Visual analysis system for inspection of football player

Abstract

The invention discloses a visual analysis system for investigation of football players. The visual analysis system comprises a navigation view and an investigation view, the navigation view supports the user to observe the role of a new player through the lineup board, identifies the player possibly replaced by the new player, displays a series of automatically recommended players through a player ranking list after specifying an inspection index, and also supports the comparison of the automatically recommended players according to a plurality of standards. The selected potential target players are supported to be added into the survey view; the survey view supports the user to find the interested tactics through the tactical list, supports the comparison of the expected action scores of the players through the player record list to determine the most appropriate player, and also supports the further survey of the difference between the simulated action and the historical action to know the influence of the players on the team tactical system. According to the system, a systematic solution is provided for a player to inspect problems, and the football player inspection process based on simulation is supported through player navigation, comparison and deep investigation.

Description

Technical Field

[0001] The invention belongs to the technical field of combining data analysis and visual analysis, and in particular relates to a visual analysis system for observing football players. Background Art

[0002] Player scouting refers to finding potential players from outside the team to provide the coach with more player options. Based on the scouting results, the staff will formulate strategies for player recruitment and transfer.

[0003] Scoring the right players for a team is a complex task. Coaches and analysts need to compare a large number of players on various performance metrics to assess whether they are suitable for the team. Many studies have proposed methods to evaluate player performance from different aspects. Traditional methods for selecting players in soccer mainly quantify player performance through multiple aspects and focus on the players with higher rankings. For example, player ranking methods based on the weighted sum of player performance characteristics are used to find the players who contribute the most to the team. Other methods focus on the contribution of players through each fine-grained action and rank players according to different types of overall action values. In addition, current works have proposed various criteria for evaluating player performance, such as defending the opposing players and passing creativity. However, the actual player scouting process is more complicated than simply selecting the best performing player, and is usually subject to constraints such as player roles and team transfer budgets. Therefore, many works summarize soccer player scouting as a multi-criteria decision making (MCDM) problem and select the most suitable players for the team by optimizing the sum of player performance metrics under multiple constraints.

[0004] Although these studies are helpful, it is difficult to know how a player will perform in a new team directly from historical game data, which is one of the most important reasons for the failure of player scouting. Therefore, simulation analysis is introduced into the process of football player scouting to estimate the performance of players in new teams. For example, football games are modeled as a semi-Markov decision process to simulate the goal difference after new players join the team and select players who can bring higher goal difference. There are also some methods that define football player scouting as a stochastic programming problem, simulating the total market value of the team after adding new players to discover players with economic potential.

[0005] However, these methods provide results of expected increments in team performance indicators through complex mathematical models. Due to the lack of interactive navigation tools, experts often have difficulties in fully exploring a large space of potential target players. In addition, without understanding the calculation process of the model, experts also face challenges in understanding how players cooperate with team tactics to realize their potential. Visual analysis technology supports experts to interact with data analysis models and interpret model results. Through visual analysis technology, coaches can integrate domain knowledge into the analysis process to adjust model inputs and explore detailed model results to improve understanding of the results. However, in the field of visual analysis of football match data, most interactive visualization systems focus on player position and movement analysis, formation change analysis, passing pattern mining and analysis, and player cross-border transfer analysis. Given the unique data structure and domain tasks, it is difficult to directly apply the above-mentioned visual analysis research on football match data to player scouting tasks.

[0006] Player performance evaluation is one of the most widely used methods in football player selection and has also attracted the attention of recent visualization research. For example, metaphor- and icon-based visualization methods are provided to compare statistical performance indicators of multiple players. Some other works have developed visual analysis systems to evaluate players by performance characteristics in different game contexts, or to visualize their performance based on their detailed actions in the game. However, it is still difficult to directly use these studies for football player assessment because they do not support the navigation of target players and the comparison of player behaviors under different tactics. Therefore, it is necessary to provide a visualization analysis system combined with a game simulation framework to support navigation and comparison in football player assessment and evaluate the results of player assessment. Summary of the invention

[0007] In view of the above, an object of the present invention is to provide a visual analysis system for football player scouting, to provide a systematic solution to the player scouting problem, and to support the simulation-based football player scouting process through player navigation, comparison and in-depth investigation.

