Monitoring a return-stroke sports game

Abstract

The invention relates to monitoring a return-stroke sports game, comprising: detecting an identity of at least one of the players (7a, 7b) involved in the return-stroke sports game by means of a detection device (2); recording a video signal of the return-stroke sports game by means of a recording device (3); analysing the recorded video signal by means of a computer-implemented analysis algorithm, comprising extracting at least one relevant value of at least one player performance parameter from the recorded video signal and comparing the relevant extracted value of the player performance parameter with a player performance parameter history, and determining, on the basis of a result of the comparison, whether or not the at least one extracted value is within a tolerance range specified as permissible; and outputting a warning signal if the at least one extracted value is not within the tolerance range, in order to promote fairness in the return-stroke sports game.

Description

[0001] title

[0002] Monitoring a racket sports match

[0003] The scope of the invention relates to methods for monitoring a racket sport game, comprising recording the identity of the players involved in the racket sport game by a recording device; recording a video signal of the racket sport game by a recording device; and analyzing the recorded video signal by a computer-implemented analysis algorithm.

[0004] background

[0005] Sporting activities are an integral part of physical fitness in society. This is especially true for racket sports such as tennis, paddle tennis, pickleball, squash, volleyball, and the like. Many athletes, however, want not only to participate in sports but also to improve their skills in their chosen sport in order to be competitive in events, to participate in events where they collect performance points, and / or to qualify for participation in events such as tournaments through the accumulation of performance points or similar means.

[0006] Game activity monitoring systems, for example for ball sports such as football, are known and include, for example, sensors embedded in the ball that detect its movements. These sensors can, for example, use a magnetic field to determine the trajectory and / or position of the ball, as described in EP 2 657 924 A1.

[0007] Other monitoring and / or information systems used in ball sports, particularly racket sports, are known as smart court systems. However, tracking the movement of the often fast-moving objects on the court is very difficult and time-consuming. Therefore, in many cases, complex systems are used that enable tracking of the object on the court and in 3D space. This allows, among other things, real-time decisions to be made during a game, such as line calls in tennis, or the collection of statistics on the course of a sporting event. These complex systems often include multiple and / or movable cameras that enable the tracking of an object, such as the ball.

[0008] The Smartcourt system of WO 2013 / 124856 Al, for example, comprises a large number of video cameras mounted in a fixed position. Using image data from the various camera perspectives, a 3D model of the court, the ball, and the players is created. Based on the movements captured within this model, conclusions are drawn about the type of stroke, the ball speed, and the points of impact, and these can be viewed by the athletes.

[0009] In the field of computer-based analysis of video signals recorded during a game, DE 10 2018 119 170 A1 should also be mentioned. This patent corrects scores calculated by players during the game using event-based scoring (backward scoring), thus counteracting the unconscious or conscious reporting of false scores. This approach therefore also promotes fair play, i.e., it monitors the game to ensure consistent score recording for all participants.

[0010] However, event-based result determination in itself does not prevent the circumvention of the so-called "best effort" principle, according to which participating players must always give their best in the interest of fair play towards third parties, and no player should give another player an unfair advantage through deliberate underperformance. Such deliberate underperformance, for example in the form of allowing another player to win, can distort competition and therefore be unfair or even fraudulent in the context of a tournament or other ranking-relevant game.

[0011] The technical challenge, therefore, is to improve monitoring in a racket sport, in particular to automatically recognize at least potentially unfair play as such.

[0012] overview

[0013] This problem is solved by the subject matter of the independent claims. Advantageous embodiments result from the dependent claims, the description, and the figures.

[0014] One aspect concerns a (at least partially, i.e., semi- or fully automated) method for monitoring a racket sport, in particular a tennis game, a paddle game, a pickleball game, a squash game, a volleyball game, and / or a badminton game. Preferably, the game involves at least two players, and preferably no more than four or six players. The racket sport is preferably a ball-based racket sport, especially one using a racket or a hand as the playing implement. Using a racket as the playing implement has the advantage that game-relevant events, including the overall course of the game, can be recorded particularly robustly and automatically, for example, based on the contact times of the racket with the ball.

