis system 2000 according to an embodiment may calculate a fitness index for the detected sports event. Specifically, the second controller 2002 may use the sports data directly to calculate a fitness index indicating whether the detected sports event is suitable for the highlight data.

In other words, when a plurality of events are detected from the reference sports data, only a specific type of sports event from among the plurality of events needs to be included in the highlight data, and as such, a specific type of sports event suitable for the highlight data. It can be expressed as the fitness index as a criterion for selecting

Here, the data analysis system 2000 may calculate a fitness index corresponding to the highlight data. Specifically, the second controller 2002 may determine whether the detected sport event is suitable for the highlight data according to the highlight data to be generated, and calculate a fitness index based on the determination result.

For example, if the highlight data is an image related to a goal scene of a sports game, a sports event related to a scoring situation will have a high fitness index, and a sports event related to a foul situation will have a low fitness index.

An example of the purpose of the highlight data and the generation of the highlight data using the fitness index will be described later in detail.

However, in the present specification, it has been described that the feature score and the fitness index are output separately, but this is only an example, and it is also possible to generate sports highlight data using other indicators that comprehensively consider the fitness index and the feature score. .

In the above, an example related to the analysis of sports data directly has been described with reference to the drawings.

Hereinafter, an example related to the analysis of indirect sports data will be described with reference to FIGS. 20 to 28 .

According to an embodiment, the data analysis system 2000 may detect a sports event from indirect sports data using a data analysis algorithm. Specifically, the second controller 2002 uses the artificial neural network stored in the second memory 1400 to

First, a general operation of the data analysis system 2000 analyzing indirect sports data will be described with reference to FIG. 20 .

20 is a schematic flowchart of an indirect sports data analysis operation according to an embodiment.

Referring to FIG. 20 , the indirect sports data analysis operation according to an embodiment includes an indirect sports data acquisition step (S3000), an indirect sports data preprocessing step (S3200), and a step of extracting a sporting event from the preprocessed indirect sports data (S3400). and calculating a feature score for each extracted event (S3600).

First, the data analysis system 2000 may acquire indirect sports data from the data acquisition system 2000 and/or the online web server 3000 ( S3000 ). Since the operation of acquiring indirect sports data is the same as described above, a detailed description thereof will be omitted.

Thereafter, the data analysis system 2000 may pre-process the obtained indirect sports data (S3200). Specifically, the second controller 2002 may pre-process the acquired indirect sports data as input data of the analysis algorithm. Here, the preprocessing of the indirect sports data may mean filtering the text included in the indirect sports. In addition, the pre-processing of the indirect sports data may be to calculate the emotional index. In addition, the pre-processing of the indirect sports data may be to detect a text increase time. Also, the pre-processing of indirect sports data may be an operation of extracting a text group corresponding to a game log. The step of pre-processing the indirect sports data ( S3200 ) will be described later in detail.

After the indirect sports data pre-processing step (S3400), the data analysis system 2000 may extract a sports event from the pre-processed indirect sports data (S3400). Specifically, the second controller 2002 is the pre-processed indirect sports data. may be input to the analysis algorithm, and a sports event may be detected based on an output result of the analysis algorithm.

Thereafter, the data analysis system 2000 may calculate a feature score for each detected sport event ( S3600 ). Specifically, the second controller 2002 may calculate a feature score corresponding to the highlight data to be obtained for each of the extracted sports events. Calculation of feature scores of indirect sports data will also be described later.

Although the step of extracting a sport event from indirect sports data (S3400) and the step of calculating a feature score for each sport event (S3600) are illustrated as separate operations in the drawings, the present invention is not limited thereto. As will be described later, the data analysis system 2000 according to an embodiment may use an artificial neural network to obtain both the detection of a sports event and the feature score calculation result from indirect sports data. This will also be described later in detail.

In the above, the operation of analyzing the indirect sports data by the data analysis system 2000 according to an embodiment has been generally described.

