Intelligent go replaying system based on multi-dimensional analysis
The Go intelligent review system uses multi-dimensional analysis to automatically process game data, identify key moves and conduct quantitative analysis, solving the subjective and inefficiency issues of traditional Go review and achieving efficient and real-time review feedback.
Patent Information
- Application Number
- CN202510769736.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-10
- Publication Date
- 2025-09-26
AI Technical Summary
Traditional Go game review relies on human experience, is highly subjective, inefficient, and lacks quantitative analysis.
The intelligent Go review system based on multi-dimensional analysis obtains game data through video recording, extracts images of chess piece placement, processes the game in segments, determines key moves, and marks the player's thinking coefficient and game impact index.
It reduces the subjectivity of Go game review, improves review efficiency and feedback efficiency, and realizes full process automation and real-time analysis.
Smart Images
Figure CN120708121A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of integrated Go intelligent review technology, and in particular to a Go intelligent review system based on multi-dimensional analysis. Background Art
[0002] As a complex strategic board game, Go requires in-depth analysis of strategic decisions, move efficiency, situational evolution, and key move judgments. Traditional Go review relies heavily on the experience and intuition of human players, which has the following limitations:
[0003] Highly subjective: Human chess players may ignore the potential impact of certain key moves due to limited perspective or lack of experience.
[0004] Inefficiency: Manual review takes a lot of time and it is difficult to fully cover all dimensions of the chess game.
[0005] Insufficient quantification: There is a lack of precise quantitative analysis of indicators such as move efficiency, point count changes, and win rate fluctuations.
[0006] Therefore, there is an urgent need for an intelligent Go review system based on multi-dimensional analysis to solve the above problems. Summary of the Invention
[0007] The purpose of the present invention is to provide an intelligent Go review system based on multi-dimensional analysis to solve the technical problems of existing Go review systems such as strong subjectivity, low efficiency and untimely feedback.
[0008] The purpose of the present invention can be achieved through the following technical solutions:
[0009] An intelligent Go review system based on multi-dimensional analysis, the system comprising:
[0010] A chess game data acquisition module is used to obtain a video recording of the chess game played by a chess player and determine the chess pieces held by the chess player. Based on the video recording and the chess pieces held, the images of the chess pieces held by the chess player are extracted in the order in which the chess pieces are placed by the chess player.
[0011] A game segmentation processing module is used to determine the number of occupied points held by the chess pieces of the chess player and the number of occupied points held by the opponent's chess pieces in each chess piece placement image, and to segment the game period based on the number of occupied points held by the chess pieces of the chess player to obtain game segments;
[0012] The key move determination module is used to determine the key moves of each game segment, obtain the player's thinking coefficient corresponding to the key moves and the game influence index of the key moves;
[0013] The replay marking module is used to mark the key moves of the game segments based on the player's thinking coefficient and the game impact index of the key moves.
[0014] Furthermore, extracting the images of the chess pieces held by the chess players in the order in which they place the pieces based on the video recording and the chess pieces held by the chess players specifically includes the following process:
[0015] Extracting the chess pieces held by the chess player in the order of chess player placement based on the video recording and the chess pieces held, wherein the chess pieces held by the chess player are frames containing other backgrounds;
[0016] Perform super-pixel segmentation on the selected drop frame to obtain a super-pixel drop image;
[0017] The features of the chess piece dropped are extracted from the super-pixel chess piece dropped image, a detection bounding box of the chess piece held by the chess player is determined based on the features of the chess piece dropped image, and the dropped image of the chess piece held by the chess player in the video recording is determined based on the detection bounding box.
[0018] Furthermore, the super-pixel segmentation of the selected sub-frame to obtain the super-pixel sub-image specifically includes the following processes:
[0019] Step 1: Convert the selected sub-frame to the CIELAB color space and combine the color vector and X and Y coordinates into a 5-dimensional feature vector. The CIELAB color space consists of three elements: brightness and color, A and B. A represents the color range from magenta to green, and B represents the color range from yellow to blue.
[0020] Step 2: Initialize the seed points and evenly distribute them in the image. The number of seed points is the same as the number of superpixels. The position of the seed point is the center of the superpixel. Assume that the selected subframe has N pixels, the number of superpixels is K, the size of each superpixel is N / K, and the distance between superpixels is
[0021] Step 3: Adjust the seed point position and select the position with the smallest gradient within the n×n range of the seed point as the new seed point position, where n is a preset value.
