Real time kinematic analyses of body motion

Number: 11514722
Published: 2022-11-29
Filed: 2020-11-12
Assignee: Disney Enterprises, Inc.
Inventors: Prince; Kevin John, Dietrich; Carlos Augusto, Van Dall; Dirk Edward
CPC: G06V20/44; G06V40/28; G06V10/762; G06V20/42; G06F18/23; G06T7/55; G06T7/251; G06V40/23
Verdict: Low Notable software
Source: Google Patents · FreePatentsOnline

The keeper's note

Real-time kinematic body-motion analysis technique (granted).

Abstract

Systems and methods are presented for generating statistics associated with a performance of a participant in an event, wherein pose data associated with the participant, performing in the event, are processed in real time. Pose data associated with the participant may comprise positional data of a skeletal representation of the participant. Actions performed by the participant may be determined based on a comparison of segments of the participant's pose data to motion patterns associated with actions of interests.

Background

BACKGROUND (1) Professional games, such as combat sports, generate an abundance of information that is hard for viewers to follow and appreciate in real time. For example, in a boxing match, the occurrence of a play action carried out by a player (such as a jab, a cross, an undercut, or a hook) as well as the play action's forcefulness and effectiveness may be hard for a viewer to immediately recognize and quantify by mere visual inspection of the player's performance during the game. Likewise, comparative and cumulative analyses of the movements of performing participants during a live event cannot be accomplished without an automated system. Analyses of the live event's video to detect and to measure the participants' performances, can facilitate statistics in real time. In a sporting event, for example, game statistics, generated as the game unfolds, may support commentary and may provide insights into the development of the game by intuitive visualization of the game statistics. (2) Analyzing a live event to detect play actions performed by participants of the event requires techniques for real time detection of actions of interest, employable on video feeds that capture the live event. Various action recognition techniques have been proposed for applications such as human-machine interfaces, video indexing and retrieval, video surveillance, and robotics, for example. However, detecting play actions in real time, that by their nature are of high motion and spatiotemporall

Claims

1. A method, comprising: receiving a sequence of pose data associated with a performance of a participant in an event; detecting, based on the sequence of pose data, one or more actions performed by the participant; and generating, based on a kinetic energy generated by the detected one or more actions, statistics associated with the performance of the participant in the event. || 12. An apparatus, comprising: at least one processor; and memory storing instructions that, when executed by the at least one processor, cause the apparatus to: receive a sequence of pose data associated with a performance of a participant in an event; detect, based on the sequence of pose data, one or more actions performed by the participant; and generate, based on a kinetic energy generated by the detected one or more actions, statistics associated with the performance of the participant in the event. || 15. A non-transitory computer-readable medium comprising instructions executable by at least one processor to perform a method, the method comprising: receiving a sequence of pose data associated with a performance of a participant in an event; detecting, based on the sequence of pose data, one or more actions performed by the participant; and generating, based on a kinetic energy generated by the detected one or more actions, statistics associated with the performance of the participant in the event. || 19. A method, comprising: receiving a sequence of pose data associated with a performance of a participant in an event; detecting, based on the sequence of pose data, one or more actions performed by the participant, the detecting comprises: extracting a segment of pose data from the sequence of pose data, comparing the extracted segment to one or more motion patterns, wherein each of the one or more motion patterns is associated with an action of interest, and determining, based on a match between the extracted segment and at least one motion pattern of the one or more motion patterns, an action performed by the participant, wherein the determined action corresponds to an action of interest associated with the at least one matching motion pattern; and generating, based on a kinetic energy generated by the detected one or more actions, statistics associated with the performance of the participant in the event.