Recognition system

Abstract

To provide a recognition system that can accurately recognize objects of multiple types. [Solution] The recognition system is a chip recognition system for a game hall having a game table, and comprises a game recording device that records the state of chips stacked on the game table as an image using a camera, and a chip determination device that analyzes the recorded image of the state of the chips to determine the number and type of chips bet by the player. The chip determination device further includes a function to store the image characteristics of a predetermined state of chips, and when it determines that the image obtained from the game recording device is an image of a predetermined state, it outputs and displays the determination result as "determination unknown".

Description

Technical Field

[0001] The present invention relates to a recognition system, and particularly to a recognition system for chips.

Background Art

[0002] In games such as baccarat games, bets are placed by a customer (player) stacking a plurality of chips on a table. Therefore, it is necessary to accurately recognize the stacked chips. Note that International Publication No. 2008 / 120749 discloses an example of a chip used in a game.

Summary of the Invention

[0003] An object of the present invention is to provide a recognition system capable of accurately recognizing objects having a plurality of types.

[0004] A chip recognition system according to a first aspect is a chip recognition system in a casino having a gaming table, a game recording device that records the state of chips stacked on the gaming table as an image by a camera, a chip determination device that performs image analysis on the recorded image of the chip state to determine the number and type of chips bet by a player, and includes the chip determination device further has a function of storing the features of an image of a predetermined state of a chip and outputting and displaying a determination result indicating that it is undetermined when it is determined that the image obtained from the game recording device at the time of determination is an image of the predetermined state.

[0005] In this configuration, the chip detection device stores images that result in low accuracy in reading the chips as images in a predetermined state. When it determines that an image obtained from the game recording device is in such a predetermined state, it does not force an answer, but instead outputs and displays a message indicating that the determination is unknown. This makes it possible to exclude the determination results from the determination of the number and type of chips that result in low accuracy in reading the chips (i.e., determination results that are likely to be incorrect). In other words, it becomes possible to determine the number and type of chips only from images that can be read accurately, and as a result, it becomes possible to recognize the chips with high accuracy.

[0006] The chip recognition system according to the second embodiment is a chip recognition system according to the first embodiment, The chip determination device includes an artificial intelligence device, which learns from multiple images used in past determinations when the chip determination device made a determination error, as training data. The chip determination device further includes a function to self-determine the accuracy of the determination based on images in which the determination result was incorrect as a result of the learning process, and to output and display as the determination result that the determination is undetermined for those in which the determination is questionable.

[0007] In this configuration, the artificial intelligence device of the chip detection device can improve the accuracy of its self-assessment by learning from multiple images used in past (incorrect) detections when an error occurred. This reduces the instances where images that can be accurately read are mistakenly output as "unknown". It can be done.

[0008] The chip recognition system according to the third embodiment is a chip recognition system according to the second embodiment, The chip determination device further includes a function to analyze the image from the game recording device when it determines that the determination is unclear, and to determine and store whether the cause of the unclear determination is due to the overlapping state of the chips stacked on the game table or whether part or all of a chip is hidden by other chips.

[0009] In this configuration, the dealer can easily identify the reason for the "undetermined" determination from the determination results stored in the chip determination device. This allows the dealer to quickly resolve the cause of the undetermined determination by rearranging the chip so that it is not obscured by other chips, or by neatly rearranging chips that are piled unevenly (the dealer may also caution the player if they are uncomfortable with the dealer touching their chips).

[0010] The chip recognition system according to the fourth embodiment is a chip recognition system according to any of the first to third embodiments, In order for the game record to be later analyzed by the chip detection device, the game recording device records the image acquired from the camera by adding an index or timestamp, or by adding a tag that identifies the stacking state of the chip.

[0011] In this configuration, the chip determination device can easily identify images of the chip state to be analyzed from the contents of the game recording device by utilizing the index, time, and tags attached to the image, thereby reducing the time required for identification.

