KR102469411B1 - Method based on image for coaching billiards and recording medium storing program to implement the method - Google Patents

Abstract

A billiard guide method and recording medium are disclosed. The billiard guide method is implemented by a computer including a memory storing computer-executable instructions and a processor that executes the computer-executable instructions, and the billiard guide method uses a video image of a billiard ball to locate the billiard ball. a billiard ball identification step of identifying a cue ball and a color, and classifying a cue ball and an object ball; And a route derivation step of deriving a recommended route so that the cue ball can move along a route that can be scored; wherein, in the route derivation step, an image similar to the video image is determined and extracted from a plurality of previously stored image information, Route information combined with the extracted image information can be derived as a recommended route.

Description

Image-based billiard guide method and recording medium {METHOD BASED ON IMAGE FOR COACHING BILLIARDS AND RECORDING MEDIUM STORING PROGRAM TO IMPLEMENT THE METHOD}

The present disclosure relates to a billiard guide method.

In general, billiards has a game method of placing a billiard ball in a rectangular billiard table and then hitting the billiard ball using a cue to score points, and is divided into three balls (three cushions), four balls, and pocket balls according to the game method . Among them, the three-cushion is to place three billiard balls, including a white ball, a red ball, and a yellow ball, on the billiard table, hit the first object ball and the second object ball using the cue ball, and hit the cushion installed on the side of the billiard table three or more times. It is a method in which points are recognized only when they collide.

In billiards, it is important to first identify the path of the ball that can be recognized as a score and move the ball along that path, but at the beginner level, it is not easy to think of a path that can score, and where on the cue ball should be hit with the cue. There was a difficult problem in figuring out how to move the apparatus to the desired path only by adjusting the strength of the force at the time of hitting. In this regard, Korean Patent Registration No. 10-1903656 has proposed a technique for a billiard practice device, but the development of a new technique that can more effectively improve the conventional problems is required.

The technical idea of the present disclosure is to solve the above problems, and an object thereof is to provide a technique for guiding a scoring path and an appropriate hitting method in a three-cushion game.

In addition, another object of the technical idea of the present disclosure is to provide a technique for deriving a scoring path and a hitting method by applying an artificial intelligence technique for recognizing an image.

The problem to be solved by the present invention is not limited to the above problem, and other technical problems not mentioned will be clearly understood by those skilled in the art from the description below.

As one embodiment of the present invention to achieve this object, the billiard guide method is implemented by a computer including a memory in which computer executable instructions are stored and a processor that executes the computer executable instructions, and the billiard guide method A billiard ball identification step of identifying the location and color of the billiard ball using the video image of the billiard ball and distinguishing between a cue ball and an object ball; And a route derivation step of deriving a recommended route so that the cue ball can move along a route that can be scored; wherein, in the route derivation step, an image similar to the video image is determined and extracted from a plurality of previously stored image information, Route information combined with the extracted image information can be derived as a recommended route.

In addition, the billiard guide method further includes an object recognition model generation step of creating an object recognition model based on an artificial intelligence neural network by analyzing and learning captured images made by photographing billiard balls disposed on a billiard table, and the object recognition model In the generating step, the captured image may include a pool table frame with four sides, a pool table cushion with four sides, and primary labeling information for classifying white balls, red balls, and yellow balls.

In addition, the billiard guide method further includes a data storage step of storing a captured image including secondary labeling information, and in the data storage step, the captured image further includes first information, second information, and third information. wherein the first information is information about the location of a user playing the cue polo, and is applied to one side of a pool table frame consisting of four sides, and the second information is the position of the first cushion that contacts when the cue polo moves As information on , it is applied to one side of a billiard table cushion composed of four sides, and the third information is information about a path of a cue ball that can score when the first information and the second information are determined, and the second information The labeling information is information provided for the apparatus to move according to the third information, and may include at least one selected from among the point, thickness, and strength of stroke of the apparatus.

In addition, in the billiard ball identification step, the frame and cushion of the billiard table included in the video image are subdivided into four sides and recognized through the object recognition model, and the cue ball and object ball can be distinguished based on the color of the identified billiard ball. have.