[0022] Step 4: Assign labels to pixels, i.e., the cluster centers corresponding to the pixels. Assign labels to each pixel in the 2S×2S region of each seed point, initialize the superpixel category array and distance array, record the seed point corresponding to the pixel and the distance from each pixel to the seed point, and calculate the distance from the pixel to the seed point in the 2S×2S region of each seed point. If it is less than the distance in the original distance array, update the distance array and superpixel category array.
[0023] Step 5: Iterative optimization, repeating steps 3 and 4 until the cluster center of each pixel no longer changes, and obtaining a superpixel image.
[0024] Furthermore, determining the detection bounding box of the chess piece held by the player based on the features of the chess piece placement image specifically includes the following process:
[0025] The image feature F of the move and the learnable target query F q and the temporal features of the image F C Input into the target decoder composed of cross attention and feedforward network, and the final target feature F is obtained by calculating the cross attention H , which is expressed as follows:
[0026] Q E =F q +F C ;K E =F;V E =F;
[0027]
[0028] F H =Layer→norm(FFN(F H1 )+F); where d E Represents feature F H1 Dimension; FFN represents feedforward network, Layer→norm(·) represents layer normalization processing;
[0029] The final target feature F H The target feature detection frame is recorded as the detection bounding box of the chess piece held by the chess player.
[0030] Furthermore, determining the number of occupied points of the chess pieces held by the chess player and the number of occupied points of the opponent's chess pieces in the chess piece placement images based on each chess piece placement image specifically includes the following process:
[0031] Use edge detection and Hough transform to detect chessboard lines and locate chessboard grids;
[0032] Live shape detection and dead shape judgment: Connected area analysis: Use DFS / BFS to traverse the connected area of the chess pieces and mark live and dead shapes; Live shape condition verification: Check whether the live shape contains at least two independent air holes, and the chess pieces are connected to form an indivisible whole; Dead shape judgment: Chess pieces that are completely surrounded by the enemy and cannot form two eyes are marked as dead;
[0033] Use the flood-fill algorithm to mark all empty points within the boundaries of the active shape. Count the number of marked empty points, which is the number of points of the active shape. Count the number of your own chess pieces in the active shape. Traverse all active shapes, accumulate the number of points and the number of pieces, and get the total number of occupied points for the player and the opponent.
[0034] Furthermore, the game period is segmented based on the number of occupied points held by the chess pieces of the chess players. The game segmentation specifically includes the following process:
[0035] The order of chess moves corresponding to the number of occupied points held by the chess pieces of the chess player is used as the horizontal axis, and the number of occupied points held by the chess pieces of the chess player corresponding to the order of chess moves is used as the vertical axis. A rectangular coordinate system is established, and a curve of the number of occupied points is generated by tracing points. The slopes of the curves of the number of occupied points between adjacent chess moves are counted, and the slopes are arranged according to the order of chess moves. The first two adjacent chess moves where the slope decreases are marked as the first-division chess moves, the chess moves corresponding to the chess moves in the first-division chess move with a later order are recorded as the first-division time point, the second two adjacent chess moves where the slope decreases for the second time are marked as the second-division chess moves, and the chess moves corresponding to the chess moves in the second-division chess move with a later order are recorded as the second-division time point. The period from the start time point of the chess game to the first-division time point is recorded as the layout game segment, the period from the first-division time point to the second-division time point is recorded as the mid-game battle game segment, and the period from the second-division time point to the end time point is recorded as the endgame game segment.
[0036] Furthermore, the key moves of each game segment specifically include the following processes:
[0037] Obtain a curve of the number of occupied points for each game segment, and record as key moves any move in which the number of occupied points exceeds a first preset threshold or is lower than a second preset threshold, where the first threshold is greater than the second threshold.