[0012] The chip recognition system according to the fifth embodiment is a chip recognition system according to any of the first to fourth embodiments, The chip determination device includes a second artificial intelligence device, which learns from multiple images and chip information used in past determinations where the determination was correct in the chip determination device, using these as training data.

[0013] In this configuration, the second artificial intelligence device of the chip determination device can improve the accuracy of its determination of the number and type of chips by learning from multiple images and chip information used in past (correct) determinations in the chip determination device as training data.

[0014] The chip recognition system according to the sixth embodiment is a chip recognition system according to any one of the first to fifth embodiments, If the chip determination device determines that it cannot determine the chips, it analyzes images recorded by a camera other than the aforementioned camera to determine the number and type of chips bet by the player.

[0015] The chip recognition system according to the seventh embodiment is a chip recognition system according to any of the first to sixth embodiments, The chip detection device, when it detects the next chip without recognizing any chips at a certain interval or longer in the vertical direction, determines that the image is in the predetermined state and outputs and displays the result as "determination unknown".

[0016] The chip recognition system according to the eighth embodiment is a chip recognition system according to any of the first to seventh embodiments, The chip determination device compares the number of chips determined from the height of the chips with the number of chips determined by image analysis of the chip state image. If they differ, it determines that the image is in the predetermined state and outputs the determination result as "determination unknown".

[0017] The recognition system relating to the ninth aspect is: A recognition system that identifies objects of different types and determines the number of each type, where the object to be judged has multiple types. A recording device that records the state of the aforementioned object as an image using a camera, A determination device including an artificial intelligence device that performs image analysis on the recorded images of the objects and determines the number of each type of object, Equipped with, The determination device learns past determination results as teacher data, has a function of self-determining the accuracy in determination, and further has a function of self-determining that there is doubt in the determination result when the level of the accuracy is below a certain level, and outputting and displaying that fact as an unclear determination in the determination result.

[0018] According to such an aspect, the determination device learns past determination results as teacher data. For a new image, first, it self-determines the accuracy in determination. When the level of the accuracy is below a certain level, instead of forcibly giving an answer, it outputs and displays that fact as an unclear determination. Thereby, from the determination results of the number of each type of object, it is possible to exclude determination results (that is, determination results with a high possibility of being incorrect) when forcibly giving an answer for images with low accuracy in determination. That is, it becomes possible to determine the number of each type of object only from images that can be accurately read, and as a result, it becomes possible to accurately recognize the object.

Brief Explanation of Drawings

[0019] [Figure 1] FIG. 1 is a diagram schematically showing a gaming hall equipped with a chip recognition system according to an embodiment. [Figure 2] FIG. 2 is a diagram for explaining the progress of a baccarat game. [Figure 3] FIG. 3 is a block diagram showing the schematic configuration of a chip recognition system according to an embodiment. [Figure 4] FIG. 4 is a flowchart for explaining a chip recognition method according to an embodiment. [Figure 5] FIG. 5 is a diagram for explaining the case where another chip is hidden behind a certain chip. [Figure 6] FIG. 6 is a diagram for explaining the case where the chips are stacked in a jagged manner.

Embodiments for Implementing the Invention

[0020] Embodiments of the present invention will be described in detail below with reference to the attached drawings. In each drawing, components having equivalent functions are denoted by the same reference numerals, and detailed descriptions of components with the same reference numerals will not be repeated.

[0021] In the embodiments described below, we will explain a chip recognition system in a gaming hall with a gaming table as an example of a recognition system that identifies objects by type and determines the number of each type. However, it goes without saying that the objects to be judged are not limited to chips, as long as they have multiple types.

[0022] First, referring to Figures 1 and 2, a game played in a game hall having a game table 4 will be described. In this embodiment, the game table 4 is a baccarat table, and an example of a baccarat game being played will be described, but the present invention is also applicable to other game halls or other games. It is Noh.