Hereinafter, an operation in which the data analysis system 2000 preprocesses indirect sports data will be described in detail with reference to FIGS. 21 to 26 .

First, a pre-processing operation related to text filtering and calculating a sentiment index will be described with reference to FIGS. 21 to 22 .

21 is a schematic flowchart of an operation of the data analysis system filtering text included in indirect sports data.

Referring to FIG. 21 , the pre-processing operation according to an embodiment may include a text filtering step ( S3220 ) and a step of calculating a sentiment index based on the filtered text ( S3240 ).

First, the data analysis system 2000 may filter the text included in the indirect sports data. (S3220) Specifically, the second controller 2002 is a sports event among sports-related texts included in Internet posts or SNS posts. You can filter texts that contain content related to . Also, the second controller 2002 may perform filtering based on the part-of-speech of the text.

Thereafter, the data analysis system 2000 may calculate a sentiment index of the filtered text. Specifically, the second controller 2002 may analyze the filtered text to calculate a sentiment index for the content included in the text. Here, the sentiment index may mean an index related to the emotion of the post creator included in the text. In addition, the indirect sports data from which the emotional index is calculated may be used as input data of an artificial neural network, which will be described later.

Hereinafter, a specific example of a preprocessing operation of indirect sports data related to text filtering and calculation of a sentiment index will be described with reference to FIG. 22 .

22 shows an example of indirect sports data processing related to text filtering and calculation of a sentiment index.

Referring to FIG. 22 , the data analysis system 2000 according to an embodiment may filter indirect sports data to extract only sports-related data. As illustrated by way of example, the indirect sports data may include data irrelevant to sports. If indirect sports data is used as an input algorithm of an analysis algorithm with such noises included, the accuracy of sports event detection may decrease. Accordingly, the accuracy of sports event detection may be improved through the indirect sports data filtering operation performed by the data analysis system 2000 .

Also, the data analysis system 2000 according to an embodiment may filter the text based on the part-of-speech. Specifically, the second controller 2002 may analyze the parts-of-speech included in the text, and calculate the sports event related content or the emotional index included in the corresponding part-of-speech.

As a specific example, in the case of a noun or a verb, it is common to include information about an object related to sports or an operation performed in sports. As illustrated by way of example, in the case of serial number t1 text, 'Lee Patent' includes content about a sports player and sports action such as 'shooting', which may correspond to content related to a sporting event. can

In addition, in the case of modifiers such as adverbs and adjectives, it is common to include information on motions performed in sports or information on emotions of text writers. As shown by way of example, in the case of the text serial number t1, the adverb 'good' modifies the sports action 'shooting', and represents the emotion felt by the post author.

As shown in the drawing, the data analysis system 2000 according to an embodiment may calculate the sentiment index s1-s5 for the text t1-t5 included in the indirect sports data through a pre-processing operation. Criteria for calculating the emotional index may vary. For example, in the case of text reflecting a positive sporting event, a high emotional index may be calculated. As another example, in the case of text reflecting a negative sporting event, a low emotional index may be calculated. In addition, the criterion of the sentiment index may be changed to correspond to the highlight data to be generated. This also means that the emotional index can be used as an index for determining the importance of a sports event.

In the above, the pre-processing operation related to text filtering and calculation of the sentiment index has been described with reference to the drawings. Hereinafter, the pre-processing operation related to the text increase moment detection will be described with reference to the drawings.

23 is a schematic flowchart of an operation in which the data analysis system detects an increase moment of text included in indirect sports data.

Referring to FIG. 23 , the preprocessing operation according to an embodiment may include detecting a text increase moment ( S3222 ) and calculating a sentiment index based on the filtered text ( S3242 ).

First, the data analysis system 2000 may detect the moment when the text included in the indirect sports data increases. (S3222) Specifically, the second controller 2002 is a sports-related text included in an Internet post or SNS post. In the middle, it is possible to calculate the amount of text increase according to time, and detect a section in which the text rapidly increases.