[0015] One process step is the identification of at least one of the players involved in the racket sport, i.e., participating players, using a data capture device. Preferably, the identities of all participating players are captured. This can be done, for example, by scanning a (QR) code using a camera unit of the data capture device and / or by facial recognition using the camera unit. Once the identities are captured, the performance parameters subsequently extracted can be assigned to the identities as player performance parameters and advantageously processed as described below.

[0016] As a further procedural step, the recording device may compare the captured identities with stored player information (in particular, releasing the players for the racket sport if the comparison is successful). The stored player information may, for example, include (in whole or in part) the player performance parameter history described below. For this purpose, the recording device may have a database or access a database or other storage structure, which may be located on a server and / or in the cloud, and in which the player information is stored. Alternatively or additionally, the player information may be stored on the player's mobile device. Release may be granted for one or more of the subsequent procedural steps.The comparison allows access to previously known individual values ​​of player performance parameters stored in the aforementioned database or another storage medium.

[0017] For example, the recording, analysis, and / or output of the warning signal described below can only occur if the comparison result is positive, meaning, for instance, that player information is stored for the recorded player and / or that the players are legally permitted to play against each other according to a predefined set of rules. For example, the predefined set of rules could include a set of permissible combinations of individual player performance reference values. As explained further below, this ensures, for example, that the players whose behavior in the game is to be monitored have a combination of age, gender, and / or performance level that is permissible according to the game rules defined in the predefined set of rules for the respective game. This is particularly advantageous for games that count towards positions in an official ranking.However, it may also be possible to perform recording and / or analysis even if the comparison result is negative, and, for example, generate and store player information or parts thereof during the recording and / or analysis process. In this case, subsequent manual verification may be required. Alternatively or additionally, it may be possible to issue a warning signal (possibly another one, different from the warning signal introduced below) directly upon a negative comparison result.

[0018] Further process steps include recording a video signal of the racket sport game using a recording device, for example, one or more camera units, and analyzing the recorded video signal using a computer-implemented analysis algorithm. The analysis of the recorded video signal can be performed using a processing unit that is part of the recording device (and therefore "on-site"), or that is part of a server and / or a cloud (and therefore "remotely").

[0019] As part of the analysis, at least one value (preferably a sequence of values) of at least one player performance parameter is extracted from the recorded video signal for at least one player (preferably for all players). Furthermore, the analysis involves comparing each extracted value of the player performance parameter with a player performance parameter history assigned to the respective player (via player identity and / or a group-specific comparison). This comparison can be performed using appropriate statistical methods. For example, a mean value, including a standard deviation, calculated from a sequence of values ​​of the corresponding player performance parameter, which is part of the player performance parameter history, can be compared with the extracted value or with the mean value of the extracted sequence, including the standard deviation.The statistical method(s) can also be implemented using a neural network or other forms of machine learning. For example, the analysis algorithm can include artificial intelligence (AI) that compares the extracted value(s) with the player's performance parameter history and then performs (or, based solely on the results of the conventional comparison without AI, performs the task described in the following paragraph). The AI ​​can thus check the extracted values ​​for anomalies or for values ​​that are atypical for the player, i.e., values ​​not expected based on the player's performance parameter history. It has been shown that such anomalies can be indicators of unfair play, i.e., that a player is intentionally playing worse than they are capable of.

[0020] As part of the analysis, at least one result of the comparison is used to determine whether the extracted value (and / or sequence of values) lies within a predefined tolerance range. This result may also include sub-results. In particular, the comparison result is not limited to a single numerical value; rather, it may also take the form of sequences of numbers, tables, matrices, or other formats. This determination thus provides information on whether the extracted value(s) of the respective player performance parameter(s) lies within an expected range of fluctuation and is therefore typical for the player, or whether atypical values ​​are present, indicating potentially unfair or even fraudulent play.The described procedure can therefore be used to determine whether the individual playing strength matches the course of the game (and / or match and / or set) and / or the player or player identity and / or player information (which is the case if at least one extracted value and / or the sequence of values ​​lies within the permissible tolerance range) or is atypical / conspicuous (which is the case if at least one extracted value and / or the sequence of values ​​does not lie within the permissible tolerance range).