[0023] Figure 1 is a schematic diagram showing a gaming area equipped with a chip recognition system 10 according to one embodiment. As shown in Figure 1, the gaming area has a roughly semicircular gaming table 4 and a number of chairs 201 arranged along the arc side of the gaming table 4, facing the dealer D. The number of chairs 201 is arbitrary, and in the example shown in Figure 1, there are six chairs 201. In addition, a betting area BA is provided on the gaming table 4 corresponding to each chair 201. That is, in the illustrated example, six betting areas BA are arranged in an arc shape.

[0024] As shown in Figure 1, each of the 201 chairs is occupied by a customer (player) C. Customers (players) C place their bets on whether the baccarat game will end with the player (PLAYER) or the banker (BANKER), or whether it will be a tie (TIE), by stacking chips W in the betting area BA located in front of the chair 201 where they are seated (hereinafter referred to as "betting").

[0025] The chips W to bet may be of one type or multiple types. The number of chips W to bet may be arbitrarily determined by the customer (player) C. The chip recognition system 10 according to this embodiment recognizes the number and types of these stacked chips W.

[0026] Dealer D, timing his call to end betting by player C, calls out "No More Bet" and makes a sideways movement of his hand. Next, Dealer D draws cards one by one from the card shooter S and places them on the game table 4. As shown in Figure 2, the first card goes to the player, the second to the banker, the third to the player, and the fourth to the banker (hereinafter, the drawing of the first to fourth cards will be referred to as "dealing").

[0027] Furthermore, all cards are drawn from the card shooter device S with their backs facing upwards. Therefore, neither the dealer D nor the player C can determine the rank (number) or suit (hearts, diamonds, spades, clubs) of the drawn cards.

[0028] After the fourth card is drawn, the player C who bet on the Player (or the player with the highest bet if there are multiple players who bet on the Player, or the Dealer D if there are no players who bet on the Player) turns the first and third cards, which are face down, face up. Similarly, the player C who bet on the Banker (or the player with the highest bet if there are multiple players who bet on the Banker, or the Dealer D if there are no players who bet on the Banker) turns the second and fourth cards face up (this act of turning face down cards face up is generally called a "squeeze").

[0029] Then, based on the rank (number) of the first four cards and the detailed rules of the baccarat game, the dealer D draws a fifth card, and then a sixth card, which become either the player's or the banker's hand. Similarly, the player C who bet on the player squeezes the cards that become the player's hand, and the player who bet on the banker squeezes the cards that become the banker's hand.

[0030] After the first four cards are drawn, the fifth and sixth cards are squeezed to determine the winner. The time until the results are known is the most exciting part for player C.

[0031] Furthermore, depending on the rank (number) of the cards, the winner may be decided after the first to fourth cards, or it may take five, or even six cards to be decided. Dealer D understands when the winner has been decided and the result based on the rank (number) of the squeezed cards, and performs tasks such as pressing the result display button on the card shooter device S to display the result on the monitor to inform the customer (player) C.

[0032] At the same time, the win / loss determination unit of the card shooter device S determines the outcome of the game. If the dealer D attempts to draw another card without displaying the win / loss result even though the winner has been determined, an error occurs. The card shooter device S detects this error and outputs an error signal. Finally, while the win / loss result is displayed, the dealer D settles the chips bet by the customers (players) C, paying the winning customers (players) C and collecting the chips bet by the losing customers (players) C. After settlement is complete, the display of the win / loss result ends and betting for the next game begins.

[0033] The above flow of the baccarat game is widely practiced in general casinos. The card shooter device S is an existing card shooter device that has a structure in which the dealer D draws out the cards, is configured to read the drawn cards, and also has result display buttons and a result display unit, and is equipped with functions for determining wins and losses and displaying the results. In a typical casino floor, a card shooter device S and monitors are placed at each of the multiple gaming tables 4 lined up, and the cards to be used are supplied to each gaming table 4 or the cabinet below it in packages, sets, or even cartons.

[0034] The chip recognition system 10 according to this embodiment relates to a system for recognizing chips W stacked and placed in the betting area BA by a customer (player) C, and more specifically, to a system for recognizing the number and / or type of chips W.