Thereafter, the data analysis system 2000 may calculate a text increase index based on the sensed text increase amount. Specifically, the second controller 2002 may calculate the text increase index by using the increased amount of text in the section in which the text is increased. Here, the text increase index may mean an index related to the text increase amount corresponding to the text increase section. In addition, the indirect sports data from which the text increase index is calculated may be used as input data of an artificial neural network, which will be described later.

Hereinafter, a specific example of a preprocessing operation of indirect sports data related to text increase index calculation will be described with reference to FIG. 24 .

24 shows an example of indirect sports data processing related to text increase index calculation.

Referring to FIG. 24 , the data analysis system 2000 may detect a section in which text increases in indirect sports data. Specifically, the second controller 2002 may detect a section in which the text increases with time by using time information included in the indirect sports data.

Referring to the drawings, for example, the text increase section in the indirect sports data is divided into a first section to a third section. Here, the first section is related to the chance situation from 19:02 to 19:03, the second section is related to the foul situation from 19:15 to 19:16, and the third section is related to the scoring situation from 19:30 Each of the time intervals from ~19:31 is shown. It will be understood that the time interval shown in the drawings is merely exemplary, and the text increase period may be set in various ways.

Also, the data analysis system 2000 may calculate a text increase index. Specifically, the second controller 2002 may calculate the text increase amount increased during each text increase section.

Referring to the drawings, the text increase indices I1 to I3 in the first to third sections are illustrated by way of example. Referring to the drawing, it is shown that the amount of text increase is sharply increased in the third section reflecting the scoring situation than in the first section and the second section reflecting the chance situation or the foul situation. In this case, the text increase indices I3 in the third section may be calculated to be higher than the text increase indices I1 and I2 in the remaining sections. That is, the data analysis system 2000 according to an embodiment may determine the importance of the sports event occurring in the corresponding text section through the text increase index. That is, the text increase index may be used as an index for determining the importance of a sports event in a corresponding section.

In the above, the preprocessing operation related to the text increase instantaneous index calculation has been described with reference to the drawings. Hereinafter, the operation related to text clustering will be described with reference to the drawings.

25 is a flowchart of an operation of the data analysis system processing indirect sports data through text clustering.

Referring to FIG. 25 , the text clustering operation according to an embodiment may include clustering the text by comparing the game log with the text ( S3226 ) and setting a highlight extraction time based on the clustering result ( S3246 ). have.

First, the data analysis system 2000 may group a plurality of texts by comparing the game log and the text. (S3226) Specifically, the second controller 2002 compares the game log and text data related to the same game, and , a plurality of text groups including at least one text matching the contents of each game log may be obtained. This will be described in detail with reference to FIG. 27 .

Thereafter, the data analysis system 2000 may acquire time information related to a sports event based on the extracted text group. Specifically, the second controller 2002 may determine the duration of the sports event to be extracted into the highlight data based on the writing time of the text included in the text group corresponding to each sports event. In this regard, it will be described in detail with reference to FIG. 27 .

26 illustrates an example of text clustering according to an embodiment.

Also, the data analysis system 2000 may partition a text group corresponding to the content included in the game log. Specifically, the second controller 2002 compares the time information corresponding to at least one or more sporting events included in the game log with the time at which a plurality of texts are written, and extracts at least one text related to each sporting event. It is possible to obtain a plurality of text groups corresponding to the sporting event of .

As an exemplary description with reference to the drawings, FIG. 28 shows that text groups (first to fourth groups) for each content of a game log of a soccer match are classified. In the case of the first group, a total of three texts are included from 19:02 to 19:03 for sporting events related to shooting, and in the second group, from 19:15 to 19:16 for sporting events related to fouls A total of two texts are included, and in the case of the third group, it is shown that a large amount of texts are included from 19:30-19:32 in relation to the scoring situation.