[0008] To achieve the above-mentioned purpose of the invention, an embodiment provides a visual analysis system for football player inspection, including a navigation view and an investigation view;

[0009] The navigation view supports the user to observe roles that need to find new players through the lineup board and identify players that may be replaced by new players. It also supports displaying a series of automatically recommended players through a player ranking list after specifying the investigation indicators. It also supports comparing the automatically recommended players according to multiple criteria and supports adding selected potential target players to the investigation view during the comparison process.

[0010] The investigation view supports the user to find a tactic of interest through a tactic list, and supports comparing the expected action scores of players through a player record list to determine the most suitable player, and also supports further investigation of the differences between simulated actions and historical actions to understand the impact of players on the team's tactical system.

[0011] Preferably, the lineup board included in the navigation view is used to indicate the players to be replaced as a reference for player recommendation, and all the players in the team are displayed according to their roles. The background of the lineup board is half a football field, and it is divided into ten areas according to the spatial positions of players with different roles on the field. Players with the same role are placed in corresponding areas of the lineup board, and players with the same role are arranged in descending order according to their playing time to show their importance, and their market value is encoded using a bar chart for effective comparison.

[0012] Preferably, the player ranking list included in the navigation view is used to compare recommended players on multiple criteria to find potential target players. Specifically, through the player recommendation function, suitable players are screened out from a large number of candidate players, and the recommended players are displayed in the form of a sortable list to help users compare players. Each row in the list represents a player and all of its indicators, including the player's name tag, basic personal information, historical performance, and expected performance. All indicators are encoded in bar graphs to facilitate comparison under multiple criteria. At the same time, accurate numerical text of basic player information is also provided.

[0013] Preferably, the navigation view includes the following interactive operations:

[0014] Select a replacement player: The user clicks on a player in the lineup board to select him as the player to be replaced;

[0015] Select evaluation index: Users can select evaluation index according to their needs through the drop-down list at the top of the lineup board. After selection, the recommended players will be displayed in the player ranking list;

[0016] Sorting and filtering: Users click the sort button at the top of the player ranking list, or drag the slider to sort or filter players according to different player comparison criteria;

[0017] Select potential target players: The user clicks the checkbox to the right of a player item in the player ranking list to add the selected player as a potential target player to the investigation view.

[0018] Preferably, the investigation view includes a tactic list for selecting a tactic of interest and analyzing a sortable list of all tactics of the team, wherein the tactic list supports users in selecting a tactic of interest to analyze the simulated actions of the players therein, each row in the tactic list represents a tactic, including a tactical map showing the tactics and their frequency and success rate, and the tactics are filtered by categories, including offensive tactics and defensive tactics.

[0019] Preferably, the tactics are visualized based on the court through a tactical map, including an abstract spatiotemporal trajectory and a heat map. The tactics are defined as representative frequent sequence patterns under similar ball possession sequences. Specifically, each action in the ball possession sequence is discretized by the court area, and the frequent sequence patterns are obtained by mining the ball possession sequence patterns after the action discretization. Therefore, the tactics are represented as a series of action sequences discretized by the court area. These court areas are connected to form an abstract spatiotemporal trajectory to display the tactics. The heat map is used to encode the overall action value in each area in the tactics to help users understand the spatial effect of the tactics.

[0020] Preferably, the player record list included in the survey view is used to observe the expected contribution of the player at the action level under the selected tactics, and specifically records the potential target players to analyze their expected contribution at the action level under specific tactics. Each row in the player record list contains the player label, the average difference between the player's simulated action and the original action, and the average action value increment. All indicators in each row are encoded by bar charts to facilitate comparison. The bar charts of the average difference and action value increment can also be expanded into area charts to show their distribution.