[0035] As shown in Figure 1, in this embodiment, a surveillance camera 212 that captures images of the state of the chips W stacked in the betting area BA is provided outside the gaming table 4. In addition, each chip W is equipped with an RFID, and the chip tray 23 managed by the dealer D is equipped with an RFID reader 22 that reads the RFID of the chips W in the chip tray 23.

[0036] The chip recognition system 10 according to this embodiment is communicated to the surveillance camera 212 and the RFID reader 22, respectively.

[0037] Figure 3 is a block diagram showing the schematic configuration of the chip recognition system 10 according to this embodiment. As shown in Figure 3, the chip recognition system 10 includes a game recording device 11, a chip determination device 12, and a determination success / failure determination device 14. At least a part of the chip recognition system 10 is implemented by a computer.

[0038] The game recording device 11 includes a fixed data storage device, such as a hard disk. The game recording device 11 records the state of the chips W stacked on the game table 4 as images captured by the camera 212. The images may be moving images or a series of still images.

[0039] The game recording device 11 assigns an index or timestamp to the images acquired from the camera 212 so that the game records can be analyzed later by the chip detection device 12, which will be described later. Alternatively, the scene of chip W being collected or paid may be recorded with a tag that identifies the specific scene.

[0040] The chip determination device 12 analyzes images of the chip state W recorded in the game recording device 11 to determine the number and type of chips W bet by the customer (player) C. The chip determination device 12 may include an artificial intelligence device that performs image recognition using, for example, deep learning technology.

[0041] By the way, if the chips W1 to W6 bet by customer (player) C are stacked in multiple piles (see Figure 5), or if the stacking of chips W1 to W6 is messy and jagged (see Figure 6), the camera 212 cannot see all of chips W1 to W6, which may reduce the accuracy of reading chips W1 to W6.

[0042] More specifically, as shown in Figure 5 with the camera 212 labeled (A), raising the height of the camera 212 makes it less likely for chips W1 to W4 in the back pile to be hidden by chips W5 and W6 in the front pile, even if chips W1 to W6 are stacked in separate piles on the front and back. However, as shown in Figure 6 with the camera 212 labeled (A), if the stacking of chips W1 to W6 is messy and jagged, one chip W1 may be hidden by another chip W2 above it, or one chip W3 may be hidden by another chip W4 or W5 above it, making it difficult to read all of chips W1 to W6.

[0043] Conversely, if the height of camera 212 is lowered, as shown in Figure 6 with the label (B), it becomes easier to read all of chips W1 to W6 even if they are stacked haphazardly and unevenly. However, if chips W1 to W6 are stacked in separate piles, such as camera 212 shown in Figure 5 with the label (B), chips W1 to W6 in the back pile tend to be hidden by chips W5 and W6 in the front pile, making it difficult to read all of chips W1 to W6.

[0044] Furthermore, even if there are no problems with the stacking of chips W1 to W6, if halation (a phenomenon where ambient light enters and the image becomes white) occurs in camera 212, the image contrast will decrease, resulting in a lower accuracy rate for reading chips W1 to W6.

[0045] Conventional artificial intelligence (AI) devices will force (and likely produce incorrect) answers to images with low reading accuracy, resulting in mismatches between the chips bet by customer (player) C and the chips in the chip tray. Stopping the game every time a mismatch occurs due to an AI reading error would be inefficient.

[0046] Taking these points into consideration, the chip determination device 12 in this embodiment further includes an artificial intelligence device (artificial intelligence device 12a for chip information determination) that recognizes image patterns with a low accuracy rate (artificial intelligence device 12b for image pattern recognition) in addition to an artificial intelligence device (artificial intelligence device 12a for chip information determination) that determines the type and number of stacked chips W.