[0038] Furthermore, obtaining the player's thinking coefficient corresponding to the key move and the game influence index of the key move specifically includes the following process:
[0039] Obtain the player's thinking time corresponding to the key move, where the player's thinking time corresponding to the key move is the period between the opponent's move and the player's completion of the key move; calculate the difference between the player's thinking time corresponding to the key move and the preset thinking time, and record the ratio of the difference to the game period as the player's thinking coefficient;
[0040] Establish a hierarchical model with a step-by-step structure, where the layers include the target layer, the criterion layer, and the measure layer; construct a judgment matrix: determine the number m of relevant influencing factors at each layer, and construct a set U of chess game influencing factors, U = {φ1, φ2, φ3}, where φ1 is the subset of the change in the AI win rate before and after the move, φ2 is the subset of the number of points before and after the move, and φ3 is the subset of the move efficiency. From the set U, take two subsets from the same layer and compare them. Use M to represent the importance ratio, and assign the corresponding importance according to the preset ratio. Combine the importance of each layer to form a judgment matrix;
[0041] Calculate the maximum eigenvalue γ of the judgment matrix:
[0042] Among them, vαβ It is the matrix obtained by normalizing each column vector of the judgment matrix. The values of α and β are 1, 2...m, w α is the matrix v αβ The elements of are added row by row to obtain the vector and then normalized into a matrix, w β is the matrix v αβ Add the elements of column by column to obtain the vector and then normalize the matrix;
[0043] Calculate the game impact index CI of key moves: Where k represents the order of the judgment matrix.
[0044] Furthermore, the key move marking process of dividing the game into segments based on the player's thinking coefficient and the game influence index of the key move specifically includes the following steps:
[0045] Calculate the product of the player's thinking coefficient and the game influence index of the key hand after normalization, load the product value threshold, where the product value threshold is set by the system and stored in the system, and judge whether the product value exceeds the product value threshold. If so, mark the key hands of the game segment; if not, do not mark the key hands of the game segment.
[0046] Compared with the existing solutions, the present invention achieves the following beneficial effects:
[0047] The present invention obtains a video recording of a chess game to which a chess player belongs and determines the chess pieces held by the chess player, and extracts the chess piece placement images of the chess pieces held by the chess player in the order in which the chess player places the chess pieces based on the video recording and the chess pieces held; determines the number of occupied points of the chess pieces held by the chess player and the number of occupied points of the opponent's chess pieces in the chess piece placement image based on each of the chess piece placement images, and divides the game time period into segments based on the number of occupied points of the chess pieces held by the chess player to obtain game segments; determines the key hands of each game segment, obtains the chess player thinking coefficient corresponding to the key hands and the chess game influence index of the key hands; and marks the key hands of the game segments based on the chess player thinking coefficient and the chess game influence index of the key hands, which can reduce the strong subjectivity in the Go review system, improve the review efficiency, and enhance the review feedback efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0049] Figure 1 This is a system block diagram of a Go intelligent review system based on multi-dimensional analysis according to an embodiment of the present invention;
[0050] Figure 2 This is a workflow diagram of the first Go intelligent review system based on multi-dimensional analysis in an embodiment of the present invention. DETAILED DESCRIPTION
[0051] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0052] In addition, the described features, structures or characteristics can be combined in any suitable manner in one or more example embodiments. In the following description, many specific details are provided to provide a full understanding of the example embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure can be practiced while omitting one or more of the specific details, or other methods, components, steps, etc. can be adopted. In other cases, well-known structures, methods, implementations or operations are not shown or described in detail to avoid obscuring various aspects of the present disclosure.
[0053] This embodiment provides a Go intelligent review system based on multi-dimensional analysis. Figure 1 This is a system block diagram of a Go intelligent review system based on multi-dimensional analysis according to an embodiment of the present invention. Figure 1 As shown, the system includes:
[0054] A chess game data acquisition module is used to obtain a video recording of the chess game played by a chess player and determine the chess pieces held by the chess player. Based on the video recording and the chess pieces held, the images of the chess pieces held by the chess player are extracted in the order in which the chess pieces are placed by the chess player.
[0055] A game segmentation processing module is used to determine the number of occupied points held by the chess pieces of the chess player and the number of occupied points held by the opponent's chess pieces in each chess piece placement image, and to segment the game period based on the number of occupied points held by the chess pieces of the chess player to obtain game segments;
[0056] The key move determination module is used to determine the key moves of each game segment, obtain the player's thinking coefficient corresponding to the key moves and the game influence index of the key moves;
[0057] The replay annotation module is used to annotate key moves in game segments based on the player's thinking coefficient and the key move's game impact index. This annotation process includes multi-dimensional annotation and visualization output.