[0021] Preferably, the simulated action map included in the survey view is used to display the detailed spatial distribution and expected contribution of the player's simulated actions. The simulated actions ending in the same grid will be aggregated and visualized through an icon, which shows the comparison results of the simulated actions with the original actions, including the frequency of the simulated actions, the average action value increment, and the average difference compared with the original actions. The average difference is encoded by the saturation of the icon background, and the lighter the color, the smaller the difference. The inner rectangle of the icon represents the spatial distribution of the simulated actions and their effects. The area of ​​the rectangle represents the frequency, and the saturation represents the average value increment.

[0022] Preferably, the survey view includes the following interactive operations:

[0023] Sorting: The user clicks the sort button at the top of the tactics list and player record list to sort the items in the list;

[0024] Filter: Users filter tactics by category via the drop-down list, filter by frequency via the slider at the top of the list, and in the simulated action map, users filter simulated actions by difference from the original action by dragging the slider;

[0025] Select tactics: The user clicks on an item in the tactics list to select a tactic, and the actions in the selected tactic will be displayed in the player record list and the simulated action diagram;

[0026] Select Player: The user clicks on an item in the player record list to view that player's actions in the simulated action map.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] In the visual analysis system for football player assessment provided by the present invention, football coaches and data analysts only need to import historical game data in the corresponding format to easily, intuitively, comprehensively and deeply select football players and make further comparisons and interpretations. Specifically, the present invention avoids unsuitable players by estimating the expected performance of players in new teams, and supports interactive player navigation, comparison and investigation. With the help of the system of the present invention, coaches and analysts can assess players who can improve team performance and understand how they can integrate into the tactical system of the new team in future games. Experts in the field have given high praise to the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0030] Figure 1 1 is a schematic diagram of the structure of a visual analysis system for football player assessment provided by an embodiment. DETAILED DESCRIPTION

[0031] To make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation methods described herein are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0032] The present invention mainly solves two problems existing in the existing football player assessment methods:

[0033] The first problem is to provide a fine-grained simulation of players' behavior in a new team. When scouting players, experts need to understand how players behave when playing in a new team to assess whether they fit the team's tactics and understand how to effectively deploy them in future games. Most existing studies simulate summary statistical indicators of players after joining the team, which is not enough for experts to fully understand the players in terms of possible behaviors. Players' behavior in a game is associated with multiple heterogeneous features, such as players' decision-making style and current team tactics. It is very difficult to extract and integrate these complex features into the simulation of player behavior.

[0034] The second issue is to develop effective visualization tools to support player navigation and comparison. In order to identify target players and conduct further research, experts need to explore the entire player space from all aspects of player performance. However, the target players may come from any other club or even league, which often makes experts feel overwhelmed in the navigation process. In addition, it is not easy to compare potential target players in multiple spatiotemporal dimensions (such as simulated behavior and its results under different team tactics) to determine who is the best fit for the team. The system should support visual comparison of target players in these complex aspects.

[0035] To solve the above two technical problems, the embodiment of the present invention provides a visual analysis system for football player assessment, which supports systematic football player assessment through simulation analysis. Users can explore the possible space of potential target players and compare their expected contributions to the game results, and can also compare the simulated behaviors of the players to determine the most suitable players.

[0036] like Figure 1 As shown, the visual analysis system for football player investigation provided by the embodiment includes two views: navigation view A and investigation view B. In navigation view A, users can observe the roles that need to find new players through the lineup board A1, and identify players that may be replaced by new players. After specifying the investigation index, the player ranking list A2 will show a series of automatically recommended players. Users can compare these players according to multiple criteria (such as basic personal information, historical performance and expected game results). During the comparison process, users can add potential target players to the investigation view. In investigation view B, users can find interesting tactics through tactics list B1, and compare the expected action scores of players through player record list B2 to determine the most suitable players. Users can also further investigate the differences between simulated actions and historical actions to understand the impact of players on the team's tactical system.

[0037] In the embodiment, the navigation view A includes a lineup board A1 for specifying players to be replaced as references for player recommendations, and a player ranking list A2 for comparing recommended players on multiple criteria to find potential target players.

