Search

EP-4034962-B1 - SYSTEM AND METHOD FOR IMPROVED STRUCTURAL DISCOVERY AND REPRESENTATION LEARNING OF MULTI-AGENT DATA

EP4034962B1EP 4034962 B1EP4034962 B1EP 4034962B1EP-4034962-B1

Inventors

  • HOBBS, Jennifer
  • LUCEY, PATRICK

Dates

Publication Date
20260506
Application Date
20200925

Claims (14)

  1. A computer-implemented method of learning player distribution and role assignments from player tracking data captured by a tracking system (102) configured to record motion of players on a playing surface during a sporting event, the method comprising: retrieving, by a computing system (104), player tracking data for a plurality of players across a plurality of events, the player tracking data comprising coordinates of player positions during each event; characterised by initializing, by the computing system, the player tracking data based on an average position of each player in the plurality of events; learning, by the computing system, an optimal formation of player positions based on the player tracking data using a Gaussian mixture model; aligning, by the computing system, the optimal formation of player positions to a global template by identifying a distance between each distribution in the optimal formation and each distribution in the global template to generate a learned formation template; and assigning, by the computing system, a role to each player in the learned formation template.
  2. A non-transitory computer readable medium comprising instructions which, when executed by a computing system (104), cause the computing system to carry out the method of claim 1.
  3. The method of claim 1 or the non-transitory computer readable medium of claim 2, further comprising: generating, by the computing system, aligned data comprising a per-frame ordered role assignment of players; and clustering, by the computing system, the aligned data to identify new formations.
  4. The method of claim 3 or the non-transitory computer readable medium of claim 3, wherein the clustering comprises a flat or hierarchical clustering algorithm.
  5. The method of claim 1, claim 3 or claim 4 or the non-transitory computer readable medium of claim 2, claim 3 or claim 4, further comprising: filtering, by the computing system, the player tracking data to identify event frames corresponding to frames of tracking data in which an event occurs.
  6. The method of claim 1 or of any of claims 3 to 5 or the non-transitory computer readable medium of claim 2 or of any of claims 3 to 5, further comprising: normalizing, by the computing system, the player tracking data so that an attacking trajectory of all players in the player tracking data is from left to right of the playing surface.
  7. The method of claim 1 or of any of claims 3 to 6 or the non-transitory computer readable medium of claim 2 or of any of claims 3 to 6, wherein learning, by the computing system, the optimal formation of player positions based on the player tracking data using the Gaussian mixture model comprises: parametrizing a distribution of player positions as a mixture of K Gaussians to identify the optimal formation.
  8. The method of claim 1 or of any of claims 3 to 7 or the non-transitory computer readable medium of claim 2 or of any of claims 3 to 7, wherein learning, by the computing system, the optimal formation of player positions based on the player tracking data using the Gaussian mixture model comprises: monitoring eigenvalues throughout the learning to determine if an eigenvalue ratio is outside a range of acceptable values; and upon determining that the eigenvalue ratio is outside the range of acceptable values, resetting the Gaussian mixture model before continuing the learning.
  9. A system to learn player distribution and role assignments from player tracking data captured by a tracking system (102) configured to record motion of players on a playing surface during a sporting event, the system comprising: a processor; and a memory having programming instructions stored thereon, which, when executed by the processor, performs operations comprising: receiving a request from a client device to identify a team's formation and role assignment across a selected subset of games, wherein the request defines a context within each game of the subset of games; retrieving player tracking data for the selected subset of games, the player tracking data comprising coordinates of player positions during each game; filtering the player tracking data to identify frames corresponding to the defined context; learning an optimal formation of player positions based on the player tracking data and the defined context using a Gaussian mixture model to generate a learned formation template; assigning a role to each player in the learned formation template; and generating a graphical representation of a structured representation of a team's formation across the subset of games for the defined context.
  10. The system of claim 9, wherein the operations further comprise: generating aligned data comprising a per-frame ordered role assignment of players; and clustering the aligned data to identify new formations.
  11. The system of claim 10, wherein the clustering comprises a flat or hierarchical clustering algorithm.
  12. The system of claim 9, claim 10 or claim 11, wherein the defined context corresponds to an in-game situation.
  13. The system of claim 9 or of any of claims 10 to 12, wherein learning the optimal formation of player positions based on the player tracking data using the Gaussian mixture model comprises: parametrizing a distribution of player positions as a mixture of K Gaussians to identify the optimal formation.
  14. The system of claim 9 or of any of claims 10 to 13, wherein learning the optimal formation of player positions based on the player tracking data using the Gaussian mixture model comprises: monitoring eigenvalues throughout the learning to determine if an eigenvalue ratio is outside a range of acceptable values; and upon determining that the eigenvalue ratio is outside the range of acceptable values, resetting the Gaussian mixture model before continuing the learning.

