Method And Device For Tracking Sports Players with Context-Conditioned Motion Models

Number: 20180197296
Published: 2018-07-12
Filed: 2017-12-29
Assignee: Disney Enterprises, Inc.
Inventors: LIU; Jingchen; Carr; G. Peter K.
CPC: G06T7/20
Verdict: Low Notable software
Source: Google Patents · FreePatentsOnline

The keeper's note

Sports-player CV tracking with motion models (PGPUB dup).

Abstract

A method and device generates a trajectory. The method includes receiving a plurality of tracklets indicative of movement of a plurality of targets over a predetermined temporal interval. The method includes determining a plurality of context data for a pair of tracklets based upon at least one additional tracklet. The method includes computing a probability that the pair of tracklets relate to a first one of the targets. The method includes generating a trajectory for the first target based upon a concatenation of select ones of the tracklets. The concatenation maximizes the probability that the pair of tracklets correspond to the first target based upon the context data associated with the pair of the tracklets.

Background

BACKGROUND INFORMATION

A model of probabilistic object motion is required to track objects as they move so that a tracking algorithm may determine a trajectory from a set of hypotheses that is the most realistic. If multiple objects are being tracked concurrently, prior data is used to evaluate the feasibility of a set of simultaneous trajectories for all the objects being tracked. Conventionally, most multi-object tracking algorithms utilize a drastic simplification to keep the inference problem tractable. Specifically, each of the objects is evaluated in isolation such that each object is tracked independently of other objects. However, object motions are not always independent. In team sports, for example, player movements are highly correlated to both nearby and distant players. It is highly inaccurate to evaluate each trajectory using the conventional method of isolating each object. However, it is also quite difficult to determine and optimize a complex model which describes all possible interactions between players.

With regard to multi-target tracking, this has been a difficult problem of broad interest in the technical field of computer vision. Surveillance is a common scenario in which multi-target tracking is utilized. Team sports are another popular domain utilizing multi-target tracking that has a wide range of applications in strategy analysis, automated broadcasting, and content-based retrieval. Recent developments in pedestrian tracking have utiliz

Claims

21. A method, comprising: receiving a plurality of tracklets indicative of movement of a plurality of targets over a predetermined temporal interval; determining context data for a pair of tracklets of the plurality of tracklets based on at least one additional tracklet of the plurality of tracklets; computing a probability that the pair of tracklets relate to a first target of the plurality of targets, wherein the probability is computed based on a regression; and generating a trajectory for the first target based on a concatenation of select ones of the plurality of tracklets, wherein the concatenation maximizes the probability that the pair of tracklets correspond to the first target based on the context data of the pair of tracklets. 31. A device, comprising: a processor coupled to a memory, wherein the processor is programmed to generate a trajectory for a first target of a plurality of targets by: receiving a plurality of tracklets indicative of movement of the plurality of targets over a predetermined temporal interval; determining context data for a pair of tracklets of the plurality of tracklets based on at least one additional tracklet of the plurality of tracklets; computing a probability that the pair of tracklets relate to the first target, wherein the probability is computed based on a regression; and generating a trajectory for the first target based on a concatenation of select ones of the plurality of tracklets, wherein the concatenation maximizes the probability that the pair of tracklets correspond to the first target based on the context data of the pair of tracklets. 40. A non-transitory computer readable storage medium with an executable program stored thereon, wherein the program instructs a processor to perform operations comprising: receiving a plurality of tracklets indicative of movement of a plurality of targets over a predetermined temporal interval; determining context data for a pair of tracklets of the plurality of tracklets based on at least one additional tracklet of the plurality of tracklets; computing a probability that the pair of tracklets relate to a first target of the plurality of targets, wherein the probability is computed based on a regression; and generating a trajectory for the first target based on a concatenation of select ones of the tracklets, wherein the concatenation maximizes the probability that the pair of tracklets correspond to the first target based on the context data of pair of the tracklets.

Claims truncated at the source; see the full document.