[0038] In the embodiment, the lineup board A1 displays all the players in the team according to their roles. The background of the lineup board is half a football field, and is divided into ten areas (i.e., goalkeeper, left back, right back, center back, defensive midfielder, midfielder, offensive midfielder, left winger, right winger and center forward) according to the spatial positions of players with different roles on the field. Players with the same role are placed in the corresponding areas of the lineup board in the view, and players with the same role are arranged in descending order according to their playing time to show their importance, and their market value is encoded using a bar chart for effective comparison.

[0039] In the embodiment, the player ranking list A2 uses the player recommendation function to screen out suitable players from a large number of candidate players, and displays the recommended players in the form of a sortable list to help users compare players. Each row A5 in the list represents a player, including the player's name tag, basic personal information A3, and historical and expected performance A4. Basic personal information includes the player's age, market value, current team and its similarity to the analyzed team. Historical performance includes the player's overall action value and playing time per 90 minutes, while expected performance is the expected winning rate increment calculated by the game result prediction model. All indicators are encoded using bar charts to facilitate comparison under multiple criteria. The view also provides precise numerical text of basic player information.

[0040] In an embodiment, the navigation view includes the following interactive operations:

[0041] Select substitution player: Users can click on a player in the lineup board to select him as the player to be substituted.

[0042] Select Scrutiny: Users can select scout metric based on their needs through the drop-down list at the top of the lineup board. These metrics include similarity, historical performance, and expected performance. Once selected, the recommended players will be displayed in the player ranking list.

[0043] Sorting and Filtering: Users can click the sort button at the top of the player ranking list, or drag the slider to sort or filter players based on different player comparison criteria.

[0044] Select potential target players: Users can click the checkbox to the right of a player item in the player ranking list to add the player to the investigation view as a potential target player.

[0045] In the embodiment, the investigation view B includes a tactic list B1 for selecting a tactic of interest, and a player record list B2 for observing the expected contribution of the player at the action level under the selected tactic. The detailed spatial distribution and expected contribution of the player's simulated action are displayed in the simulated action graph B3. The direction of the court in this view is attacking from left to right.

[0046] In the embodiment, the tactic list B1 is a sortable list of all tactics of the analysis team. Through this list, the user can select the tactic of interest to analyze the simulated actions of the players in it. Each row B4 in the list represents a tactic, including a tactical map showing the tactic and its frequency and success rate. Tactics can be filtered by category, because coaches and analysts often need to identify specific categories of tactics for analysis based on their tactical preferences. Categories include offensive tactics, such as simple passes, corner kicks, and free kicks, and defensive tactics, such as tackles and interceptions.

[0047] In an embodiment, a tactical map is a visualization of tactics based on the court, including an abstract spatiotemporal trajectory and a heat map. Tactics are defined as representative frequent sequence patterns under similar ball possession sequences. Specifically, each action in a ball possession sequence can be discretized by court regions, and frequent sequence patterns are obtained by mining the ball possession sequence patterns after the discretization of the actions. Therefore, tactics can be represented as a series of action sequences discretized by court regions. These court regions can be connected to form an abstract spatiotemporal trajectory to display tactics. A heat map is used in the view to encode the overall action value in each region in the tactics to help users understand the spatial effect of the tactics.

[0048] In the embodiment, the player record list B2 records potential target players to analyze their expected contribution at the action level under specific tactics. Each row B6 in the list contains a player tag, the average difference between the player's simulated action and the original action, and the average action value increment. These two indicators indicate the compatibility of the player with the tactics and the performance under these tactics. The player tag also includes the player's age and market value to help users decide on target players. All indicators are encoded through bar graphs for easy comparison. The bar graphs of average difference and action value increment can also be expanded into area graphs B5 to show their distribution.