Also, the data analysis system 2000 may determine the duration of the sports event by using time information included in each text group. Specifically, the second controller 2002 may determine the duration of the sports event to be reflected in the highlight data based on the writing times of the plurality of texts included in the text group.

Referring to the drawings, the duration from 19:02 to 19:03 is illustrated for the first group corresponding to the shooting-related sporting event of FIG. 28 . That is, the second controller 2002 may extract sports data corresponding to the time from 19:02 to 19:03 from among the sports data acquired in the corresponding sports event and use it as highlight data.

As another example, the duration from 19:30 to 19:32 is illustrated for the third group corresponding to the sporting event related to the scoring situation. In this case, the second controller 2002 may extract sports data corresponding to a time from 19:30 to 19:32 from among the sports data acquired in the corresponding sports event and use it as highlight data.

Also, the data analysis system 2000 may calculate the importance of the sports event based on the text included in the text group. Specifically, the second controller 2002 may calculate the feature score or fitness index of the sports event corresponding to the text group in consideration of the total amount or duration of posts included in the text group partitioned for each sport event.

As for the relay image of the sporting event, an image related to the sporting event may be transmitted for a time different from the time at which the sporting event is actually performed. For example, in the case of a sports event related to a soccer scoring situation, a sports video may be transmitted for a significantly longer time than the time taken for an actual score through repeated replays, and even during the time the replay video is transmitted, viewers can view the corresponding sports event It is common to write related posts.

On the other hand, in the case of a sports event in which a scoring situation frequently occurs, such as basketball, the viewer's reaction may be differentiated even in the case of the same sports event (scoring). That is, for example, if a goal is scored through a cool action such as a dunk shot, the duration of the text group may be prolonged as well as the total amount of text included in the text group may increase. The duration will not be long and the total amount of texts included in the text group will not be large either.

That is, the total amount or duration of text included in the text group may reflect various characteristics of a sports event, and the data detection system 1000 according to an embodiment determines the total amount or duration of text included in the text group. Considering this, it is possible to extract sporting events more accurately and efficiently.

In the description of one embodiment of the present specification, the pre-processing operation has been described as an operation for processing sports data into input data of an artificial neural network, but the spirit of the present invention is not limited thereto. That is, according to an embodiment of the present specification, it is also possible to generate sports highlight data based on the result of performing the pre-processing operation, and it is also possible to detect only the sports event.

After all, it will be understood that all operations performed by the data analysis system 2000 described above or described below may be performed by omitting a part thereof, and a plurality of operations may be combined and performed. will be.

In the above, an example in which the data analysis system 2000 pre-processes indirect sports data and performs a text clustering operation has been described.

Hereinafter, an operation in which the data analysis system 2000 analyzes indirect sports data using an analysis algorithm according to an embodiment will be described with reference to the drawings.

The data analysis system 2000 according to an embodiment may analyze indirect sports data using an analysis algorithm. Specifically, the second controller 2002 may input the indirect sports data to the analysis algorithm stored in the second memory 1200 and detect the sports event based on the output result to the analysis algorithm.

Here, the data analysis system 2000 according to an embodiment may use various analysis algorithms.

As an example, the data analysis algorithm may be provided as a machine learning model. The machine learning model according to an embodiment may be provided in the form of an artificial neural network. Representative examples of artificial neural networks include, but are not limited to, artificial neural networks of the deep learning series including an input layer, an output layer, and a hidden layer that processes data between the input layer and the output layer. Neural networks may also be used.

Also, the data analysis system 2000 may use a machine learning technique other than an artificial neural network. That is, the data analysis system 2000 can use various machine learning techniques such as a dimensionality reduction technique such as PCA, a nearest neighbor algorithm (KNN), and a random forest. In addition, even forms combined in various ways may be used as an analysis method algorithm according to an embodiment.

Furthermore, the analysis algorithm in the present specification is not necessarily limited to a machine learning model. That is, the analysis algorithm may include various judgment/decision algorithms other than the machine learning model.

