Exploring the representation capabilities of the HOG descriptor

TY - GEN

T1 - Exploring the representation capabilities of the HOG descriptor

AU - Tatu, Aditya Jayant

AU - Lauze, Francois Bernard

AU - Nielsen, Mads

AU - Kimia, Benjamin

PY - 2011

Y1 - 2011

N2 - Object recognition strategies are increasingly based on regional descriptors such as SIFT or HOG at a sparse set of points or on a dense grid of points. Despite their success on databases such as PASCAL and CALTECH, the capability of such a representation in capturing the essential object content of the image is not well-understood: How large is the equivalence class of images sharing the same HOG descriptor? Are all these images from the same object category, and if not, do the non-category images resemble random images which cannot generically arise from imaged scenes? How frequently do images from two categories share the same HOG-based representation? These questions are increasingly more relevant as very large databases such as ImageNet and LabelMe are being developed where the current object recognition strategies show limited success. We examine these questions by introducing the metameric class of moments of HOG which allows for a target image to be morphed into an impostor image sharing the HOG representation of a source image while retaining the initial visual appearance. We report that two distinct images can be made to share the same HOG representation when the overlap between HOG patches is minimal, and the success of this method falls with increasing overlap. This paper is therefore a step in the direction of developing a sampling theorem for representing images by HOG features.

AB - Object recognition strategies are increasingly based on regional descriptors such as SIFT or HOG at a sparse set of points or on a dense grid of points. Despite their success on databases such as PASCAL and CALTECH, the capability of such a representation in capturing the essential object content of the image is not well-understood: How large is the equivalence class of images sharing the same HOG descriptor? Are all these images from the same object category, and if not, do the non-category images resemble random images which cannot generically arise from imaged scenes? How frequently do images from two categories share the same HOG-based representation? These questions are increasingly more relevant as very large databases such as ImageNet and LabelMe are being developed where the current object recognition strategies show limited success. We examine these questions by introducing the metameric class of moments of HOG which allows for a target image to be morphed into an impostor image sharing the HOG representation of a source image while retaining the initial visual appearance. We report that two distinct images can be made to share the same HOG representation when the overlap between HOG patches is minimal, and the success of this method falls with increasing overlap. This paper is therefore a step in the direction of developing a sampling theorem for representing images by HOG features.

KW - image recognition

KW - image representation

KW - very large databases

KW - visual databases

KW - CALTECH databases

KW - HOG descriptor

KW - ImageNet

KW - LabelMe

KW - PASCAL databases

KW - SIFT

KW - dense grid of points

KW - impostor image sharing

KW - initial visual appearance

KW - metameric class

KW - noncategory images

KW - object content

KW - object recognition strategies

KW - random images

KW - regional descriptors

KW - representation capabilities

KW - source image

KW - sparse set of points

KW - Educational institutions

KW - Equations

KW - Histograms

KW - Mathematical model

KW - Object recognition

KW - Upper bound

KW - Vectors

U2 - 10.1109/ICCVW.2011.6130416

DO - 10.1109/ICCVW.2011.6130416

M3 - Article in proceedings

SN - 978-1-4673-0062-9

SP - 1410

EP - 1417

BT - 2011 IEEE International Conference on Computer Vision Workshops (ICCV Workshops)

PB - IEEE

T2 - International Conference of Computer Vision (ICCV)

Y2 - 8 November 2011 through 13 November 2011

ER -