[0049] In the embodiment, the simulated action diagram B3 shows the spatial distribution of simulated actions under specific tactics. The court is divided into 96 rectangular grids in the view, each grid is 12 columns and 8 rows, and the width of each grid is 8.75 meters and the height is 8.5 meters. The simulated actions ending in the same grid will be aggregated and visualized through an icon. The icon shows the comparison results of the simulated actions with the original actions, including the frequency of the simulated actions, the average action value increment, and the average difference compared with the original actions. The average difference is encoded by the saturation of the icon background, and the lighter the color, the smaller the difference. The inner rectangle of the icon represents the spatial distribution of the simulated actions and their effects, the area of ​​the rectangle represents the frequency, and the saturation represents the average value increment.

[0050] In an embodiment, the survey view includes the following interactive operations:

[0051] Sorting: Users can click the sort button at the top of the tactics list and player record list to sort the items in the list.

[0052] Filter: Users can filter plays by category via the drop-down list, and by frequency via the slider at the top of the list. In the Simulated Plays Map, users can filter simulated plays by difference from the original play by dragging the slider.

[0053] Select a tactic: Users can click on an item in the tactic list to select a tactic, and the actions in that tactic will be displayed in the player record list and the simulated action diagram.

[0054] Select Player: Users can click on an item in the player record list to view that player's actions in the simulated action map.

[0055] Based on the visual analysis system for football player assessment, the visual analysis process of football player assessment by analysts includes:

[0056] The analyst first clicks on a player on the lineup board and selects expected performance as the criteria to find players who can improve the team's performance. Next, the analyst looks at the top 10 players recommended by the system and finds that most of them meet expectations. The analyst notices a player C who ranks highest in terms of win rate improvement. At the same time, the player's current team is similar to the target team and has a relatively good historical performance. Based on these findings, the analyst concludes that this player may be the best player to improve the team's performance and adds him to the investigation view for further analysis. In addition, the analyst finds a series of possible target players and adds these players to the investigation view.

[0057] After selecting the potential target player, the analyst moves to the investigation view to observe the detailed simulated movements of the players by tactic. The analyst selects the simple pass tactic and filters out the tactics with a frequency below 0.05 and sorts them by success rate to focus on the team's successful tactics. The analyst notices that the left-wing build-up tactic is the most successful tactic for the team. Therefore, the analyst selects this tactic and moves to the player record list to compare the players' simulated movements.

[0058] In the player record list, the analyst sorted the list by the average value of the action value increment and found that player C ranked the highest with a value greater than zero, which means that player C can perform more valuable actions under the left-wing organization tactics compared to other potential target players. The analyst also sorted the list by action decision difference and found that player C's simulated actions differed the most from the original players. This indicates that player C will improve the left-wing organization tactics through different action decisions. The analyst further selected these simulated actions of player C and turned to the simulated action graph to examine the details. In the simulated action graph, the analyst found that the high-value simulated actions mainly ended in the midfield. Through this process, the analyst concluded that player C can improve the target team's winning rate by improving the action selection under the left-wing organization tactics.

[0059] The specific implementation methods described above have described in detail the technical solutions and beneficial effects of the present invention. It should be understood that the above is only the most preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, supplements and equivalent substitutions made within the scope of the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A visual analysis system for football player assessment, characterized in that: Includes navigation view and investigation view; The navigation view supports the user to observe roles that need to find new players through the lineup board and identify players that may be replaced by new players. It also supports displaying a series of automatically recommended players through a player ranking list after specifying the investigation indicators. It also supports comparing the automatically recommended players according to multiple criteria and supports adding selected potential target players to the investigation view during the comparison process. The investigation view supports the user to find a tactic of interest through a tactic list, and supports comparing the expected action scores of players through a player record list to determine the most suitable player, and also supports further investigation of the differences between simulated actions and historical actions to understand the impact of players on the team's tactical system.

2. The visual analysis system for football player assessment according to claim 1, characterized in that: The lineup board included in the navigation view is used to indicate the players to be replaced as a reference for player recommendation, and all the players in the team are displayed according to their roles. The background of the lineup board is half a football field, and it is divided into ten areas according to the spatial positions of players with different roles on the field. Players with the same role are placed in the corresponding areas of the lineup board, and players with the same role are arranged in descending order according to their playing time to show their importance, and their market values ​​are encoded using a bar chart for effective comparison.