And, in the path derivation step, an arbitrary side of the pool table frame consisting of four sides is designated as the user's position (hereinafter referred to as a 'first position'), and any side of the pool table cushion consisting of four sides is designated as the cue ball. A position designation step of designating a position of a first cushion (hereinafter referred to as a 'second position') that is in contact when moving; an image extraction step of extracting images corresponding to the first and second positions from among the captured images stored in the data storing step; The degree of similarity between the positional coordinates of the white ball, red ball, and yellow ball included in the photographed image extracted in the image extraction step and the positional coordinates of the white ball, red ball, and yellow ball included in the video image are determined, and the degree of similarity is determined by a similarity algorithm. a similar image selection step of selecting a photographed image having the highest value; and a result output step of outputting third information and secondary labeling information included in the captured image selected in the similar image selection step.

In addition, in the billiard ball identification step, the position coordinates of the billiard ball are derived using a bounding box, which is a virtual rectangle surrounding the billiard ball included in the video image, and the position coordinates of the billiard ball are the intersection of two diagonals of the bounding box. can be applied as

In addition, the billiard guide method further includes a motion determination step of determining the motion of the billiard ball using a motion picture of the billiard ball, and in the motion determination step, an image is extracted from the video in units of a predetermined time, and the extracted It may be determined that the billiard ball is in a stationary state when there is no change in the measured RGB value by measuring the change in the RGB value of each pixel of the image.

As another embodiment of the present invention to achieve this object, a program for performing the above-described billiard guide method may be recorded in a computer-readable recording medium.

The solutions to the problems described above are merely illustrative and should not be construed as limiting the present invention. In addition to the exemplary embodiments described above, there may be additional embodiments described in the drawings and detailed description.

As described above, according to various embodiments of the present invention, it is important to automatically recognize billiards video images through an object recognition model learned based on an artificial intelligence neural network and derive intuitive results based on data labeling. It is possible. That is, in various embodiments of the present invention, since it is possible to guide the user to a route that can be scored and to present information necessary for the apparatus to move along the route (eg, score, thickness, strength of stroke, etc.) It can stimulate interest and contribute to the improvement of skills.

In addition, according to various embodiments of the present invention, the location of the user and the location of the first cushion are not specified because the location of the user and the location of the first cushion are first specified, and the image similarity is determined after extracting corresponding images. Compared to the process of determining image similarity in a state where the similarity is not detected, the similarity determination operation performance is optimized.

Effects according to various embodiments of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

1 is a block diagram schematically showing each component of a billiard guide system according to an embodiment of the present invention.
2 is a flowchart schematically illustrating a billiard guide method according to an embodiment of the present invention.
Figure 3 schematically illustrates a state in which a photographing unit photographs a billiard table in one embodiment of the present invention.
4 schematically illustrates a video image captured in one embodiment of the present invention.
5 schematically illustrates a captured image including first information, second information, third information, and secondary labeling information in one embodiment of the present invention.
Figure 6 schematically illustrates identifying the position of a billiard ball from a video image in one embodiment of the present invention.
7 schematically illustrates how images corresponding to specific first and second positions are extracted in one embodiment of the present invention.
8 schematically shows how the scoring path and the hitting method corresponding to the image with the highest degree of similarity are output in one embodiment of the present invention.

A preferred embodiment of the present invention will be described in more detail with reference to the accompanying drawings, but the already well-known technical parts will be omitted or compressed for conciseness of description.

It should be noted that references herein to “one” or “an” embodiment of the invention are not necessarily to the same embodiment, they mean at least one.

In the following embodiments, terms such as first and second are used for the purpose of distinguishing one component from another component without limiting meaning.

In the following examples, expressions in the singular number include plural expressions unless the context clearly indicates otherwise.

In the following embodiments, terms such as include or have mean that features or elements described in the specification exist, and do not preclude the possibility that one or more other features or elements may be added.

When an embodiment is otherwise implementable, a specific process sequence may be performed differently from the described sequence. For example, two processes described in succession may be performed substantially simultaneously, or may be performed in an order reverse to the order described. That is, each step of the methods described herein may be suitably performed in any order unless otherwise stated in the specification or clearly contradicted by context.

Since each component shown in the drawings is arbitrarily shown for convenience of explanation, the present invention is not necessarily limited to the bar shown.