[0021] Accordingly, as a further process step, a warning signal is issued if at least one extracted value is not within the specified tolerance range. The warning signal can be visual, audible, and / or electronic (which can, for example, be transmitted to another device). The warning signal can be graded according to further specified criteria and / or assigned to several (different) tolerance ranges. For example, a green signal can be issued, similar to a traffic light, if the extracted value is within the tolerance range, a yellow signal if the extracted value is outside the tolerance range but sufficiently close, and a red signal if the extracted value is significantly outside the tolerance range.In particular, the warning signal can be an electronic warning signal by which the recorded video signal is marked (for human review) and / or by which critical, especially game-deciding, sections of the game are marked in the video signal for which a value outside the tolerance range has been determined. While visual and / or audible warning signals have the advantage that, for example, a human referee's attention is drawn to the relevant player(s) during the game, the aforementioned electronic marking has the advantage that the human review of the game behavior can also be carried out reliably and efficiently afterward, i.e., with large quantities of available recorded video signals.Accordingly, the warning signal can be issued during the game, ideally in real time, or after the game has ended (e.g., to avoid influencing the competition).

[0022] Overall, the described method offers the advantage of detecting unfair play in an automated manner (i.e., with reduced personnel). This makes it particularly useful for uncovering instances where a player intentionally underperforms, for example, to award performance points to another player or to manipulate a bet on the outcome of a match. The proposed method is based on the understanding that fairness in a game can only be determined through the individual comparison of player behavior over different time periods (represented here by the previously known player performance parameter history on the one hand and the currently extracted values ​​of the respective player performance parameters on the other), and not by fixed, objective parameter values.This is a major problem, especially in amateur leagues, as they have significantly less financial resources compared to professional leagues, making it difficult to conduct a sufficient number of league-relevant matches due to staff shortages. Therefore, the described approach also promotes the sport overall and makes it more attractive, particularly for young players. Even outside of a professional setting, such as a tournament, league-relevant achievements, like earning ranking points in tennis, can be demonstrated, for example, on a private (tennis) court without the presence of a certified referee. Appropriate statistical methods can also be used to conduct an objective and verifiable assessment of fairness. The selected statistical method can be empirically validated and / or determined. Does the statistical method include...If the analysis algorithm is an application of a neural network or artificial intelligence, it can be trained (empirically) using known methods such as supervised learning and / or self-supervised learning and / or unsupervised learning.

[0023] Besides fair partner play, in which the players involved in the game are monitored, the procedure can also be used to combat betting fraud, where it may be sufficient to monitor one of the players, e.g. the favorite.

[0024] In one embodiment, it is provided that at least one of the player performance parameters can be determined in real time and / or includes a player reaction time and / or a player running speed and / or a player's field of vision orientation and / or a player's posture (in particular, a player's body language, i.e., one or more expressive values ​​derived from the player's posture) and / or net ball frequency and / or net ball severity and / or out-of-bounds frequency and / or out-of-bounds severity. Net ball or out-of-bounds severity can, for example, be understood here as a vertical distance to the net edge or a horizontal distance to the out-of-bounds line at the time of the fault. In this case, in a ball game in which a point is gained on a fault ball with a greater distance to the net edge or...The severity of an offense is greater than in a ball game where a point is won due to a faulty shot closer to the net or the line. Depending on the player's performance history, these parameters can be particularly effective in proving unfair play. Especially in amateur sports, players are typically not skilled enough, both generally and, as can be demonstrated by their performance history, specifically in tennis, to deliberately place a ball just a few centimeters out of bounds. Statistically speaking, a manipulative amateur player will therefore place an intentional ball out of bounds not at a closer distance (e.g., a few centimeters) but at a greater distance (e.g., several centimeters), making the deliberate underperformance easily detectable using the described method.As described below, this applies particularly to game-deciding points or important game scores.