[0058] In summary, the present invention obtains the video recording of the chess game to which a chess player belongs and determines the chess pieces held by the chess player, and extracts the chess piece placement images of the chess pieces held by the chess player in the order in which the chess player places the chess pieces based on the video recording and the chess pieces held; determines the number of occupied points of the chess pieces held by the chess player and the number of occupied points of the opponent's chess pieces in the chess piece placement image based on each placement image, and segments the game period based on the number of occupied points of the chess pieces held by the chess player to obtain game segments; determines the key hands of each game segment, obtains the chess player thinking coefficient corresponding to the key hands and the chess game influence index of the key hands; marks the key hands of the game segment based on the chess player thinking coefficient and the chess game influence index of the key hands, which can reduce the strong subjectivity in the Go review system, improve the review efficiency, and enhance the review feedback efficiency.
[0059] Furthermore, the present invention can improve the automation and efficiency of Go review.
[0060] Full process automation: From video extraction to key move annotation, no manual intervention is required, and the review efficiency is improved by more than 90%. Real-time analysis: supports real-time analysis during the game (such as live review), providing instant feedback to players.
[0061] Data-driven in-depth analysis: Through indicators such as the number of points, win rate fluctuations, and thinking time, chess game analysis is transformed from subjective experience into objective data.
[0062] Comparative Analysis: Supports comparison of different players or different games of the same player to reveal differences in player styles.
[0063] In some embodiments, Figure 2 This is a workflow diagram of the first Go intelligent review system based on multi-dimensional analysis according to an embodiment of the present invention. Figure 2 As shown, based on the video recording and the chess pieces held by the chess players, sequentially extracting the images of the chess pieces placed by the chess players in the order in which the chess pieces are placed includes the following steps:
[0064] Step S201: extracting the chess pieces held by the chess player in the order of placement based on the video recording and the chess pieces held by the chess player, wherein the chess pieces held by the chess player are frames containing other backgrounds;
[0065] Step S202: performing super-pixel segmentation on the selected child frame to obtain a super-pixel child image;
[0066] Step S203: extracting the image features of the super-pixel chess piece image, determining a detection bounding box of the chess piece held by the chess player based on the image features of the chess piece, and determining the chess piece image of the chess piece held by the chess player in the video recording based on the detection bounding box.
[0067] In some embodiments, performing super-pixel segmentation on the selected sub-frame to obtain a super-pixel sub-image specifically includes the following process:
[0068] Step 1: Convert the selected sub-frame to the CIELAB color space and combine the color vector and X and Y coordinates into a 5-dimensional feature vector. The CIELAB color space consists of three elements: brightness and color, A and B. A represents the color range from magenta to green, and B represents the color range from yellow to blue.
[0069] Step 2: Initialize the seed points and evenly distribute them in the image. The number of seed points is the same as the number of superpixels. The position of the seed point is the center of the superpixel. Assume that the selected subframe has N pixels, the number of superpixels is K, the size of each superpixel is N / K, and the distance between superpixels is
[0070] Step 3: Adjust the seed point position and select the position with the smallest gradient within the n×n range of the seed point as the new seed point position, where n is a preset value.
[0071] Step 4: Assign labels to pixels, i.e., the cluster centers corresponding to the pixels. Assign labels to each pixel in the 2S×2S region of each seed point, initialize the superpixel category array and distance array, record the seed point corresponding to the pixel and the distance from each pixel to the seed point, and calculate the distance from the pixel to the seed point in the 2S×2S region of each seed point. If it is less than the distance in the original distance array, update the distance array and superpixel category array.
[0072] Step 5: Iterative optimization, repeating steps 3 and 4 until the cluster center of each pixel no longer changes, and obtaining a superpixel image.
[0073] In some embodiments, determining the detection bounding box of the chess piece held by the player based on the features of the chess piece placement image specifically includes the following process:
[0074] The image feature F of the move and the learnable target query F q and the temporal features of the image F C Input into the target decoder composed of cross attention and feedforward network, and the final target feature F is obtained by calculating the cross attention H , which is expressed as follows:
[0075] Q E =F q +F C ;K E =F;V E =F;
[0076]
[0077] F H =Layer→norm(FFN(F H1 )+F); where dE Represents feature F H1 Dimension; FFN represents feedforward network, Layer→norm(·) represents layer normalization processing;
[0078] The final target feature F H The target feature detection frame is recorded as the detection bounding box of the chess piece held by the chess player.