1 is a block diagram schematically showing each component of a billiard guide system according to an embodiment of the present invention, and FIG. 3 is a schematic diagram of a photographing unit photographing a billiard table in an embodiment of the present invention, Figure 4 schematically shows a video image captured in an embodiment of the present invention, Figure 5 is a first information, second information, third information and secondary labeling information included in an embodiment of the present invention A photographed image is schematically shown, and FIG. 6 schematically illustrates identifying a position of a billiard ball from a video image in one embodiment of the present invention. Figure 8 schematically shows how the images corresponding to the position and the second position are extracted, and FIG. it is depicted 1 to 8, the billiard guide system 10 according to an embodiment includes a photographing unit 100, a motion determination unit 200, an object recognition model unit 300, a data storage unit 400, a location It may include a designation unit 500, an image extraction unit 600, an image selection unit 700, and a result output unit 800. In one embodiment, billiard guide system 10 may be computer implemented. Here, the computer may include a communication device communicating through at least one communication network, a display outputting screen information, a memory storing computer executable instructions, and a processor electrically connected to the memory.

In one embodiment, the motion determination unit 200, the object recognition model unit 300, the location designation unit 500, the image extraction unit 600, the image selection unit 700, and the result output unit 800 can be executed by a computer. It may be stored in a memory in the form of a command, and a processor may implement the function of each component by executing the command stored in the memory.

The photographing unit 100 is a device for generating an image by capturing a billiard ball placed on the billiard table 20 in real time. In one embodiment, the photographing unit 100 is installed above the billiard table 20, and the photographing unit 100 may generate photo information by photographing the billiard table 20 at regular intervals, and may continuously photograph the video information. can also create

The motion determination unit 200 may determine the motion of the billiard ball using video information captured by the photographing unit 100 . For example, the motion determination unit 200 extracts an image from video information in a unit of time (eg, unit of second), and measures a change in RGB value of each pixel included in the extracted image to measure RBG. When there is no change in value, it can be determined that the billiard ball is in a stationary state.

The object recognition model unit 300 is an artificial intelligence neural network having an input layer, a hidden layer, and an output layer, and includes one or more deep neural networks such as a deep neural network, a convolutional neural network, and a recurrent neural network. The learning algorithm is modularized. The object recognition model unit 300 may pre-store object recognition criteria learned by analyzing the shape of a pool table frame, the shape of a pool table cushion, the position of a billiard ball, the color of a billiard ball, etc. included in a plurality of image samples. . Therefore, the object recognition model unit 300 automatically recognizes the positions of the pool table frame and cushion included in the video image, recognizes the position and color of the billiard ball, and distinguishes between the cue ball and the object ball according to the recognized position and color. can For example, a ball recognized as white by the object recognition model unit 300 may be determined as a cue ball, a ball recognized as red may be determined as a first object ball, and a ball recognized as yellow may be determined as a second object ball. . In this specification, the cushion is installed on the inner surface of the billiard table frame, and refers to a surface in direct contact with the billiard ball.

In addition, as shown in FIG. 6, in one embodiment, the object recognition model unit 300 uses a bounding box (B), which is a virtual rectangle surrounding the billiard ball included in the video image, to position coordinates (P) of the billiard ball. ) can be derived. As a specific example, the object recognition model unit 300 may determine the center point of the billiard ball by calculating the intersection of the two diagonal lines of the bounding box B.

The data storage unit 400 is a storage device in which a plurality of previously captured images are stored. In one embodiment, the data storage unit 400 may be applied as a relational database management system. In this specification, a video image is an image captured in real time by the shooting unit 100 during an actual game, and the shot image is created in advance before an actual game, and refers to an image taken in advance. According to an embodiment, first information, second information, third information, primary labeling information, and secondary labeling information may be included in a captured image stored in the data storage unit 400 . As shown in FIG. 5 , first information, second information, third information, and secondary labeling information may be stored in the data storage unit 400 in a mapped state in one captured image.

In one embodiment, the first information is information about the location of a user playing cue polo, and may be applied to one side (ie, one of F1, F2, F3, and F4) of a pool table frame composed of four sides. The second information is information about the position of the first cushion contacted when the cue ball moves, and may be applied to one side (ie, any one of C1, C2, C3, and C4) among the four pool table cushions. The third information is information about the path of the cue ball that can score when the first information and the second information are specified. The primary labeling information may be applied as information for identifying the shape, size, position, color, etc. of a pool table frame, a pool table cushion, a white ball, a red ball, and a yellow ball included in a photographed image. For example, the primary labeling information is information indicating that an image having a specific size and located in a specific area within a photographed image is marked as a 'white ball'.

In addition, the secondary labeling information is information provided for the apparatus to move according to the third information, and is information required to successfully execute the path of the apparatus included in the third information. For example, the secondary labeling information may be applied to at least one selected from among the point and thickness of the cue ball and the strength of force during stroke. In this specification, the target point of the cue ball means the point where the end of the cue stick touches the cue ball, and the thickness means the degree to which the two balls overlap when the cue ball hits the object ball.