Therefore, it is clarified in advance that the analysis algorithm in the present specification should be interpreted in a comprehensive sense including all types of algorithms capable of data analysis and sports event detection using sports data.

However, for convenience of explanation, in the following description in the present specification, an analysis algorithm using a machine learning model related to an artificial neural network will be mainly described. However, due to this, it will be clear that the analysis algorithm in the present specification is not limited to an artificial neural network-based machine learning model.

In the present specification, the artificial neural network may receive indirect sports data and output a sports event detection result.

27 is an exemplary diagram of an artificial neural network according to an embodiment.

Referring to FIG. 27 , the artificial neural network may include an input layer that receives indirect sports data, an output layer that receives a sports event detection result, and at least one hidden layer disposed between the input layer and the output layer.

Indirect sports data may be input to the input layer. The input layer may include a plurality of input nodes. Preprocessed indirect sports data may be input to each of the input nodes. For example, the aforementioned emotional index, text content, text increase index, text clustering result, etc. may be input to each of the input nodes. Alternatively, indirect sports data that has not been pre-processed may be input as it is. In the following description of the present specification, for convenience of explanation, data input to the artificial neural network will be referred to as 'input data'.

The output layer may output a sports event detection result.

For example, when the artificial neural network outputs a sports event detection result in the form of binary classification, one or two output nodes may be included in the output layer. An artificial neural network that outputs a result value in the form of binary classicization can mainly detect whether a sports event or a highlight event including a single content is performed.

As another example, when the artificial neural network outputs a sports event detection result in the form of multiclassification, the output layer may include a plurality of output nodes. An artificial neural network that outputs a result value in the form of multi-classification can mainly output a complex detection result for highlight or not, event type, characteristic indicator, and the like.

The hidden layer may include a plurality of layers, and each layer may include at least one hidden node.

The artificial neural network may be trained using, as a running set, labeling data in which a sports event detection value obtained in advance from sports data and data based on indirect sports data are tagged with each other. Accordingly, the sufficiently trained artificial neural network may output a sports event detection result when indirect sports data is input. Here, the sports data for acquiring the previously confirmed sports event may include both the aforementioned direct sports data and indirect sports data.

Specifically, the artificial neural network that determines the sports event from the indirect sports data may learn the relation between the content of the text corresponding to each text increase index and the corresponding sports event. In addition, the artificial neural network may learn about the difference between the time at which the sports event occurs and the time at which the text occurs (hereinafter referred to as 'delay time'). Also, the artificial neural network may learn the duration of a sports event for extracting an appropriate highlight image. Here, the sufficiently trained artificial neural network may receive data related to indirect sports data and output a sports event detection result.

Reference is again made to FIG. 27 .

Referring to FIG. 27 , the data analysis system 2000 may detect a sports event from indirect sports data using an artificial neural network.

Specifically, the second controller 2002 uses the artificial neural network stored in the second memory 1200 to input a plurality of input data obtained from the indirect sports data to the input layer of the artificial neural network, and based on the indirect sports data A detection result for a sports event may be obtained according to the result of the computational processing performed by the artificial neural network.

The artificial neural network according to an embodiment may use various data that may be obtained from indirect sports data as input data. For example, the input data may be an amount of text for a predetermined time. As another example, the input data may be the emotional index of the above-described text. As another example, the input data may be content included in text. Specifically, the content included in the text may be a word or a sentence. As another example, the input data may be a clustering result.

Here, as described above, the artificial neural network may receive all of several types of input data to the input node, may receive only one type of input data, and may receive a plurality of inputs of a single type of input data.

The artificial neural network may output a sports event detection result through an operation process in the hidden layer.

For example, the sports event detection result may be the type of event included in the indirect sports data. Here, the type of event may include information on a plurality of sports events included in the indirect sports data in the form of multi-classification.

As another example, the sports event detection result may be an index indicating whether a sports event included in the indirect sports data may be included in the highlight data. In this case, the indicator may be output as a probability value or a score. In addition, the index may correspond to the above-mentioned fitness index.