3. The visual analysis system for football player assessment according to claim 1, characterized in that: The player ranking list included in the navigation view is used to compare recommended players based on multiple criteria to find potential target players. Specifically, through the player recommendation function, suitable players are screened out from a large number of candidate players, and the recommended players are displayed in the form of a sortable list to help users compare players. Each row in the list represents a player and all of its indicators, including the player's name tag, basic personal information, historical performance, and expected performance. All indicators are encoded in bar charts to facilitate comparison under multiple criteria. At the same time, precise numerical text of basic player information is also provided.

4. The visual analysis system for football player assessment according to claim 1, characterized in that: The navigation view includes the following interactive operations: Select a replacement player: The user clicks on a player in the lineup board to select him as the player to be replaced; Select evaluation index: Users can select evaluation index according to their needs through the drop-down list at the top of the lineup board. After selection, the recommended players will be displayed in the player ranking list; Sorting and filtering: Users click the sort button at the top of the player ranking list, or drag the slider to sort or filter players according to different player comparison criteria; Select potential target players: The user clicks the checkbox to the right of a player item in the player ranking list to add the selected player as a potential target player to the investigation view.

5. The visual analysis system for football player assessment according to claim 1, characterized in that: The investigation view includes a tactic list for selecting a tactic of interest and analyzing a sortable list of all team tactics. The tactic list supports users in selecting a tactic of interest to analyze the simulated actions of the players therein. Each row in the tactic list represents a tactic, including a tactical graph showing the tactics and their frequency and success rate. Tactics are filtered by category, including offensive tactics and defensive tactics.

6. The visual analysis system for football player assessment according to claim 5, characterized in that: The tactics are visualized based on the court through a tactical graph, including an abstract spatiotemporal trajectory and a heat map. Tactics are defined as representative frequent sequence patterns under similar ball possession sequences. Specifically, each action in the ball possession sequence is discretized by the court area, and the frequent sequence patterns are obtained by mining the ball possession sequence patterns after the action discretization. Therefore, the tactics are represented as a series of action sequences discretized by the court area. These court areas are connected to form an abstract spatiotemporal trajectory to display the tactics. The heat map is used to encode the overall action value in each area in the tactics to help users understand the spatial effect of the tactics.

7. The visual analysis system for football player assessment according to claim 1, characterized in that: The investigation view includes a player record list for observing the expected contribution of the player at the action level under the selected tactics. It specifically records potential target players to analyze their expected contribution at the action level under specific tactics. Each row in the player record list includes a player label, the average difference between the player's simulated action and the original action, and the average action value increment. All indicators in each row are encoded through bar charts for easy comparison. The bar charts of the average difference and action value increment can also be expanded into area charts to show their distribution.

8. The visual analysis system for football player assessment according to claim 1, characterized in that: The simulated action map included in the survey view is used to display the detailed spatial distribution and expected contribution of the player's simulated actions. The simulated actions that end at the same grid will be aggregated and visualized through an icon, which shows the comparison results of the simulated actions with the original actions, including the frequency of the simulated actions, the average action value increment, and the average difference compared with the original actions. The average difference is encoded by the saturation of the icon background. The lighter the color, the smaller the difference. The inner rectangle of the icon represents the spatial distribution of the simulated actions and their effects. The area of ​​the rectangle represents the frequency, and the saturation represents the average value increment.

9. The visual analysis system for football player assessment according to claim 1, characterized in that: The survey view includes the following interactive operations: Sorting: The user clicks the sort button at the top of the tactics list and player record list to sort the items in the list; Filter: Users filter tactics by category via the drop-down list, filter by frequency via the slider at the top of the list, and in the simulated action map, users filter simulated actions by difference from the original action by dragging the slider; Select tactics: The user clicks on an item in the tactics list to select a tactic, and the actions in the selected tactic will be displayed in the player record list and the simulated action diagram; Select Player: The user clicks on an item in the player record list to view that player's actions in the simulated action map.

Images