Systems and methods for visual object matching

Patent metadata

Field	Value
Publication number	US-9020246-B2
Application number	US-201314045207-A
Country	US
Kind code	B2
Filing date	Oct 3, 2013
Priority date	Jul 12, 2011
Publication date	Apr 28, 2015
Grant date	Apr 28, 2015

Abstract

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.

First claim

What is claimed is:

1. A computer-implemented method comprising:
obtaining a query image that has been submitted by a user of an image search system;
determining, by the image search system, that the query image is indicated to be associated with an object;
obtaining, for the query image: (i) one or more training images of a corpus of training images that are indicated to be associated with the object, and (ii) for each one of the one or more training images that are indicated…

Assignees

• Google Inc

Inventors

• Li Yuan
• Adam Hartwig

Classifications

• G06F18/214 Primary

  Physics · mapped topic

• G06K9/62

  Physics · mapped topic

• G06K9/6256

  Physics · mapped topic

• G06K9/66 Primary

  Physics · mapped topic