By a News Reporter-Staff News Editor at Politics & Government Week -- According to news reporting originating from Washington, D.C., by VerticalNews journalists, a patent application by the inventors HONG, Zhibin (Sydney, AU); MEI, Xue (Ann Arbor, MI); CHEN, Zhe (Sydney, AU); WANG, Chaohui (Bussy-Saint-Georges, FR); PROKHOROV, Danil (Ann Arbor, MI); TAO, Dacheng (Sydney, AU), filed on May 19, 2015, was made available online on December 1, 2016.
The assignee for this patent application is Toyota Motor Engineering & Manufacturing North America, Inc.
Reporters obtained the following quote from the background information supplied by the inventors: "The background description provided herein is for the purpose of generally presenting the context of the disclosure. Work of the presently named inventors, to the extent the work is described in this background section, as well as aspects of the description that may not otherwise qualify as prior art at the time of filing, are neither expressly nor impliedly admitted as prior art against the present disclosure.
"Variations in the appearance of a tracked object, such as changes in geometry, photometry, camera viewpoint, illumination, or partial occlusion, pose a major challenge to object tracking. The object tracking devices developed thus far can primarily be classified into two categories: short-term tracking devices and long-term tracking devices, respectively.
"Short-term tracking devices are based on a short-term memory model and employ techniques such as incremental learning of a low-dimensional subspace of the target representation and sparse representation of tracking. The short-term memory of the target appearance is modelled using a small set of target instances and discriminative techniques that consider both background and foreground information. Short-term tracking devices depend on spatiotemporal consistency of visual clues and tend to be inefficient in complex object tracking applications.
"In contrast, long-term tracking devices employ a long-term memory model and can be differentiated based on their choice of appearance models and inference algorithms. A long term object tracker such as the tracking-learning and detection (TLD) tracker employs two experts to identify the false negatives and false positives in order to train the detector. Additionally, some long term trackers model target appearances using oversampled local features, whereas other long term trackers are based on self-paced learning schemes in which the target appearance is learned by selecting trustworthy frames. A drawback of the above long term trackers is that they update the appearance model in a rather conservative manner and are thus not able to handle fast appearance changes that occur in a short time period.
"Accordingly, there is a requirement for an object tracking device that addresses the above stated drawbacks of object trackers and boosts tracking performance."
In addition to obtaining background information on this patent application, VerticalNews editors also obtained the inventors' summary information for this patent application: "An object tracking device according to an exemplary embodiment, boosts tracking performance and addresses problems of complex object tracking, wherein the object may become occluded or leave the field-of-view. Specifically, the present disclosure provides for an object tracking device that adapts to changes in object appearances during tracking. The object tracking device (referred to herein as a multi-store tracker) is based on the Atkinson-Shiffrin memory model. The multi-store tracker includes one short-term memory store component and one long-term memory store component, which collaboratively process the input image and track the object.
"According to one embodiment, an integrated correlation filter (ICF) that stores short-term memory and depends on spatiotemporal consistency is employed in the short-term store to perform short-term tracking via a two-stage filtering process. Additionally, the multi-store tracker includes a long-term memory component that is based on key-point matching-tracking and random-sample-consensus (RANSAC) estimation. The long-term memory component interacts with a key-point feature database and controls the final output as well as the short-term memory states. Further, in order to maintain a reasonable size for the key-point feature database, the multi-store tracker updates the key-point feature database based on a forgetting curve model, thereby retaining only the useful object features.
"The foregoing paragraphs have been provided by way of general introduction, and are not intended to limit the scope of the following claims. The described embodiments together, with further advantages, will be best understood by reference to the following detailed description taken in conjunction with the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
"Various embodiments of this disclosure that are provided as examples will be described in detail with reference to the following figures, wherein like numerals reference like elements, and wherein:
"FIG. 1 depicts an exemplary schematic diagram illustrating an Atkinson-Shiffrin memory model;
"FIG. 2 depicts, according to an embodiment, a block diagram illustrating components of an object tracking device;
"FIG. 3 illustrates an exemplary flowchart depicting the steps performed by the multi-store tracker to track an object;
"FIGS. 4A and 4B depict exemplary graphs illustrating the performance of the multi-store tracker;
"FIG. 5 depicts a graph illustrating a comparison of an F-score of the multi-store tracker and other trackers; and
"FIG. 6 illustrates a block diagram of a computing device according to one embodiment."
For more information, see this patent application: HONG, Zhibin; MEI, Xue; CHEN, Zhe; WANG, Chaohui; PROKHOROV, Danil; TAO, Dacheng. Apparatus and Method for Object Tracking. Filed May 19, 2015 and posted December 1, 2016. Patent URL: http://appft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&u=%2Fnetahtml%2FPTO%2Fsearch-adv.html&r=2647&p=53&f=G&l=50&d=PG01&S1=20161124.PD.&OS=PD/20161124&RS=PD/20161124
Keywords for this news article include: Toyota Motor Engineering & Manufacturing North America Inc.
Our reports deliver fact-based news of research and discoveries from around the world. Copyright 2016, NewsRx LLC