The position specifying unit 500 designates an arbitrary side of the pool table frame found in the video image of the billiard table 20 as the user's position (hereinafter, referred to as a 'first position'), and selects an arbitrary side of the billiard table cushions. can be designated as the position of the first cushion (hereinafter referred to as 'second position') that comes into contact when the apparatus moves. For example, the position specifying unit 500 designates a pool table frame having the closest vertical distance to the location coordinates of the cue ball shown in the video image as the first position, and makes contact with the cue ball when moving toward the point where the first object ball is located. The position of the first cushion to be used may be designated as the second position. Referring to FIG. 4 , the position specifying unit 500 may designate the first frame F1 as a first position and designate a third cushion C3 as a second position.

The image extractor 600 may extract a captured image having specific information from among captured images stored in the data storage 400 . Here, the specific information is information corresponding to the first location and the second location designated by the location specifying unit 500 . For example, as shown in FIG. 7, the image extraction unit 600 is in a state in which the first frame F1 is pre-determined as the first position and the third cushion C3 is pre-determined as the second position. A plurality of in-photographed images may be extracted.

The image selection unit 700 determines the relationship between the position coordinates of the white ball, red ball, and yellow ball included in the captured image extracted by the image extraction unit 600 and the position coordinates of the white ball, red ball, and yellow ball included in the video image. similarity can be judged. In an embodiment, the image selector 700 determines the similarity between location coordinates using a similarity algorithm (eg, cosine similarity, Euclidean distance, etc.), and selects a captured image having the highest similarity value according to the similarity algorithm. can be selected As shown in FIG. 7 , the image selector 700 may select a captured image (image processed by a dotted line) having the highest similarity value.

The result output unit 800 may check third information and secondary labeling information included in the captured image selected by the image selector 700 and output the corresponding contents. As shown in FIG. 8, in the selected photographic image, third information, which is information about the path of the cue ball capable of scoring, and hitting information necessary for the cue ball to move according to the third information (eg, score, thickness, stroke) magnitude of force) is included, the result output unit 800 may output this information.

2 is a flowchart schematically illustrating a billiard guide method according to an embodiment of the present invention. The billiard guide method according to an embodiment of the present invention will be described according to the flow chart shown in FIG. 2 and described with reference to the drawings shown in FIGS. 1 to 8, but will be described in order for convenience.

1. Video recording step <S201>

In this step, the photographing unit 100 may generate an image by capturing a billiard ball placed on the billiard table 20 in real time. The image generated in this step may be transmitted to the motion determination unit 200 .

2. Motion judgment step <S202>

In this step, the motion determination unit 200 may determine the motion of the billiard ball using video information captured by the photographing unit 100 in step S201. For example, the motion determination unit 200 extracts a video image from video information in a unit of time (eg, unit of second) and measures a change in RGB values of each pixel included in the extracted video image. When there is no change in the RBG value, it can be determined that the billiard ball is in a stationary state.

3. Object Recognition Model Creation Step <S203>

In this step, the processor may create an object recognition model based on an artificial intelligence neural network by analyzing and learning a plurality of photographed images of billiard balls disposed on the pool table 20 in advance. The object recognition model unit 300 according to an embodiment is an artificial intelligence neural network having an input layer, a hidden layer, and an output layer, and one or more deep learning algorithms are modularized. The object recognition model unit 300 may be equipped with an object recognition criterion created through analysis and learning of the shape of the frame of the pool table, the shape of the cushion of the pool table, the position of the billiard ball, and the color of the billiard ball included in the photographed image.

In addition, in the photographed image used for learning of the object recognition model unit 300 in this step, there are four billiard table frames F1 to F4 and four billiard table cushions C1 to C4. , primary labeling information for distinguishing white, red, and yellow balls may be included. The object recognition model unit 300 may establish object recognition criteria through a process of analyzing and learning primary labeling information included in a photographed image.

4. Data storage step <S204>

In this step, the processor may store the captured image including the secondary labeling information in the data storage unit 400 . Since the first information, the second information, and the third information are further included in the captured image stored in the data storage unit 400, the first information, the second information, the third information, and the secondary labeling information are included in one captured image. can be mapped together.