[0047] The artificial intelligence device 12a for determining chip information analyzes images of the state of chips W recorded in the game recording device 11 to determine the number and type of chips W bet by customer (player) C. The artificial intelligence device 12a for determining chip information may further determine the position of the chips W bet by customer (player) C on the betting area BA.

[0048] The artificial intelligence device 12a for determining chip information analyzes images of the state of chips W recorded in the game recording device 11 to determine the state of chips W in the chip tray 23 before settlement for each game. You may also determine the number and type of cards.

[0049] As shown in Figure 3, the chip detection device 12 outputs the detection result to the output device 15. The output device 15 may output the detection result of the chip detection device 12 as text information to a monitor on the game table 4, or as audio information to the dealer D's headset, etc.

[0050] Furthermore, the artificial intelligence device 12b for image pattern recognition stores the characteristics of images of chips W in predetermined states, and determines whether the image obtained from the game recording device 11 is an image of that predetermined state. Here, "an image of chips W in a predetermined state" refers to an image in which, when the number and type of chips are determined by image analysis, the level of accuracy of the determination may fall below a certain level, i.e., an image in which the determination is questionable. Specifically, these include, for example, images of chips W1 to W6 stacked in multiple piles captured and stored by a camera 212 located at a low position (see camera 212 indicated by the code (B) in Figure 5), images of chips W1 to W6 stacked haphazardly and jaggedly captured and stored by a camera 212 located at a high position (see camera 212 indicated by the code (A) in Figure 6), and images where halation has occurred.

[0051] When the chip determination device 12 determines that the image obtained from the game recording device 11 is in a predetermined state by the artificial intelligence device 12b for image pattern recognition, that is, when it determines that the level of accuracy of the determination falls below a certain level, it outputs the determination result as "determination unknown" to the output device 15.

[0052] Furthermore, the chip determination device 12 may also be equipped with a function to analyze the image obtained from the game recording device 11 when it determines that the determination is unclear, and to determine and store whether the cause of the unclear determination is (1) the overlapping state of the chips stacked on the game table, or (2) part or all of chip W being hidden by other chips.

[0053] Referring to Figures 5 and 6, if the chip determination device 12 determines that the image obtained from camera 212 indicated by (A) or camera 212 indicated by (B) is in a predetermined state and therefore cannot be determined, it may use an image captured by another camera 212 indicated by (C) to read the chip W. Viewing from a different angle using cameras in different orientations or positions allows for a more objective view of the chip W. In particular, if the reason for the determination being unclear is that part or all of chip W is hidden by other chips, as shown in Figure 5, using the camera 212 on the opposite side indicated by (C) will prevent the chip from being hidden by other chips. Furthermore, the chip determination device 12 may output the reading results of the chip W using the images captured by each camera 212. In that case, the chip determination device 12 may output the accuracy of the determination of each reading result together, or it may consider the result with the highest number of reads to be the most likely to be correctly recognized.

[0054] When the chip detection device 12 counts chips W, it may determine that it cannot determine the chip if it detects the next chip W within a cluster of chips W without recognizing any chips at a certain vertical interval or longer. In other words, if it detects the next chip W without recognizing any chips at a certain vertical interval or longer, it is highly likely that the chips in between are hidden and not visible.

[0055] The chip determination device 12 may be configured to output a determination result indicating that determination is not possible if the number of chips W determined from the height of the chips W is different from the result obtained when determining the type and number of chips W. The number of chips may also be determined by determining a specific point (such as the center of the outline of the topmost chip) from the shape of the chip and using a method such as triangulation. .

[0056] The judgment correctness determination device 14 is a device that determines whether the judgment result of the chip judgment device 12 is correct or incorrect. The judgment correctness determination device 14 grasps the actual total amount V0 of chips W in the chip tray 23 when the settlement of chips bet by customer (player) C is complete, that is, when all payments to winning customer (player) C and the collection of chips W (losing chips) bet by losing customer (player) C are complete.