As another example, the sports event detection result may be an index reflecting a viewer's reaction to the sports event. In this case, the index reflecting the viewer's reaction may correspond to the above-described feature score.

As another example, the sport event detection result may be time information on the sport event. Specifically, the sports event detection result may include information such as the occurrence time, end time, and duration of the sports event.

Hereinafter, an output result of an artificial neural network according to an embodiment will be described with reference to the drawings.

28 is an example of a sports event detection output result using an artificial neural network according to an embodiment.

Referring to FIG. 28 , the data analysis system 2000 according to an embodiment may obtain a sports event detection result using an artificial neural network.

Specifically, the data analysis system 2000 may acquire time information related to a sports event from indirect sports data. Specifically, the second controller 2002 may acquire time information including all time information related to the sport event, such as the occurrence time, end time, and duration of the sport event, using an artificial neural network.

However, here, the time information does not necessarily have to match the occurrence time or end time of the actual sports event. For example, in generating highlight data reflecting a viewer's demand, the context of a sporting event may also be a major consideration. That is, the data detection system 1000 may determine the duration of the sports event to which the buffer is added in consideration of the situation before and after the sports event. Conversely, it is also possible to obtain the sporting event duration to be shorter than the time during which the actual sporting event was performed.

Also, the data analysis system 2000 may determine the type of sports event included in the indirect sports data. Specifically, the second controller 2002 may extract a sports event from indirect sports data using an artificial neural network and determine what type of event the extracted sports event is. Also, here, time information and an event type for the time may be tagged together with the extracted sports event. In addition, as will be described later, time information, event type, feature score, fitness index, etc. may be separately or compositely tagged in the extracted sports event.

Also, the data analysis system 2000 may calculate a feature score for each sport event included in the indirect sports data. Specifically, the second controller 2002 may extract a sports event from indirect sports data using an artificial neural network and calculate a feature score of the extracted sports event. Since the content related to the feature score of the indirect sports data is similar to the feature score of the direct sports data described above, a detailed description thereof will be omitted.

Also, the data analysis system 2000 may calculate a fitness index for each sport event included in the indirect sports data. Specifically, the second controller 2002 may extract a sports event from indirect sports data using an artificial neural network and calculate a fitness index of the extracted sports event. Since the content related to the fitness index of the indirect sports data is similar to the fitness index of the direct sports data described above, a detailed description thereof will be omitted.

As a specific example with reference to the drawings, the artificial neural network according to an embodiment detected n sports events from 1-n from indirect sports data. In addition, the artificial neural network according to an embodiment acquires time information for each of n sports events. In addition, the artificial neural network according to an embodiment determines the event type for each of n sports events. In addition, the artificial neural network according to an embodiment calculates a feature score and a fitness index for each of the extracted n sports events.

To summarize the above, the data analysis system 2000 according to an embodiment may detect at least one sports event from indirect sports data using an artificial neural network, and display a plurality of indices for the detected sports event. can be tagged. In this case, the data analysis system 2000 according to an exemplary embodiment may generate highlight data using a sport event detection result in which a plurality of indices detected from indirect sports data are tagged.

Hereinafter, an operation in which the data analysis system 2000 generates highlight data using the aforementioned various sports event detection results will be described.

According to an embodiment, the data analysis system 2000 may generate highlight data using a sports event detection result.

Specifically, the second controller 2002 may extract highlight data from the base sports data by using the sports event detection result obtained from the reference sports data.

First, a flow of a highlight generating operation performed in the data analysis system 2000 will be described with reference to the drawings, and elements that may be considered when generating a highlight will be described with reference to the drawings.

29 is a schematic flowchart of an operation of generating highlight data of a data analysis system according to an exemplary embodiment.

