Systems and methods for re-identifying objects in images

Number: 10127668
Published: 2018-11-13
Filed: 2016-03-04
Assignee: Disney Enterprises, Inc.
Inventors: Bak; Slawomir W.; Carr; George Peter
CPC: G06V40/103; G06V20/52; G06T7/73
Verdict: Low Notable software
Source: Google Patents · FreePatentsOnline

The keeper's note

CV object re-identification.

Abstract

There is provided a system including a memory and a processor configured to receive a first image depicting a first object and a second image depicting a second object, divide the first image into a first plurality of patches and the second image into a second plurality of patches, extract plurality of feature vectors from each of the first plurality of patches and a second plurality of feature vectors from the second plurality of patches, determine dissimilarities based on a plurality of patch metrics, each patch dissimilarity measure being a dissimilarity between corresponding patches of the first plurality of patches and the second plurality of patches, compute an image dissimilarity between the first image and the second image based on an aggregate of the plurality of patch dissimilarity measures, evaluate the image dissimilarity to determine a probability of whether the first object and the second object are the same.

Background

BACKGROUND(1) Re-identifying individuals in images can be a difficult task, because many images are not taken with sufficiently high resolution to use facial recognition software. Conventional methods of re-identification depend on a comparison of a first total image to a second total image. Comparing the two total images, however, requires compressing image data for each image by one or more orders of magnitude, resulting in a significant loss of data and resolution. As a result, conventional methods are error prone and may return false negatives due to, among other things, differing conditions between the images being compared, such as different lighting and a change in pose of the individual.SUMMARY(2) The present disclosure is directed to systems and methods for re-identifying objects in images, substantially as shown in and/or described in connection with at least one of the figures, as set forth more completely in the claims.

Claims

1. A method for use by a system comprising a non-transitory memory and a hardware processor, the method comprising: receiving, using the hardware processor, a first image depicting a first object and a second image depicting a second object; dividing, using the hardware processor, the first image into a first plurality of patches and the second image into a second plurality of patches; extracting, using the hardware processor, a first plurality of feature vectors from each of the first plurality of patches and a second plurality of feature vectors from each of the second plurality of patches; repositioning, using the hardware processor, each patch of the first plurality of patches based on a deformation cost for each patch of the first plurality of patches; determining, using the hardware processor, a plurality of patch dissimilarity measures based on a plurality of patch metrics, each patch dissimilarity measure being a dissimilarity between corresponding patches of the repositioned first plurality of patches and the second plurality of patches; computing, using the hardware processor, an image dissimilarity between the first image and the second image based on an aggregate of the plurality of patch dissimilarity measures; and evaluating, using the hardware processor, the image dissimilarity to determine a probability of whether the first object and the second object are the same. 11. A system comprising: a non-transitory memory storing an executable code; and a hardware processor executing the executable code to: receive a first image depicting a first object and a second image depicting a second object; divide the first image into a first plurality of patches and the second image into a second plurality of patches; extract a first plurality of feature vectors from each of the first plurality of patches and a second plurality of feature vectors from each of the second plurality of patches; reposition each patch of the first plurality of patches based on a deformation cost for each patch of the first plurality of patches; determine a plurality of patch dissimilarity measures based on a plurality of patch metrics, each patch dissimilarity measure being a dissimilarity between corresponding patches of the repositioned first plurality of patches and the second plurality of patches; compute an image dissimilarity between the first image and the second image based on an aggregate of the plurality of patch dissimilarity measures; and evaluate the image dissimilarity to determine a probability of whether the first object and the second object are the same.