[0079] In some embodiments, determining the number of occupied points of the chess pieces held by the chess player and the number of occupied points of the opponent's chess pieces in the chess piece placement images based on each chess piece placement image specifically includes the following process:
[0080] Use edge detection and Hough transform to detect chessboard lines and locate chessboard grids;
[0081] Live shape detection and dead shape judgment: Connected area analysis: Use DFS / BFS to traverse the connected area of the chess pieces and mark live and dead shapes; Live shape condition verification: Check whether the live shape contains at least two independent air holes, and the chess pieces are connected to form an indivisible whole; Dead shape judgment: Chess pieces that are completely surrounded by the enemy and cannot form two eyes are marked as dead;
[0082] Use the flood-fill algorithm to mark all empty points within the boundaries of the active shape. Count the number of marked empty points, which is the number of points of the active shape. Count the number of your own chess pieces in the active shape. Traverse all active shapes, accumulate the number of points and the number of pieces, and get the total number of occupied points for the player and the opponent.
[0083] In some embodiments, the game period is segmented based on the number of occupied points held by the chess pieces of the chess players, and obtaining the game segments specifically includes the following process:
[0084] The order of chess moves corresponding to the number of occupied points held by the chess pieces of the chess player is used as the horizontal axis, and the number of occupied points held by the chess pieces of the chess player corresponding to the order of chess moves is used as the vertical axis. A rectangular coordinate system is established, and a curve of the number of occupied points is generated by tracing points. The slopes of the curves of the number of occupied points between adjacent chess moves are counted, and the slopes are arranged according to the order of chess moves. The first two adjacent chess moves where the slope decreases are marked as the first-division chess moves, the chess moves corresponding to the chess moves in the first-division chess move with a later order are recorded as the first-division time point, the second two adjacent chess moves where the slope decreases for the second time are marked as the second-division chess moves, and the chess moves corresponding to the chess moves in the second-division chess move with a later order are recorded as the second-division time point. The period from the start time point of the chess game to the first-division time point is recorded as the layout game segment, the period from the first-division time point to the second-division time point is recorded as the mid-game battle game segment, and the period from the second-division time point to the end time point is recorded as the endgame game segment.
[0085] In some embodiments, the key moves of each game segment specifically include the following process:
[0086] Obtain a curve of the number of occupied points for each game segment, and record as key moves any move in which the number of occupied points exceeds a first preset threshold or is lower than a second preset threshold, where the first threshold is greater than the second threshold.
[0087] Furthermore, obtaining the player's thinking coefficient corresponding to the key move and the game influence index of the key move specifically includes the following process:
[0088] Obtain the player's thinking time corresponding to the key move, where the player's thinking time corresponding to the key move is the period between the opponent's move and the player's completion of the key move; calculate the difference between the player's thinking time corresponding to the key move and the preset thinking time, and record the ratio of the difference to the game period as the player's thinking coefficient;
[0089] Establish a hierarchical model with a step-by-step structure, where the layers include the target layer, the criterion layer, and the measure layer; construct a judgment matrix: determine the number m of relevant influencing factors at each layer, and construct a set U of chess game influencing factors, U = {φ1, φ2, φ3}, where φ1 is the subset of the change in the AI win rate before and after the move, φ2 is the subset of the number of points before and after the move, and φ3 is the subset of the move efficiency. From the set U, take two subsets from the same layer and compare them. Use M to represent the importance ratio, and assign the corresponding importance according to the preset ratio. Combine the importance of each layer to form a judgment matrix;
[0090] Calculate the maximum eigenvalue γ of the judgment matrix:
[0091] Among them, v αβ It is the matrix obtained by normalizing each column vector of the judgment matrix. The values of α and β are 1, 2...m, w α is the matrix v αβ The elements of are added row by row to obtain the vector and then normalized into a matrix, w β is the matrix v αβ Add the elements of column by column to obtain the vector and then normalize the matrix;
[0092] Calculate the game impact index CI of key moves: Where k represents the order of the judgment matrix.
[0093] Furthermore, the key move marking process of dividing the game into segments based on the player's thinking coefficient and the game influence index of the key move specifically includes the following steps:
[0094] Calculate the product of the player's thinking coefficient and the game influence index of the key hand after normalization, load the product value threshold, where the product value threshold is set by the system and stored in the system, and judge whether the product value exceeds the product value threshold. If so, mark the key hands of the game segment; if not, do not mark the key hands of the game segment.