Referring to FIG. 29 , the highlight data generation operation according to an embodiment includes the step of selecting a highlight event based on the reference sports data analysis result ( S4000 ) and the step of extracting the highlight target event from the base sports data ( S4200 ). may include

According to an embodiment, the data analysis system 2000 may select a highlight target event based on the reference sports data analysis result. Specifically, the second controller 2002 may select a sports event to be included in the highlight data in consideration of indices related to the sports event according to the reference sports data analysis result.

Here, the highlight data may be generated for various purposes. For example, the highlight data may be generated for the purpose of outputting a major sporting event of a sporting event. As a specific example, the highlight data may include all of a scoring situation, a shooting situation, a chance situation, etc. of a sports game, or only a scoring situation.

As another example, the highlight data may be generated for the purpose of outputting a sporting event related to a specific player. As a specific example, the highlight data may include all of a specific player's dribbling breakthrough situation, shooting situation, chance situation, and scoring situation, or may include only some situations, or may include only the ball touch situation of a specific player.

In the audio-visual training situation of a professional sports team or a youth sports team, extracting and viewing only a play image of a player requiring training can greatly improve training efficiency. The highlight generating method according to an embodiment may improve training efficiency by accurately and efficiently generating and providing highlight data to include only sporting events requested by viewers.

In addition to the purposes mentioned as examples, the highlight data may include all sporting events corresponding to the purposes required in the industrial field. That is, the data analysis system 2000 according to an embodiment may generate highlight data including only a sports event corresponding to a purpose requested by a viewer.

After the step of extracting the highlight target sporting event ( S4000 ), the data analysis system 2000 may extract relevant highlight data from the base sport data. Specifically, the second controller 2002 may extract highlight data from the base sports data based on the highlight target sports event obtained from the reference sports data.

Here, as described above, the reference sports data and the base sports data may be singular or plural, respectively, and the reference sports data and the base sports data may have the same format. Illustratively, it is also possible to extract highlight data from the base sports data by synthesizing the sports event detection results of the plurality of reference sports data, and based on the sports event detection results obtained from the plurality of reference sports data, the plurality of base data Of course, it is also possible to extract highlight data from

Hereinafter, examples of highlight data extracted from the base sports data by the data analysis system 2000 according to an exemplary embodiment based on the analysis result of the reference sports data will be described with reference to the drawings.

30 illustrates an example of highlight data according to an embodiment.

Referring to FIG. 30 , the data analysis system 2000 according to an embodiment may extract a sports event suitable for highlight data from a plurality of base sports data. Specifically, the second controller 2002 may select base sports data suitable for the purpose of the highlight data from among the plurality of base sports data, extract a sports event from the base sports data, and generate the highlight data.

For example, the second controller 2002 may generate highlight data by selecting image data reflecting a sporting event meeting the purpose of the highlight data from among the plurality of image data.

As a specific example with reference to the drawings, if the purpose of the highlight data is to generate an image including a sports event for a scoring situation or to generate an image including a sports event for a home run situation, the highlight data includes a scoring situation or a home run situation It would be desirable to include this best-reflected sports data. In this case, the second controller 2002 may generate highlight data by selecting image data that best reflects the scoring situation or home run situation among the image data acquired from a plurality of cameras installed in the stadium so as to meet the purpose of the highlight data. .

As another specific example with reference to the drawings, if the purpose of the highlight data is to generate an image to reflect the tactical performance of the team, the highlight data includes sports data in which information about the tactical movement performed by the team is well reflected. it is preferable In this case, the second controller 2002 may generate the highlight data by selecting the image data that best reflects the tactical movement of the team from among the plurality of images so as to match the purpose of the highlight data. In this case, the image data in which the tactical movement is best reflected may be image data obtained by a camera photographed at a wide angle. Here, as described above, the plurality of images may be acquired from a plurality of cameras installed at different positions of the stadium.

As another example, the second controller 2002 may generate highlight data by selecting image data reflecting a characteristic of a sporting event from among a plurality of image data.