5. Billiard ball identification step <S205>

In this step, the object recognition model unit 300 may identify the position and color of the billiard ball included in the video image by using a pre-stored object recognition criterion, and distinguish between the cue ball and the object ball. For example, the object recognition model unit 300 subdivides and recognizes the frame and cushion of the pool table 20 included in the video image into four sides, recognizes the position and color of the billiard ball, and recognizes the identified billiard ball. Based on the position and color of the cue ball, it is possible to distinguish the cue ball from the object ball. In one specific example, the object recognition model unit 300 may determine a ball recognized as white as a cue ball and a ball recognized as red as a first object ball.

In addition, in this step, the object recognition model unit 300 calculates the location coordinates (P) of the billiard ball by using a bounding box (B), which is a virtual rectangle surrounding the billiard ball, to determine the position of the billiard ball included in the video image. can be derived. In one specific embodiment, the object recognition model unit 300 derives coordinates of vertices of the bounding box B, and determines the central point of the billiard ball by Equation 1 below.

[Equation 1]

(Where (X ₀ , Y ₀ ) is the center point of the billiard ball, X _min is the minimum value of the X coordinate among the coordinates of the vertex of the bounding box, X _max is the maximum value of the X coordinate among the coordinates of the vertex of the bounding box, Y _min is The minimum value of the Y coordinate among the coordinates of the vertex of the bounding box, Y _max is the maximum value of the Y coordinate among the coordinates of the vertex of the bounding box)

6. Route derivation step <S206>

In this step, a process of deriving a recommended route may be performed so that the cue polo can move on a route that can be scored. For example, in this step, an image similar to a video image is determined and extracted from a plurality of captured image information stored in the data storage unit 400, and route information combined with the extracted image information is derived as a recommended route. can The route derivation step according to an embodiment may include a location designation step, an image extraction step, a similar image selection step, and a result output step.

6-1. Positioning step <S2061>

In this step, the position specifying unit 500 designates an arbitrary side among the four sides of the pool table frame identified in the video image as the first position, and selects an arbitrary side among the four sides of the pool table cushion as the second position. can be specified as For example, the position specifying unit 500 designates the first frame F1 having the closest vertical distance to the position coordinates of the cue ball (white ball) shown in the video image as the first position, and the first object ball (red ball). The position of the first cushion C3 that comes into contact when the cue ball (white ball) moves toward the point where there is may be designated as the second position.

6-2. Image extraction step <S2062>

In this step, the image extraction unit 600 may extract images corresponding to the first location and the second location from among the captured images stored in the data storage unit 400 in step S204. For example, the image extraction unit 600 may extract captured images corresponding to the first and second positions designated in step S2061.

6-3. Similar image selection step <S2063>

In this step, the image selector 700 determines the location coordinates of the white ball, the red ball, and the yellow ball included in the photographed image extracted in step S2062, and the white ball included in the video image determined to be in a stationary state in step S202, The similarity between the location coordinates of the red ball and the yellow ball may be determined, and a photographed image having the highest similarity value according to the similarity algorithm may be selected. The similarity algorithm used in this step is not limited to a specific similarity algorithm, and various known similarity algorithms such as cosine similarity or Euclidean distance may be applied.

6-4. Result output step <S2064>

In this step, the result output unit 800 may output third information and secondary labeling information included in the captured image selected in step S2063. Therefore, in this step, the user can check the path of the cue ball capable of scoring and the strike information necessary for the cue ball to move along the path of the cue ball capable of scoring.

The billiard guide method according to various embodiments of the present invention may be implemented as computer readable code on a computer readable recording medium. A computer-readable recording medium includes all types of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage devices. In addition, computer-readable recording media may be distributed to computer systems connected through a network to store and execute computer-readable codes in a distributed manner.

For convenience of understanding, there are cases in which each of the aforementioned components is described as being used, but those skilled in the art may include a plurality of processing elements or a plurality of types of processing elements in a processing device. it can be seen that there is

As described above, according to various embodiments of the present invention, the billiard video image is automatically recognized through an object recognition model learned based on an artificial intelligence neural network, and intuitive results are derived because it is based on data labeling. It is possible. That is, in various embodiments of the present invention, in an actual game situation, the user is guided to a path that can be scored, and information necessary for the apparatus to move along the path (eg, score, thickness, strength of stroke, etc.) is presented. Therefore, it can induce interest of users and contribute to improvement of skills.