[0095] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.
[0096] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0097] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0098] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is only for some logical functions. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0099] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.
[0100] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.
Claims
1. A Go intelligent review system based on multi-dimensional analysis, characterized by: The system includes: A chess game data acquisition module is used to obtain a video recording of the chess game played by a chess player and determine the chess pieces held by the chess player. Based on the video recording and the chess pieces held, the images of the chess pieces held by the chess player are extracted in the order in which the chess pieces are placed by the chess player. A game segmentation processing module is used to determine the number of occupied points held by the chess pieces of the chess player and the number of occupied points held by the opponent's chess pieces in each chess piece placement image, and to segment the game period based on the number of occupied points held by the chess pieces of the chess player to obtain game segments; The key move determination module is used to determine the key moves of each game segment, obtain the player's thinking coefficient corresponding to the key moves and the game influence index of the key moves; The replay marking module is used to mark the key moves of the game segments based on the player's thinking coefficient and the game impact index of the key moves.
2. The Go intelligent review system based on multi-dimensional analysis according to claim 1, characterized in that: The specific process of extracting the chess pieces placed by the chess players in sequence based on the video recording and the chess pieces they hold includes the following steps: Extracting the chess pieces held by the chess player in the order of chess player placement based on the video recording and the chess pieces held, wherein the chess pieces held by the chess player are frames containing other backgrounds; Perform super-pixel segmentation on the selected drop frame to obtain a super-pixel drop image; The features of the chess piece dropped are extracted from the super-pixel chess piece dropped image, a detection bounding box of the chess piece held by the chess player is determined based on the features of the chess piece dropped image, and the dropped image of the chess piece held by the chess player in the video recording is determined based on the detection bounding box.
3. The Go intelligent review system based on multi-dimensional analysis according to claim 2, characterized in that: The process of performing super-pixel segmentation on the selected sub-frame to obtain the super-pixel sub-image is as follows: Step 1: Convert the selected sub-frame to the CIELAB color space and combine the color vector and X and Y coordinates into a 5-dimensional feature vector. The CIELAB color space consists of three elements: brightness and color, A and B. A represents the color range from magenta to green, and B represents the color range from yellow to blue. Step 2: Initialize the seed points and evenly distribute them in the image. The number of seed points is the same as the number of superpixels. The position of the seed point is the center of the superpixel. Assume that the selected subframe has N pixels, the number of superpixels is K, the size of each superpixel is N / K, and the distance between superpixels is Step 3: Adjust the seed point position and select the position with the smallest gradient within the n×n range of the seed point as the new seed point position, where n is a preset value. Step 4: Assign labels to pixels, i.e., the cluster centers corresponding to the pixels. Assign labels to each pixel in the 2S×2S region of each seed point, initialize the superpixel category array and distance array, record the seed point corresponding to the pixel and the distance from each pixel to the seed point, and calculate the distance from the pixel to the seed point in the 2S×2S region of each seed point. If it is less than the distance in the original distance array, update the distance array and superpixel category array. Step 5: Iterative optimization, repeating steps 3 and 4 until the cluster center of each pixel no longer changes, and obtaining a superpixel image.
4. The Go intelligent review system based on multi-dimensional analysis according to claim 2, characterized in that: Determining the detection bounding box of the chess piece held by the player based on the features of the chess piece image specifically includes the following process: The image feature F of the move and the learnable target query F q and the temporal features of the image F C Input into the target decoder composed of cross attention and feedforward network, and the final target feature F is obtained by calculating the cross attention H , which is expressed as follows: Q E =F q +F C ;K E =F;V E =F; F H =Layer→norm(FFN(F H1 )+F); where d E Represents feature F H1 Dimension; FFN represents feedforward network, Layer→norm(·) represents layer normalization processing; The final target feature F H The target feature detection frame is recorded as the detection bounding box of the chess piece held by the chess player.
