Yuan Ming Li

8438163, May 7, 2013

Dec 7, 2011

13/314061

Yuan Li - Los Angeles CA, US
Hartwig Adam - Marina del Rey CA, US

Google Inc. - Mountain View CA

G06F 17/30

707737, 707706, 707722, 707754, 707915, 382159, 382195, 382209

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for automatically extracting logos from images. Methods include generating a query list including a plurality of logo search queries, for each logo search query of the plurality of logo search queries: generating a plurality of image search results, each image search result including image data, and clustering the plurality of image search results into a plurality of clusters, each cluster including a plurality of images of the plurality of image search results, extracting, for each cluster of the plurality of clusters, a representative image to provide a plurality of representative images, and a name corresponding to the representative image to provide a plurality of names, and providing the plurality of representative images and the plurality of names to a logo index, the logo index being accessible to identify one or more logo images in a query image.

8548195, Oct 1, 2013

Jun 13, 2008

12/664588

Haizhou Ai - Beijing, CN
Yuan Li - Los Angeles CA, US
Shihong Lao - Kyoto, JP
Takayoshi Yamashita - Kizugawa, JP

Omron Corporation - Kyoto
Tsinghua University - Beijing

G06K 9/00

382103, 382165

The present invention relates to a tracking method and a tracking device adopting multiple observation models with different life spans. The tracking method is suitable for tracking an object in a low frame rate video or with abrupt motion, and uses three observation models with different life spans to track and detect a specific subject in frame images of a video sequence. An observation model I performs online learning with one frame image prior to the current image, an observation model II performs online learning with five frames prior to the current image, and an observation model III is offline trained. The three observation models are combined by a cascade particle filter so that the specific subject in the low frame rate video or the object with abrupt motion can be tracked quickly and accurately.

8577131, Nov 5, 2013

Jul 12, 2011

13/181015

Yuan Li - Los Angeles CA, US
Hartwig Adam - Marina del Rey CA, US

Google Inc. - Mountain View CA

G06K 9/62

382159, 382157, 382160, 382161, 382209

Systems and methods for improving visual object recognition by analyzing query images are disclosed. In one example, a visual object recognition module may determine query images matching objects of a training corpus utilized by the module. Matched query images may be added to the training corpus as training images of a matched object to expand the recognition of the object by the module. In another example, relevant candidate image corpora from a pool of image data may be automatically selected by matching the candidate image corpora against user query images. Selected image corpora may be added to a training corpus to improve recognition coverage. In yet another example, objects unknown to a visual object recognition module may be discovered by clustering query images. Clusters of similar query images may be annotated and added into a training corpus to improve recognition coverage.

8625887, Jan 7, 2014

Jul 13, 2011

13/182076

Yuan Li - Los Angeles CA, US
Hartwig Adam - Marina del Rey CA, US

Google Inc. - Mountain View CA

G06K 9/62

382159, 382209

Systems and methods for modeling the occurrence of common image components (e. g. , sub-regions) in order to improve visual object recognition are disclosed. In one example, a query image may be matched to a training image of an object. A matched region within the training image to which the query image matches may be determined and a determination may be made whether the matched region is located within an annotated image component of the training image. When the matched region matches only to the image component, an annotation associated with the component may be identified. In another example, sub-regions within a plurality of training image corpora may be annotated as common image components including associated information (e. g. , metadata). Matching sub-regions appearing in many training images of objects may be down-weighted in the matching process to reduce possible false matches to query images including common image components.

56852819, Nov 11, 1997

Apr 22, 1996

8/635663

Yuan Li - Alhambra CA

F02M 2900

123590

A gas vortex device for an internal combustion engine includes at least a guider body installed in a predetermined position of a flowing passage provided between an inlet chamber of a cylinder body and an exhaling end of an air inputting arrangement of the internal combustion engine. The guider body which has an inlet end and an outlet end installed in such a manner that the outlet end should be more proximate to the inlet chamber of the cylinder body than the inlet end. The guider body further has an axial portion and at least two guiding wings extending symmetrically, outwardly and radially from the axial portion. Therefore, a gas mixture including the air and the atomized fuel that are sucked into the flowing passage from the air inputting arrangement are forced to flow through the guider body, inhaling through the inlet end and exhaling from the outlet end thereof before sucking into the inlet chamber of the cylinder body, so that the gas flow is guided by the guiding wings to spin and speed up such whirling motion. Therefore the gas mixture sucking into the inlet chamber of the cylinder body is spinning continuously in vortex form, so as to further atomize the atomized fuel particles to more diminutive tiny particles and more evenly and completely mix the air and atomized fuel particles.

20180260759, Sep 13, 2018

Mar 7, 2018

15/914416

- Seattle WA, US
Daryn Edward Nakhuda - Bainbridge Island WA, US
Angela Beth Hugeback - Bainbridge Island WA, US
Yuan Li - Bellevue WA, US
Peter VanTuyl Bentley - Seattle WA, US
Joseph Delovino Sunga - Seattle WA, US
Aaron Matthew Hedquist - Seattle WA, US
Matthew Cameron Herz - Seattle WA, US

G06Q 10/06
G06T 7/10

Disclosed is a configuration for segmenting an image. For each user, the configuration computes an accuracy and contribution score. The configuration determines multiple tasks for segmenting an image. For each task, the configuration assigns a particular user to work on a task based on an accuracy or a contribution score of the particular user, receives an indication of a completed task from the particular user, and assesses an accuracy of the completed task based on the accuracy of the particular user. Responsive to determining all multiple tasks are completed accurately, the configuration combines the completed multiple tasks to form a segmented image.

20170373958, Dec 28, 2017

Jun 27, 2016

15/193113

- Redmond WA, US
Yuan Li - Sammamish WA, US

H04L 12/26
H04L 29/08
H04L 29/06

Aspects extend to methods, systems, and computer program products for orchestrating operations at applications. Aspects of the invention use a side channel (instrumentation messages generated by a service) as a mechanism to discover when a service has completed an activity. Use of a side channel, allows a (e.g., client) application to create behaviors similar to service side interfaces and/or protocols without modifying a service. Accordingly, functionality can be added incrementally, safely, and cheaply without having to revise an underlying implementation. In one aspect, an instrumentation collector and publisher (ICP) facilitates the synchronization between services and an application. ICP is a scalable infrastructure that provides applications a way to interact with servers through instrumentation.

20170323211, Nov 9, 2017

May 5, 2017

15/588478

- Seattle WA, US
Angela Beth Hugeback - Bainbridge Island WA, US
Yuan Li - Bellevue WA, US
Daryn Edward Nakhuda - Bainbridge Island WA, US
Patrick Emmett O'Donnell - Seattle WA, US
Matthew Newman Shobe - Mercer Island WA, US

G06N 5/04
G06Q 10/06

Disclosed is a system (and process) for determining the accuracy of computerized tasks in a task batch. The system calculates a number of reviews to assess the accuracy of a task based on a source accuracy and a reviewer accuracy. The source accuracy is based factors calculated by a predictive model, the factors including a historical accuracy of an authoring user. The reviewer accuracy is based on a true positive rate and a true negative rate of one or more reviewers of the task batch. The system transmits sourced tasks to a same number of reviewers. The system collects reviews and assesses if the task passes review based on the collected number of reviews.