[0057] In this embodiment, the judgment success / failure determination device 14 acquires RFID information of the chips W in the chip tray 23 from the RFID reader 22, and based on the acquired RFID information, determines the type and number of chips W in the chip tray 23 after settlement for each game, and grasps the actual total amount V0.

[0058] Furthermore, the judgment success / failure determination device 14 obtains information on the number and type of chips W as a judgment result from the chip judgment device 12, and based on the obtained information on chips W, calculates the total amount of chips W (winning chips) bet by the winning customer (player) C (i.e., the amount deducted from the chip tray 23 in that game) V2 and the total amount of chips W bet by the losing customer (player) C (i.e., the amount increased from the chip tray 23 in that game) V3. The judgment success / failure determination device 14 then subtracts the amount deducted from the chip tray 23 in that game V2 from the total amount of chips W in the chip tray 23 before settlement of each game V1, and adds the amount increased from the chip tray 23 in that game V3 to calculate the total amount of chips that should be in the chip tray 23 V4 (=V1-V2+V3).

[0059] The judgment correctness determination device 14 compares the expected total value V4 of the chips W in the chip tray 23 with the actual total value V0 of the chips W in the chip tray 23. If there is a difference between the expected total value V4 and the actual total value V0 (V4 ≠ V0), the device determines that the judgment result of the chip determination device 12 was incorrect. On the other hand, if the expected total value V4 and the actual total value V0 match (V4 = V0), the device determines that the judgment result of the chip determination device 12 is correct.

[0060] The chip determination device 12 obtains the correctness of its determination result from the determination correctness judgment device 14. If the determination correctness judgment device 14 determines that the determination result of the chip determination device 12 is correct, the artificial intelligence device 12a for chip information determination learns the image used in the past (correct) determination when the determination was correct, and the information on the number and type of chips W as the (correct) determination result, as training data. By repeatedly performing this learning, the artificial intelligence device 12a for chip information determination can improve the accuracy of determining the number and type of chips W.

[0061] On the other hand, if the judgment correctness determination device 14 determines that the judgment result of the chip judgment device 12 is incorrect, the artificial intelligence device 12b for image pattern recognition learns the image used in the past (incorrect) judgment when the judgment was incorrect as training data for "images in a predetermined state". Images in a predetermined state (such as an image where one chip is hidden behind another, an image where chips are stacked in a jagged pattern, or an image with halation) may be selected by a person and used to train the artificial intelligence device 12b for image pattern recognition. Images in a predetermined state (such as an image where one chip is hidden behind another, an image where chips are stacked in a jagged pattern, or an image with halation) may be intentionally created by a person or artificial intelligence and used to train the artificial intelligence device 12b for image pattern recognition. By repeatedly performing such learning, the artificial intelligence device 12b for image pattern recognition becomes able to accurately extract images where the accuracy of the judgment may fall below a certain level, that is, it can improve the accuracy of its self-judgment when self-judging the accuracy of the judgment.

[0062] Next, with reference to Figure 4, an example of the operation (chip recognition method) of the chip recognition system 10 according to this embodiment will be described.

[0063] As shown in Figure 4, first, when a customer (player) C places chips W in a stack in the betting area BA of the game table 4 (the chips W are bet), the game recording device 11 captures and records the state of the stacked chips W as an image using the camera 212 (step S31).

[0064] Next, the chip determination device 12 acquires the image recorded in the game recording device 11. The image acquired by the chip determination device 12 may be selected by the game recording device 11 based on an index, time, or a tag that identifies the chip W collection scene or payment scene.

[0065] In the chip determination device 12, the artificial intelligence device 12b for image pattern recognition determines whether the image obtained from the game recording device 11 is an image in a predetermined state (step S32). More specifically, as described above, the artificial intelligence device 12b for image pattern recognition learns from multiple images used in past determinations where the chip determination device 12 made a judgment error, using these images as training data. As a result of this learning, it performs a self-assessment of the accuracy of the chip W determination based on the images in which the determination result was incorrect, and determines whether the accuracy of the determination falls below a certain level.