Description

Cross-Reference to Related Applications This application claims priority to U.S. Application Serial No. 62/907,133, filed September 27, 2019. Field of the Disclosure The present disclosure generally relates to a system, non-transitory computer readable medium, and method for learning player distribution and role assignments in sports. Background Increasingly, sports fans and data analysts have become entrenched in sports analytics. In some situations, especially on the team-side and analyst-side of sports analytics, predicting an opponent's formation could be critical to a team's strategy heading into a game or match. The act of predicting an opponent's or team's formation has not been a trivial task, however. There is an inherent permutation disorder in team sports, which increases the difficulty at which a system can predict a team's formation or a positioning of a team's players on a playing surface given limited information. The publication "Modelling and Predicting Adversarial Behaviour using Large Amounts of Spatiotemporal Data" by Xinyu (Felix) Wei, 2016, Ph.D. thesis, Queensland University of Technology, XP093083474, https://eprints.qut.edu.au/101959/1/ Xinyu_Wei_Thesis.pdf relates to modelling and predicting adversarial behaviour in both a single agent environment (i.e. tennis) as well as multi-agent environments (i.e. soccer, basketball). Given large amounts of spatio-temporal data, the goal is to learn an accurate model that can predict the behaviour of a "specific" player or a team against a "specific" opponent in a given match context. Summary The invention is defined by the independent claims, the dependent claims relate to embodiments. Brief Description of the Drawings So that the manner in which the above recited features of the present disclosure can be understood in detail, a more particular description of the disclosure, briefly summarized above, may be had by reference to embodiments, some of which are illustrated in the appended drawings. It is to be noted, however, that the appended drawings illustrated only typical embodiments of this disclosure and are therefore not to be considered limiting of its scope, for the disclosure may admit to other equally effective embodiments. Figure 1 is a block diagram illustrating a computing environment, according to example embodiments.Figures 2A is a block diagram illustrating a portion of a formation-learning process, according to example embodiments.Figures 2B is a block diagram illustrating a portion of a formation-learning process, according to example embodiments.Figure 3A is a flow diagram illustrating a method of predicting a formation of a team, according to example embodiments.Figure 3B is a flow diagram illustrating a method of identifying formation templates for teams across a subset of games, according to example embodiments.Figure 4A illustrates charts showing an exemplary player distribution, according to example embodiments.Figure 4B illustrates charts showing an exemplary player distribution, according to example embodiments.Figure 5A is a block diagram illustrating a computing device, according to example embodiments.Figure 5B is a block diagram illustrating a computing device, according to example embodiments. To facilitate understanding, identical reference numerals have been used, where possible, to designate identical elements that are common to the figures. It is contemplated that elements disclosed in one embodiment may be beneficially utilized on other embodiments without specific recitation. Detailed Description Central to all machine leaming algorithms is data representation. For multi-agent systems, selecting a representation which adequately captures the interactions among agents is challenging due to the latent group structure which tends to vary depending on various contexts. However, in multi-agent systems with strong group structure, a system may be configured to simultaneously learn this structure and map a set of agents to a consistently ordered representation for further learning. One or more techniques provided herein include a dynamic alignment method that provides a robust ordering of structured multi-agent data, which allows for representation learning to occur in a fraction of the time compared to conventional methods. The natural representation for many sources of unstructured data is generally intuitive. For example, for images, a two-dimensional pixel representation; for speech, a spectrogram or linear filter-bank features; and for text, letters and characters. All of these representations possess a fixed, rigid structure in space, time, or sequential ordering which may be amenable for further learning. For other unstructured data sources, such as point clouds, semantic graphs, and multi-agent trajectories, such an initial ordered structure does not naturally exist. These data sources may generally be set or graph-like in nature, and therefore, the natural representation may be unordered, posing a signific