[0025] In a further embodiment, the comparison (and / or determination) is or includes a game-point-event-induced comparison (and / or) determination (and is therefore only feasible retrospectively, i.e., not in real time, since knowledge of the event is necessary to evaluate the player's behavior before the event). In this comparison (and / or determination), the value of the player performance parameter extracted for a predetermined period during the course of the game immediately before the game-point event (for example, the ball landing outside the court or in the net, or being struck on the ball) is compared with the corresponding player performance parameter history (or the determination is only performed for the value extracted within this predetermined period).The specified time period can be defined as an absolute time value, for example, the last 10 seconds before the game point / game point win event, and / or as a flexible time value, for example, determined by a number (e.g., three) of the last ball contacts before the game point / game point win event. The course of play can be considered the entire period from the beginning to the end of the respective game, as defined in the relevant rules. The course of play can thus encompass, for example, the course of a single game, a match, or a complete set. The score progression can be defined as the change in scores throughout the course of play.In particular, a predetermined number of game-deciding events (and thus alternatively or additionally one or more parts of the game) that occur before reaching a game-deciding score can be classified as game-deciding.

[0026] For example, the aforementioned comparison result for a period such as the last three ball contacts before the end of a game, match, or set (as game-deciding) can be weighted more heavily than an earlier period of that same game, match, or set. This has the advantage of further enhancing the reliability of the automated monitoring, as the game-deciding events and thus time periods are particularly well represented in the comparison result.

[0027] The process can be designed to be or include a comprehensive analysis in which the comparative results for several, preferably all, game-point events during the entire course of the game are considered holistically to determine whether the extracted values ​​fall within the specified tolerance range. The comparative results can then form the basis of the aforementioned comparison. This has the advantage that the trend of the comparative results over, for example, a game, match, and / or set can also be evaluated, further improving the quality of the monitoring. For instance, it makes it particularly easy to identify if a player initially plays fairly, i.e., to the best of their ability, and their skill level decreases as soon as it becomes apparent that they are clearly leading or likely to win.

[0028] In particular, when considering the overall picture as described above for the comparison result, the extracted values ​​corresponding to the different game-point events can be weighted differently, especially those corresponding to game-point events classified as decisive according to a predefined criterion. This also improves the monitoring results. For example, it makes it particularly easy to detect if a player's skill diminishes (especially whenever the other player has a chance of winning, for example, a game, match, or set). This is a particularly strong indicator of cheating and therefore unfair play.

[0029] In another embodiment, it is provided that, during the determination process, any deviation between the extracted value and the player performance parameter history is quantified using a single value designated as a fairness value. Based on this fairness value, it is determined whether the extracted value is within the tolerance range or not. The fairness value is a measure of the deviation, which preferably does not have the physical dimension of the extracted value, and in particular is dimensionless. For example, deviations measured in one dimension, such as reaction times measured in milliseconds (ms), can thus be compared with deviations measured in another dimension, such as ball speeds measured in meters per second (m / s). The fairness value can be quantified, in particular, by a predefined heuristic, for example, by determining a relative deviation (e.g.,Converting the deviation in absolute numbers into a relative deviation, which quantifies the deviation measured in standard deviations, is a process. The fairness score can then be aggregated into an overall fairness score, which, as a single value, characterizes the monitored racket sport game throughout its (especially complete) course. For example, fairness scores quantified in standard deviations for individual player performance parameters can be averaged and thus converted into an overall fairness score. The fairness score can, for instance, be derived empirically for each player performance parameter. This has the advantage that an automatic classification, especially the issuing or non-issuance of the warning signal, can be easily tracked, for example, graphically as a trend of the (overall) fairness score over time.Neural networks are particularly well-suited for quantification using a single value, as they can effectively supplement incomplete datasets, such as those containing player performance parameters. Consequently, even player performance parameters with low frequency and therefore low statistical relevance can be used to assess fairness.

[0030] In a further embodiment, the player performance parameter history is provided for to be, or comprise, a player-specific player performance parameter history and includes a multitude of values ​​already known for the respective player of at least one player performance parameter. This multitude can include a predetermined minimum number of, for example, 50, 100, or 1000 known values ​​for the respective player performance parameter. In this case, the player performance parameter history is assigned to the respective player via their identity. This increases the statistical relevance and further improves the accuracy of the method.

