- Number
- 10140575
- Published
- 2018-11-27
- Filed
- 2016-05-16
- Assignee
- Disney Enterprises, Inc.
- Inventors
- Lucey; Patrick J. et al.
- CPC
- G06V20/42; G06N5/022; G06F3/04842; G06F3/0481
- Verdict
- Set aside sports analytics/business
- Source
- Google Patents · FreePatentsOnline
Abstract
Approaches are described for formation retrieval. Embodiments receive positional data, across an interval window, including a respective agent trajectory for each agent and an object trajectory for one or more objects. The interval window is partitioned into frames and, at each frame, embodiments calculate a cost of assigning a role to each agent based on one or more exemplar formations. A formation is determined by assigning a role to each agent based on the calculated cost. Each frame of the formation is compared to a corresponding frame of a stored formation, by calculating a distance between a position of each assigned role in the frame and a position of a corresponding role in the stored formation and by comparing the object trajectory for the one or more objects with a corresponding object trajectory in the stored formation. Based on the comparisons, a list of stored formations is generated.
Background
BACKGROUND OF THE INVENTION(1) Field of the Invention(2) The present invention relates to artificial intelligence systems used to analyze participants engaged in an activity and, in particular, to tracking player role using non-rigid formation priors.(3) Description of the Related Art(4) Vision-based systems have been deployed to detect and track players engaged in adversarial team sports. Such vision-based systems may record the position of each player multiple times per second over the course of play. For example, a vision-based system may generate location data for each player at a rate of thirty times per second. These systems may support applications configured to perform various analysis functions, including, characterizing offensive patterns, recognizing individual plays, and predicting the evolution of play in a game or match. Typically, such systems characterize team behavior using a macroscopic representation of the team, such as the position of the centroid of the team players or the distribution pattern of the team players on the playing field, or analyzing a single player over time. Such representations fail to characterize how individual players or teams perform over time, particularly if the role of one or more players changes as play progresses.(5) In addition, tracking data for players or teams may exhibit high dimensionality, where the quantity of samples collected over long periods of play may be too numerous for applications to efficiently analyze and prod
Claims
1. A method of sports play retrieval, comprising: receiving positional data, across an interval window, comprising a respective agent trajectory for each of a plurality of agents during the interval window; partitioning the interval window into a plurality of frames; at each of the plurality of frames, calculating a cost of assigning a role to each agent of the plurality of agents based on one or more exemplar formations; determining a formation by assigning a role to each agent based on the calculated cost; comparing each frame of the plurality of frames of the formation to a corresponding frame of a stored formation, by calculating a distance between a position of each assigned role in the frame and a position of a corresponding role in the stored formation; and generating, based on the comparisons, a list of stored formations that are similar to the formation.
10. A non-transitory computer-readable storage medium including instructions that, when executed by a processor, cause the process to retrieve sports plays, by performing an operation comprising: receiving positional data, across an interval window, comprising a respective agent trajectory for each of a plurality of agents during the interval window; partitioning the interval window into a plurality of frames; at each of the plurality of frames, calculating a cost of assigning a role to each agent of the plurality of agents based on one or more exemplar formations; determining a formation by assigning a role to each agent based on the calculated cost; comparing each frame of the plurality of frames of the formation to a corresponding frame of a stored formation, by calculating a distance between a position of each assigned role in the frame and a position of a corresponding role in the stored formation; and generating, based on the comparisons, a list of stored formations that are similar to the formation.
19. A computing system, comprising: a memory that is configured to store instructions for a program; and a processor that is configured to execute the instructions for the program to retrieve sports plays, by performing an operation comprising: receiving positional data, across an interval window, comprising a respective agent trajectory for each of a plurality of agents during the interval window; partitioning the interval window into a plurality of frames; at each of the plurality of frames, calculating a cost of assigning a role to each agent of the plurality of agents based on one or more exemplar formations; determining a formation by assigning a role to each agent based on the calculated cost; comparing each frame of the plurality of frames of the formation to a corresponding frame of a stored formation, by calculating a distance between a position of each assigned role in the frame and a position of a corresponding role in the stored formation; and generating, based on the comparisons, a list of stored formations that are similar to the formation.