[0066] If the artificial intelligence device 12b for image pattern recognition determines that the image obtained from the game recording device 11 is in a predetermined state (step S33: YES), the chip determination device 12 outputs this as the determination result to the output device 15 (step S40). The determination result of the chip determination device 12 may be output as text information to the monitor on the game table 4 by the output device 15, or as audio information to the dealer D's headset, etc.

[0067] On the other hand, if the artificial intelligence device 12b for image pattern recognition determines that the image obtained from the game recording device 11 is not an image of a predetermined state (step S33: NO), the artificial intelligence device 12a for chip information determination analyzes the image of the state of the chips W recorded in the game recording device 11 to determine the number and type of chips W bet by the customer (player) C (step S34).

[0068] In step S34, the chip determination device 12 may perform image analysis on the image of the state of the chips W recorded in the game recording device 11 to determine the number and type of chips W bet by customer (player) C, as well as the position of the chips W bet by customer (player) C on the betting area BA, or it may determine the number and type of chips W in the chip tray 23 before settlement of each game.

[0069] The number and type of chips W determined by the chip determination device 12 are output to the output device 15 (step S35). The determination result of the chip determination device 12 may be output as text information to the monitor on the game table 4 by the output device 15, or as audio information to the dealer D's headset, etc.

[0070] The information regarding the number and type of chips W determined by the chip determination device 12 is also input to the determination correctness determination device 14. The determination correctness determination device 14 determines whether the determination result of the chip determination device 12 is correct or incorrect (step S36).

[0071] If the judgment correctness determination device 14 determines that the judgment result of the chip determination device 12 is correct (step S37: YES), the image used for the (correct) judgment of the chip determination device 12, and the information on the number and type of chips W as the (correct) judgment result are input as training data to the artificial intelligence device 12a for chip information determination, and the artificial intelligence device 12a for chip information determination performs learning (step S38).

[0072] On the other hand, if the judgment correctness determination device 14 determines that the judgment result of the chip judgment device 12 was incorrect (step S37: NO), the image used for the (incorrect) judgment of the chip judgment device 12 is input to the artificial intelligence device 12b for image pattern recognition as training data for "images in a predetermined state," and the artificial intelligence device 12b for image pattern recognition performs learning (step S39).

[0073] Artificial intelligence can make significant mistakes when there is an error in its judgment (it will confidently give the wrong answer), so by training the AI ​​on patterns of images that are prone to being mistaken, it can be made to recognize those patterns.

[0074] As described above, according to this embodiment, the chip determination device 12 stores images that result in low accuracy in reading the chips W as images in a predetermined state. When determining the number and type of chips W, if the device determines that the image obtained from the game recording device 11 is an image in such a predetermined state, it does not force an answer but instead outputs and displays a message indicating that the determination is unknown. This makes it possible to exclude determination results (i.e., determination results that are likely to be incorrect) that were forced to answer for images that result in low accuracy in reading the chips W from the determination results for the number and type of chips W. In other words, it becomes possible to determine the number and type of chips W only from images that can be read accurately, and as a result, it becomes possible to recognize the chips W with high accuracy.

[0075] In other words, for example, if the chip detection device 12 detects 1000 images and the accuracy rate is 99.9%, and in the 0.1% of incorrect cases, the accuracy rate is low because 9 out of 10 images have shaded chips or the chips are stacked in an uneven manner, then the accuracy rate can be increased by another level by excluding such cases from the denominator.

[0076] Furthermore, according to this embodiment, the artificial intelligence device 12b for image pattern recognition can improve the accuracy of its self-assessment when it makes a self-assessment of the accuracy of its judgment by learning from multiple images used in past (incorrect) judgments when an error occurred in the judgment. This reduces the number of cases in which images that could be accurately read are mistakenly output and displayed as "unknown judgment".