[0031] In another embodiment, the player performance parameter history includes or considers one or more values ​​(in particular, the values ​​can be a parameter value or an individual value derived from one or more parameter values) that were extracted or generated during or in connection with the currently monitored racket sport match, especially on the current day. Alternatively or additionally, the player performance parameter history can include or consider one or more values ​​that were extracted or generated during a previously completed racket sport match, especially on a previous day. The method is therefore suitable for both established and new players.for players for whom a player performance parameter history already exists, and also for players who do not yet have a player performance parameter history, which expands flexibility and range of applications.

[0032] In a further embodiment, the player performance parameter history, particularly as part of the player information, includes at least one individual player performance reference value, specifically a player's league position and / or performance score and / or age and / or gender and / or height, and the individual player performance reference value is associated with a range of values ​​for the respective player performance parameter for comparison. For example, based on the respective individual player performance reference value, statistical data from a large number of players can be used to predict which values ​​of a given player performance parameter are to be expected and are therefore typical, i.e., within the specified permissible tolerance range.The video signal extracted from the recorded video signal is thus compared to the value range of the respective player performance parameter associated with the player's individual reference value.

[0033] For example, a higher player ranking suggests greater running speed than a lower ranking, especially if one or more physiological factors such as player age, gender, and / or height are similar or identical. Accordingly, a player's individual performance reference value can be associated with a predefined range of values ​​for the respective player performance parameter of one or more other players with sufficiently similar individual performance reference values, or vice versa. This allows for comparisons using the performance parameter values ​​of other, ideally similar, players.For example, even rarely seen maneuvers by a particular player (which are therefore difficult to statistically record) can reveal whether the player's behavior corresponds to the general expectations for players of their skill level or not (possibly intentionally). In this case, the player's performance parameter history is assigned to the respective player via a group-specific comparison. This also improves the quality of the assessment of player behavior and thus the reliability of the presented method.In another embodiment, supplementary information, in particular court size and / or game duration and / or final score and / or number of sets and / or ball type, is extracted and compared with a predefined reference supplementary information. Specifically, a warning signal is also issued if a deviation is detected between the captured supplementary information and the predefined reference supplementary information. This allows player-independent parameters that affect the comparability of the game with other games in the same racket sport to be integrated into the process. This has the advantage of further increasing fairness, i.e., making cheating or manipulation more difficult, since, for example, the chosen ball type can significantly influence the difficulty of the game.

[0034] In a further embodiment, the game score is considered as a player performance parameter, wherein a game score value (in particular, the score progression and thus the time-dependent game score) is automatically extracted as an extracted value (in particular, the extracted value progression) of this player performance parameter during the analysis of the recorded video signal, and the game score value (in particular, the score progression) is manually recorded by the player as a player performance parameter history, and both game scores are compared. This corresponds to considering the game score as a player performance parameter by comparing the automatically extracted game score value with a manually recorded game score value.

[0035] Alternatively, the score (especially the score history) can be automatically recorded by the player as a player performance parameter history, and the score can then be manually extracted as an extracted value (especially the extracted score history) of this player performance parameter. For the automatic extraction of the score, the "backwards scoring" method explained in DE 10 2018 119 170 Al can be used. This has the advantage of further verifying the plausibility of the result and making manipulation more difficult.

[0036] Another aspect concerns a device for monitoring a racket sport game, comprising a detection device for recording the identity of the players involved in the racket sport game; a recording device for capturing a video signal of the racket sport game; and an analysis device for analyzing the recorded video signal using a computer-implemented analysis algorithm. The analysis device is designed to extract at least one value of at least one player performance parameter from the recorded video signal, to compare the extracted value of the player performance parameter with a player performance parameter history, and to determine, based on the result of this comparison, whether the extracted value lies within a predefined tolerance range.The analysis device or a warning device may be further configured to issue a warning signal if at least one extracted value is outside the tolerance range. The detection device may, in particular, be configured to compare the detected identities with stored player information and to release the players for the racket sport game if the comparison is successful.