5. The Go intelligent review system based on multi-dimensional analysis according to claim 1, characterized in that: Determining the number of occupied points held by the chess pieces of the chess player and the number of occupied points held by the opponent's chess pieces in each chess piece placement image based on each chess piece placement image specifically includes the following process: Use edge detection and Hough transform to detect chessboard lines and locate chessboard grids; Live shape detection and dead shape judgment: Connected area analysis: Use DFS / BFS to traverse the connected area of the chess pieces and mark live and dead shapes; Live shape condition verification: Check whether the live shape contains at least two independent air holes, and the chess pieces are connected to form an indivisible whole; Dead shape judgment: Chess pieces that are completely surrounded by the enemy and cannot form two eyes are marked as dead; Use the flood-fill algorithm to mark all empty points within the boundaries of the active shape. Count the number of marked empty points, which is the number of points of the active shape. Count the number of your own chess pieces in the active shape. Traverse all active shapes, accumulate the number of points and the number of pieces, and get the total number of occupied points for the player and the opponent.
6. The Go intelligent review system based on multi-dimensional analysis according to claim 1, characterized in that: The game period is segmented based on the number of occupied points held by the chess pieces of the players. The specific process of obtaining the game segmentation includes the following: The order of chess moves corresponding to the number of occupied points held by the chess pieces of the chess player is used as the horizontal axis, and the number of occupied points held by the chess pieces of the chess player corresponding to the order of chess moves is used as the vertical axis. A rectangular coordinate system is established, and a curve of the number of occupied points is generated by tracing points. The slopes of the curves of the number of occupied points between adjacent chess moves are counted, and the slopes are arranged according to the order of chess moves. The first two adjacent chess moves where the slope decreases are marked as the first-division chess moves, the chess moves corresponding to the chess moves in the first-division chess move with a later order are recorded as the first-division time point, the second two adjacent chess moves where the slope decreases for the second time are marked as the second-division chess moves, and the chess moves corresponding to the chess moves in the second-division chess move with a later order are recorded as the second-division time point. The period from the start time point of the chess game to the first-division time point is recorded as the layout game segment, the period from the first-division time point to the second-division time point is recorded as the mid-game battle game segment, and the period from the second-division time point to the end time point is recorded as the endgame game segment.
7. The Go intelligent review system based on multi-dimensional analysis according to claim 6, characterized in that: The key moves in each game segment specifically include the following processes: Obtain a curve of the number of occupied points for each game segment, and record as key moves any move in which the number of occupied points exceeds a first preset threshold or is lower than a second preset threshold, where the first threshold is greater than the second threshold.
8. The Go intelligent review system based on multi-dimensional analysis according to claim 1, characterized in that: The specific process of obtaining the player's thinking coefficient corresponding to the key move and the key move's game influence index includes the following steps: Obtain the player's thinking time corresponding to the key move, where the player's thinking time corresponding to the key move is the period between the opponent's move and the player's completion of the key move; calculate the difference between the player's thinking time corresponding to the key move and the preset thinking time, and record the ratio of the difference to the game period as the player's thinking coefficient; Establish a hierarchical model with a step-by-step structure, where the layers include the target layer, the criterion layer, and the measure layer; construct a judgment matrix: determine the number m of relevant influencing factors at each layer, and construct a set U of chess game influencing factors, U = {φ1, φ2, φ3}, where φ1 is the subset of the change in the AI win rate before and after the move, φ2 is the subset of the number of points before and after the move, and φ3 is the subset of the move efficiency. From the set U, take two subsets from the same layer and compare them. Use M to represent the importance ratio, and assign the corresponding importance according to the preset ratio. Combine the importance of each layer to form a judgment matrix; Calculate the maximum eigenvalue γ of the judgment matrix: Among them, v αβ It is the matrix obtained by normalizing each column vector of the judgment matrix. The values of α and β are 1, 2...m, w α is the matrix v αβ The elements of are added row by row to obtain the vector and then normalized into a matrix, w β is the matrix v αβ Add the elements of column by column to obtain the vector and then normalize the matrix; Calculate the game impact index CI of key moves: Where k represents the order of the judgment matrix.
9. The Go intelligent review system based on multi-dimensional analysis according to claim 1, characterized in that: Based on the player's thinking coefficient and the key hand's chess game influence index, the key hand marking processing of the game segmentation is specific The following processes are included: Calculate the product of the player's thinking coefficient and the game influence index of the key hand after normalization, load the product value threshold, where the product value threshold is set by the system and stored in the system, and judge whether the product value exceeds the product value threshold. If so, mark the key hands of the game segment; if not, do not mark the key hands of the game segment.