Cross-depiction problem: Recognition and Synthesis of Photographs and Artwork

Peter Hall, Hongping Cai, Qi Wu, Tadeo Corradi

Research output: Contribution to journalArticle

Abstract

The cross-depiction is the recognition -- and synthesis -- of objects whether they are photographed, painted, drawn, etc. It is a significant yet under-researched problem. Emulating the remarkable human ability to recognise and depict objects in an astonishingly wide variety of depictive forms is likely to advance both the foundations and the applications of Computer Vision.

In this paper we motivate the cross-depiction problem, explain why it is difficult, discuss some current approaches. Our main conclusions are (i) appearance-based recognition systems tend to be over-fitted to one depiction, (ii) models that explicitly encode spatial relations between parts are more robust, and (iii) recognition and non-photorealistic synthesis are related tasks.
LanguageEnglish
Pages91-103
JournalComputational Visual Media
Volume1
Issue number2
DOIs
StatusPublished - 2015

Fingerprint

Computer vision

Cite this

Cross-depiction problem: Recognition and Synthesis of Photographs and Artwork. / Hall, Peter; Cai, Hongping; Wu, Qi; Corradi, Tadeo.

In: Computational Visual Media, Vol. 1, No. 2, 2015, p. 91-103.

Research output: Contribution to journalArticle

Hall, Peter ; Cai, Hongping ; Wu, Qi ; Corradi, Tadeo. / Cross-depiction problem: Recognition and Synthesis of Photographs and Artwork. In: Computational Visual Media. 2015 ; Vol. 1, No. 2. pp. 91-103
@article{aa1b4f8522c64e529baef4db34286ee4,
title = "Cross-depiction problem: Recognition and Synthesis of Photographs and Artwork",
abstract = "The cross-depiction is the recognition -- and synthesis -- of objects whether they are photographed, painted, drawn, etc. It is a significant yet under-researched problem. Emulating the remarkable human ability to recognise and depict objects in an astonishingly wide variety of depictive forms is likely to advance both the foundations and the applications of Computer Vision.In this paper we motivate the cross-depiction problem, explain why it is difficult, discuss some current approaches. Our main conclusions are (i) appearance-based recognition systems tend to be over-fitted to one depiction, (ii) models that explicitly encode spatial relations between parts are more robust, and (iii) recognition and non-photorealistic synthesis are related tasks.",
author = "Peter Hall and Hongping Cai and Qi Wu and Tadeo Corradi",
year = "2015",
doi = "10.1007/s41095-015-0017-1",
language = "English",
volume = "1",
pages = "91--103",
journal = "Computational Visual Media",
issn = "2096-0433",
number = "2",

}

TY - JOUR

T1 - Cross-depiction problem: Recognition and Synthesis of Photographs and Artwork

AU - Hall,Peter

AU - Cai,Hongping

AU - Wu,Qi

AU - Corradi,Tadeo

PY - 2015

Y1 - 2015

N2 - The cross-depiction is the recognition -- and synthesis -- of objects whether they are photographed, painted, drawn, etc. It is a significant yet under-researched problem. Emulating the remarkable human ability to recognise and depict objects in an astonishingly wide variety of depictive forms is likely to advance both the foundations and the applications of Computer Vision.In this paper we motivate the cross-depiction problem, explain why it is difficult, discuss some current approaches. Our main conclusions are (i) appearance-based recognition systems tend to be over-fitted to one depiction, (ii) models that explicitly encode spatial relations between parts are more robust, and (iii) recognition and non-photorealistic synthesis are related tasks.

AB - The cross-depiction is the recognition -- and synthesis -- of objects whether they are photographed, painted, drawn, etc. It is a significant yet under-researched problem. Emulating the remarkable human ability to recognise and depict objects in an astonishingly wide variety of depictive forms is likely to advance both the foundations and the applications of Computer Vision.In this paper we motivate the cross-depiction problem, explain why it is difficult, discuss some current approaches. Our main conclusions are (i) appearance-based recognition systems tend to be over-fitted to one depiction, (ii) models that explicitly encode spatial relations between parts are more robust, and (iii) recognition and non-photorealistic synthesis are related tasks.

UR - http://dx.doi.org/10.1007/s41095-015-0017-1

UR - http://dx.doi.org/10.1007/s41095-015-0017-1 Official URL

U2 - 10.1007/s41095-015-0017-1

DO - 10.1007/s41095-015-0017-1

M3 - Article

VL - 1

SP - 91

EP - 103

JO - Computational Visual Media

T2 - Computational Visual Media

JF - Computational Visual Media

SN - 2096-0433

IS - 2

ER -