[0037] Advantages and advantageous embodiments of the latter aspect correspond to the advantages and advantageous embodiments described for the former aspect and vice versa.

[0038] The described features and combinations of features, including those in the general introduction, as well as the features and combinations of features disclosed in the figure description or the figures themselves, can be used not only alone or in the described combination, but also with other features or without some of the disclosed features, without departing from the scope of the invention. Consequently, embodiments that are not explicitly depicted and described in the figures, but which can be generated by separately combining the individual features disclosed in the figures, are also part of the invention. Therefore, embodiments and combinations of features that do not include all the features of an originally formulated independent claim are also to be considered disclosed.Furthermore, embodiments and combinations of features that differ from the combinations of features or go beyond those described in the dependencies of the claims are to be considered disclosed.

[0039] Detailed description

[0040] Exemplary embodiments are described in more detail below with reference to schematic drawings. These show

[0041] Fig. 1 shows an exemplary embodiment of a device for monitoring a racket sport game.

[0042] The illustrated embodiment of the device 1 for monitoring a racket sport game is, in this case, configured for monitoring a tennis game. The device 1 comprises a detection device 2, a recording device 3, an analysis device 4, and, in this case, also a warning device 5.

[0043] The detection device 2 is designed to capture the identity of the players participating in the racket sport game, for example, by means of a mobile device displaying a player-specific QR code or another identification method. The recording device 3 is designed to record a video signal of the racket sport game and, accordingly, comprises one or more cameras directed at the playing field 6. The analysis device 4 is designed to analyze the recorded video signal using a computer-implemented analysis algorithm. The warning device 5 is designed to issue a warning signal. The analysis device 4 and / or the warning device 5 may also be located separately from the other parts of the device 1; for example, a server or cloud for the analysis device 4 may be located away from the playing field 6.If two players 7a and 7b want to conduct an automatically monitored game, for example, so that the points scored in the game are counted towards an official ranking, their respective identities are first recorded by the data capture device 2. By comparing the recorded identities with stored player information, a player performance parameter history stored in or with the stored player information can be accessed. Alternatively or additionally, the player performance parameter history can be recreated and / or supplemented with the recorded identity and the subsequently extracted values ​​of the respective player performance parameters.

[0044] The recording device can also be configured to release players only upon a positive match result for the game, i.e., for the recording and / or analysis and / or warning signal output steps described below. For example, release to a payment and / or a match based on individual player performance reference values ​​of the two players may depend on this. The individual player performance reference values ​​may be part of the player performance parameter history and / or player information and / or individual player performance reference values ​​captured on-site by the recording device.This ensures that the points in the game correspond to the relevant known specifications for a rating within the official ranking, based on the corresponding stored specifications regarding the matching of, for example, player age and / or player gender and / or player performance score and / or player league position as an individual player performance reference value.

[0045] The recording device 3 captures a video signal of the game, which is then analyzed by the analysis device 4 using a computer-implemented analysis algorithm. In this process, at least one value (preferably several values) of at least one player performance parameter (preferably several player performance parameters) is extracted from the captured video signal for each player using known image recognition methods. For example, a player reaction time and / or a player running speed can be extracted for players 7a and 7b. The player performance parameters can be extracted situationally or continuously, for example, event-induced and / or analyzed as an average for the game.For example, one could consider how fast players 7a and 7b run on average during the game and / or how fast they run during specific events, such as before a game-winning event (with player running speed as a player performance parameter). In a highly simplified version of the proposed method, a comparison of player 7a's running speed before a game-winning event favoring player 7b with player 7a's average running speed could reveal whether player 7a intends to let player 7b win.

[0046] Accordingly, the extracted values ​​of the player performance parameter(s) are generally compared with the player performance parameter history, as described in the procedure. This comparison is performed for each player (7a, 7b). The extracted values ​​can be directly compared with the corresponding values ​​for the player performance parameter(s), for example, an extracted player running speed with a player running speed stored in the history. In this case, the player performance parameter history used can be referred to as a player-specific player performance parameter history. The comparison is performed individually, relying solely on the player's identity and their individual data.