As a specific example with reference to the drawings, even when the type of the detected sporting event is a scoring situation or tactical movement of the team, it is preferable that the highlight data include sports data in which information on the tactical movement or scoring situation of the team is well reflected. something to do. In this case, the second controller 2002 may generate highlight data by selecting image data that best reflects the tactical movement or scoring situation of the team from among a plurality of images so that the detected sports event characteristics are reflected. Here, as described above, the plurality of images may be acquired from a plurality of cameras installed at different positions of the stadium.

As another specific example with reference to the drawings, even in a situation in which the type of the detected sporting event is a home run or a dunk shot, it is preferable that the highlight data include sports data in which information of the corresponding sporting event is well reflected, in this case, the second controller (2002) It goes without saying that highlight data can be generated by selecting image data so that the characteristics of the corresponding sports event are well reflected.

31 illustrates another example of highlight data according to an embodiment.

Referring to FIG. 31 , the data detection system 1000 may generate highlight data in which a plurality of sports data is reflected.

Specifically, the second controller 2002 may generate highlight data in which a plurality of sports data related to a sports event are comprehensively reflected.

That is, the second controller 2002 may extract information related to a sports event from a plurality of base sports data and generate highlight data so that the extracted information is comprehensively reflected.

For example, image data, dynamic data, heat map data, text data, etc. related to a sports event may be reflected in the highlight data. Specifically with reference to the drawings, when generating a highlight image including a sports event related to a dribble breakthrough situation, the second controller 2002 includes the sports player's speed data and the sports player's speed data in the image data of the corresponding sports event. Location heat map data, text data for a corresponding sports event, etc. can be merged and output.

Also, when generating a highlight image, the second controller 2002 may augment several graphic elements in the highlight data.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic such as floppy disks. - includes magneto-optical media, and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with reference to the limited embodiments and drawings, various modifications and variations are possible by those skilled in the art from the above description. For example, the described techniques are performed in a different order than the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

Claims (11)

A method for generating a sports highlight video, the method comprising:
acquiring first image data in which an image of a first sports game is captured;
First online text data related to the image of the first sports game - The first online text data includes a writing time that is information about a time when the first online text data is generated and a writing content related to content information of the first online text data comprising - obtaining;
determining an amount of the first online text data written for a predetermined time based on the writing time of the first online text data; and
Second online text data obtained in advance and reflecting the second sporting event in second image data including a plurality of second sporting events occurring in a second sporting event, wherein the second online text data is the second online text Using an artificial neural network learned using a learning set labeled with the analysis result of - including the writing time, which is the time information at which the data is generated, and the writing content related to the content information of the second online text data, the first image Extracting sport event data including at least one first sporting event reflected by the first online text data from the data, the first sporting event occurring in the first sporting event;
The artificial neural network is learned based on the amount of the second image data and the second online text data related to the second sports event
How to create a sports highlight video.
The method of claim 1,
Calculating the emotional index for the first online text by analyzing the written content included in the first online text data; further comprising,
The artificial neural network is to be learned using the correlation of the emotional index between the second sports event and the second online text data included in the second image data
How to create a sports highlight video.

delete The method of claim 1,
Calculating a time difference between an occurrence time of the first sporting event and a writing time of the online text based on the writing time of the first online text data;
The artificial neural network is to be learned based on the correlation between the time difference information between the occurrence time of the second sport event included in the second image data and the creation time of the second online text data
How to create a sports highlight video.
The method of claim 1,
The sporting event data is tagged with a feature index reflecting the importance of the first sporting event to the first sporting event,
extracting at least one first sporting event for generating a highlight image from the sporting event data based on the characteristic indicator; further comprising
How to create a sports highlight video.
The method of claim 1,
The first image data includes a first image photographed at a first angle and a second image photographed at a second angle different from the first angle,
the sporting event data reflects the type of the first sporting event;
In consideration of the type of the first sporting event, selecting at least one of the first image and the second image to reflect the information on the first sporting event; further comprising
How to create a sports highlight video

delete delete delete delete delete

Images