[0077] Furthermore, according to this embodiment, when the chip determination device 12 determines that the determination is unclear, it determines and stores whether the cause of the unclear determination is (1) the overlapping state of the chips W stacked on the game table 4, or (2) part or all of the chips W being hidden by other chips W. This allows the dealer D to easily confirm the cause of the unclear determination. As a result, the dealer D can quickly resolve the cause of the unclear determination by rearranging the chips W so that they are not hidden by other chips W, or by neatly rearranging the chips W that are stacked unevenly (if the customer (player) C is uncomfortable with the dealer D touching the chips W, the dealer D may warn the customer (player) C).

[0078] Furthermore, according to this embodiment, the game recording device 11 assigns an index or time to the image acquired from the camera 212, or a tag that identifies the stacking state of the chip W. By assigning and recording a tag, the chip determination device 12 can easily identify the image of the chip W to be analyzed from the contents recorded by the game recording device 11 by using the index, time, and tag assigned to the image, thereby reducing the time required for identification.

[0079] Furthermore, according to this embodiment, the artificial intelligence device 12a for determining chip information can improve the accuracy of its determination when determining the number and type of chips W by learning from multiple images used in past (correct) determinations when the chip determination device 12 was correct, and from information on chips W as (correct) determination results, as training data.

[0080] The embodiments described above are intended to enable persons with ordinary skill in the art to implement the present invention. Various modifications of the above embodiments can be made naturally by those skilled in the art, and the technical idea of ​​the present invention can be applied to other embodiments as well. Therefore, the present invention is not limited to the embodiments described, but should be in the broadest scope according to the technical idea defined by the claims.

Claims

1. A chip recognition system for a gaming parlor having gaming tables, A game recording device that records the state of the chips stacked on the aforementioned game table as an image using a camera, A chip detection device is provided, The chip determination device is The function includes an image analysis of the recorded chip state to determine the number and type of chips bet by the player, A function that stores the image characteristics of a predetermined state of the chip, and determines whether the image obtained from the game recording device is an image of the predetermined state at the time of determination, A chip recognition system comprising a function that outputs and displays as the determination result "determination unknown" when it is determined that the image obtained from the game recording device is an image of the predetermined state.

2. A chip recognition system according to claim 1, The chip determination device includes an artificial intelligence device, and the artificial intelligence device learns from multiple images used in past determinations when the chip determination device made a determination error, as training data. The chip recognition system further includes a function that self-determines the accuracy of the determination based on images in which the determination result was incorrect as a result of the learning process, and outputs a determination result indicating that the determination is unknown for those in doubt.

3. A chip recognition system according to claim 2, The chip recognition system further includes a function to analyze the image from the game recording device when the chip determination device itself determines that the determination is unclear, and to determine and store whether the cause of the determination being unclear is due to the overlapping state of the chips stacked on the game table or whether part or all of a chip is hidden by other chips.

4. A chip recognition system according to any one of claims 1 to 3, The game recording device is a chip recognition system that records images acquired from a camera by assigning an index or timestamp, or by assigning a tag that identifies the stacking state of the chip, so that the record of the game can be later analyzed by the chip detection device.

5. A chip recognition system according to any one of claims 1 to 4, The chip recognition system includes a second artificial intelligence device, the second artificial intelligence device learns from multiple images and chip information used in past judgments where the chip recognition device made a correct judgment, as training data.

6. A chip recognition system according to any one of claims 1 to 5, The chip determination device, when it determines that the determination is unclear, is a chip recognition system that analyzes images recorded by a camera other than the aforementioned camera to determine the number and type of chips bet by the player.

7. A chip recognition system according to any one of claims 1 to 6, The chip detection device is a chip recognition system that, when it detects the next chip without recognizing any chips at a certain interval or more in the vertical direction, determines that the image is in the predetermined state and outputs and displays the result as "determination unknown".

8. A chip recognition system according to any one of claims 1 to 7, The chip recognition system compares the number of chips determined from the height of the chips with the number of chips determined by image analysis of an image of the chip's state. If the numbers are different, the system determines that the image is of the predetermined state and outputs the result as "undetermined".

Images