[0047] The extracted values ​​can also be compared with the corresponding values ​​of other players with comparable individual player performance reference values.

[0048] For example, the extracted player running speed can be compared with the known running speeds of other players who, however, have matching individual player performance reference values ​​according to a predefined set of criteria, such as similar player age, gender, and performance score. In this case, the used player performance parameter history can be referred to as a group-specific player performance parameter history. The comparison is performed as a group-specific comparison using the player's identity, their individual data, and the data of matching other players. Individual and group-specific comparisons can also be combined to improve accuracy.

[0049] Based on the results of the comparison, it is then determined whether the extracted value(s) fall within a predefined tolerance range, i.e., whether they are individually expected and therefore typical for the respective player 7a, 7b. If this is not the case, a warning signal can be issued to draw attention to potential manipulation. Depending on the extent of the deviation from typical values, i.e., depending on how far the extracted value(s) lie outside the permissible tolerance range, the warning signal can be adjusted.For example, a yellow signal can indicate a smaller deviation from typical values, where, according to general experience, the deviation may also be random, and a red signal can indicate a larger deviation from typical values, where, according to general experience, the deviation can no longer be random, so manipulation is clearly to be expected.

[0050] Whether the extracted value is still within the permissible tolerance range can also be determined indirectly. For example, a single value can be quantified / derived from the extracted value(s), which can be referred to as a fairness value. If the extracted value is no longer within the tolerance range, this fairness value falls below a predefined limit, and the game is no longer considered fair. This approach has the advantage of compensating for opposing effects in a meaningful way: For example, during a game, player 7a's running speed may decrease due to fatigue. At the same time, player 7a gets to know player 7b better, resulting in improved reaction time. The fairness value, as a single value, offers a way to prevent false accusations of unsportsmanlike conduct in such a situation.

[0051] In this sense, comparison can also be structured as a game-point-event-induced comparison, particularly with weighting. Since game points are of particular importance, and especially before game-deciding game-point events (i.e., before a game / match / set), players are expected to give their best to win, the values ​​extracted in such a situation—that is, before the corresponding (especially game-deciding) game-point event—or the comparison results for the values ​​extracted in such situations, can be weighted more heavily. Such weighting can, for example, refer to a period such as 10 seconds before the event, but also to a specific point in the game, such as the last three ball contacts before the event.

[0052] In our example, if a player's reaction time is poor shortly before a game-scoring event, a warning signal is more likely to be issued than if the same player reaction time occurs generally during the game, without the other player subsequently scoring a point. This weighted comparison is particularly beneficial for the outcome of the described procedure when determining the overall outcome.

[0053] Alternatively, the tolerance range for the time before game-deciding events can be reduced in the same way. This simplifies implementation when using a single value such as the fairness value for indirect determination.

[0054] In the illustrated example game, a ball 8 is served by player 7a. The ball 8 flies along curve R over the net 9. If it were to get caught in the net 9, it would have a vertical distance dv from the net edge. The magnitude of such a distance can be incorporated into the described weighting as a measure of the severity of the error, analogous to the time of the error (e.g., before the events mentioned above): A greater distance dv of the ball 8 from the net edge is (depending on the skill level of player 7a) more likely to be attributed to an intentional net ball than a smaller distance dv.

[0055] Ball 8 now hits court 6 at position A and is returned by player 7b over net 9. The time of impact at position A can also be used to determine the relevant or particularly important sections of the game. Ball 8 lands at position B with a horizontal distance dv to court 6, out of bounds. Player 7a receives the corresponding points. The described procedure now allows for an automatic decision as to whether it is a manipulated point (in the case of a greater distance dv and a better player 7b) or a fairly won point (in the case of a shorter distance and a weaker player 7b). For this purpose, the considered section of the game before the point event can be appropriately selected as explained, and the available data can be analyzed using all methods of statistical mathematics.