In addition, according to various embodiments of the present invention, the location of the user and the location of the first cushion are not specified because the location of the user and the location of the first cushion are first specified, and the image similarity is determined after extracting corresponding images. Compared to the process of determining image similarity in a state where the similarity is not detected, the similarity determination operation performance is optimized. In addition, according to various embodiments of the present invention, a billiard guide method optimized for a 3-cushion game can be presented.

As described above, the detailed description of the present invention has been made by the embodiments with reference to the accompanying drawings, but since the above-described embodiments have only been described as preferred examples of the present invention, the present invention is limited to the above embodiments. It should not be understood as being, and the scope of the present invention should be understood as the following claims and equivalent concepts thereof.

10: billiard guide system
100: shooting unit
200: motion determination unit
300: object recognition model unit
400: data storage unit
500: location designation unit
600: image extraction unit
700: image selection unit
800: result output unit
B: bounding box
P: position coordinates of the billiard ball
20: pool table
F1: 1st frame
F2: 2nd frame
F3: 3rd frame
F4: 4th frame
C1: 1st cushion
C2: 2nd cushion
C3: 3rd cushion
C4: 4th cushion

Claims (8)

A billiard guide method implemented by a computer including a memory in which computer executable instructions are stored and a processor that executes the computer executable instructions,
A billiard ball identification step of identifying the location and color of the billiard ball using the video image of the billiard ball and distinguishing between a cue ball and an object ball; and
Including; a path derivation step of deriving a recommended path so that the apparatus can move on a path that can be scored;
In the route derivation step, an image similar to the video image is determined and extracted from a plurality of previously stored image information, and route information combined with the extracted image information is derived as a recommended route;
The billiard guide method
An object recognition model generation step of creating an object recognition model based on an artificial intelligence neural network by analyzing and learning captured images made by photographing billiard balls placed on the billiard table; further comprising,
In the object recognition model generation step, the photographed image includes a pool table frame consisting of four sides, a pool table cushion consisting of four sides, and primary labeling information for distinguishing white balls, red balls, and yellow balls. would,
The billiard guide method
A data storage step of storing a captured image including secondary labeling information; further comprising,
In the data storage step, the captured image further includes first information, second information, and third information, and the first information is information about the position of a user playing water polo, among the four sides of the pool table frame It is applied to one side, and the second information is information about the position of the first cushion that contacts when the cue ball moves, and is applied to one side among the four side cushions of the billiard table. This is information about the path of the cue polo that can be scored when the first and second information are determined, and the secondary labeling information is information provided for the cue polo to move according to the third information, and the score, thickness and stroke of the cue ball are provided. Including at least one selected from the strength of the force,
The path derivation step is
A random side of the pool table frame consisting of four sides is designated as the user's position (hereinafter referred to as 'first position'), and any side of the pool table cushion consisting of four sides is the first contact when the cue ball moves. A position designation step of designating a position of a cushion (hereinafter, referred to as a 'second position');
an image extraction step of extracting images corresponding to the first and second positions from among the captured images stored in the data storing step;
The degree of similarity between the positional coordinates of the white ball, red ball, and yellow ball included in the photographed image extracted in the image extraction step and the positional coordinates of the white ball, red ball, and yellow ball included in the video image are determined, and the degree of similarity is determined by a similarity algorithm. a similar image selection step of selecting a photographed image having the highest value; and
And a result output step of outputting third information and secondary labeling information included in the captured image selected in the similar image selection step;
In the step of identifying the billiard ball, the position coordinates of the billiard ball are derived using a bounding box that is a virtual rectangle surrounding the billiard ball included in the video image, and the position coordinates of the billiard ball are the intersection of two diagonals of the bounding box. characterized in that applied
An image-based billiard guide method. delete delete According to claim 1,
In the billiard ball identification step, the frame and cushion of the billiard table included in the video image are subdivided and recognized into four sides through the object recognition model, and the cue ball and the object ball are distinguished based on the color of the identified billiard ball. characterized
An image-based billiard guide method. delete delete According to claim 1,
The billiard guide method
Further comprising a motion determination step of determining the motion of the billiard ball using a video of the billiard ball,
In the motion determination step, an image is extracted from the video at a predetermined time unit, and a change in RGB values for each pixel of the extracted image is measured, and when there is no change in the measured RGB values, it is determined that the billiard ball is in a stationary state. doing
An image-based billiard guide method. A computer-readable recording medium on which a program for performing the method according to any one of claims 1, 4, and 7 is recorded.

Images