Claims

Claims 1. Procedure for monitoring a racket sport game, with the following procedural steps: - Recording the identity of at least one of the players involved in the racket sport game (7a, 7b) by means of a recording device (2); - Recording a video signal of the racket sport game by a recording device (3); - Analyzing the recorded video signal using a computer-implemented analysis algorithm, characterized by - extracting at least one value of at least one player performance parameter from the recorded video signal, as well as - a comparison of the respective extracted value of the player performance parameter with a player performance parameter history, and - Determine, based on the result of a comparison, whether at least one extracted value lies within a specified tolerance range or not; - Issue a warning signal if at least one extracted value is not within the tolerance range.

2. Method according to claim 1, characterized in that at least one of the player performance parameters comprises a player reaction time and / or a player running speed and / or a player field of vision orientation and / or a player posture and / or net ball frequency and / or out-of-bounds frequency.

3. Method according to one of the preceding claims, characterized in that the comparison is or comprises a game point event-induced comparison in which the value of the player performance parameter extracted for a given period of time in the course of the game directly before the game point event is compared with the player performance parameter history.

4. Method according to the preceding claim, characterized in that the determination is or comprises a holistic determination in which the comparison results for several, preferably all, game point events during the complete course of the game are taken into account holistically in order to determine whether the extracted values ​​are within the specified tolerance range or not.

5. Method according to the preceding claim, characterized in that, when considering the totality of the data, the extracted values ​​corresponding to the different game-point events are weighted differently, in particular those deemed decisive according to a predetermined criterion. The corresponding values ​​for classified game point events will be weighted more heavily.

6. Method according to one of the preceding claims, characterized in that, in determining a deviation between extracted value and player performance parameter history, a deviation is quantified with an individual value designated as a fairness value, and it is determined on the basis of the fairness value whether the extracted value is within the tolerance range or not, wherein the fairness value is in particular combined to form an overall fairness value, which as an individual value characterizes the monitored racket sport game over the course of the game.

7. Method according to one of the preceding claims, characterized in that the player performance parameter history is a player-specific player performance parameter history and comprises a plurality of values ​​of the at least one respective player performance parameter that are already known for the respective player.

8. Method according to one of the preceding claims, characterized in that the player performance parameter history includes or takes into account one or more values ​​which were extracted or generated during or in connection with the currently monitored racket sport game, in particular on the current day, and / or values ​​which were extracted or generated during a previously completed racket sport game, in particular on a previous day.

9. Method according to one of the preceding claims, characterized in that the player performance parameter history, in particular as part of the player information, comprises at least one player performance individual reference value, in particular a player league position and / or a player performance score and / or a player age and / or a player gender, and the player performance individual reference value is associated for comparison with a range of values ​​of the respective player performance parameter.

10. Method according to the preceding claim, characterized in that the player performance individual reference value of a player (7a, 7b) is associated for comparison with a stored value range of the respective player performance parameter of one or more other players with sufficiently similar player performance individual reference value or vice versa.

11. Method according to one of the preceding claims, characterized in that supplementary information, in particular a playing field size and / or a playing duration and / or a final score and / or a number of sets, is provided. and / or a ball type is extracted and compared with a predetermined reference supplementary information, whereby in particular the warning signal is also issued if a deviation between recorded supplementary information and predetermined reference supplementary information is detected.

12. Method according to one of the preceding claims, characterized in that the game score is taken into account as a player performance parameter, wherein a game score value is automatically extracted, in particular retrospectively, when analyzing the recorded video signal, and the game score value is manually recorded by the player, and both game scores are compared with each other.

13. Device (1) for monitoring a racket sport game, comprising: - a recording device (2) for recording the identity of the players involved in the racket sport game; - a recording device (3) for recording a video signal of the racket sport game; and - an analysis device (4) for analyzing the recorded video signal using a computer-implemented analysis algorithm, characterized in that the analysis device (4) is designed - to extract at least one value of at least one player performance parameter from the recorded video signal, as well as - to compare the respective extracted value of the player performance parameter with a player performance parameter history, and - to determine, based on a comparison result, whether at least one extracted value lies within a predefined tolerance range or not; and by - a warning device (5) for issuing a warning signal if at least one extracted value is not within the tolerance range.

Images