Divided we fall

Resolving occlusions using causal reasoning

Paul R. Cooper, Lawrence A Birnbaum, Daniel Halabe, Matthew Brand, Peter N. Prokopowicz

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

An image of a scene with occlusions can yield only partial knowledge about disconnected fragments of the scene. If this were the only knowledge available, programs attempting to interpret the scene would have to conclude that the scene fragments would collapse in a jumble. But they won't. We describe a program that exploits commonsense knowledge of naive physics to make sense of scenes with occlusion. Our causal analysis focuses on the static stability of structures: what supports what. Occluded connections in a link-and-junction scene are inferred by determining the stability of each subassembly in the scene, and connecting parts when they are unstable. The causal explanation that is generated reflects a deeper understanding of the scene than mere model matching; it allows the seeing agent to predict what will happen next in the scene, and determine how to interact with it.

Original languageEnglish (US)
Title of host publicationComputer Vision — ECCV 1994 - 3rd European Conference on Computer Vision, Proceedings
PublisherSpringer Verlag
Pages535-540
Number of pages6
Volume800 LNCS
ISBN (Print)9783540579564
StatePublished - Jan 1 1994
Event3rd European Conference on Computer Vision, ECCV 1994 - Stockholm, Sweden
Duration: May 2 1994May 6 1994

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume800 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other3rd European Conference on Computer Vision, ECCV 1994
CountrySweden
CityStockholm
Period5/2/945/6/94

Fingerprint

Occlusion
Reasoning
Fragment
Model Matching
Physics
Unstable
Partial
Predict
Knowledge

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Cooper, P. R., Birnbaum, L. A., Halabe, D., Brand, M., & Prokopowicz, P. N. (1994). Divided we fall: Resolving occlusions using causal reasoning. In Computer Vision — ECCV 1994 - 3rd European Conference on Computer Vision, Proceedings (Vol. 800 LNCS, pp. 535-540). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 800 LNCS). Springer Verlag.
Cooper, Paul R. ; Birnbaum, Lawrence A ; Halabe, Daniel ; Brand, Matthew ; Prokopowicz, Peter N. / Divided we fall : Resolving occlusions using causal reasoning. Computer Vision — ECCV 1994 - 3rd European Conference on Computer Vision, Proceedings. Vol. 800 LNCS Springer Verlag, 1994. pp. 535-540 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{d1a3df985f5040b2a903dcf5322ca40f,
title = "Divided we fall: Resolving occlusions using causal reasoning",
abstract = "An image of a scene with occlusions can yield only partial knowledge about disconnected fragments of the scene. If this were the only knowledge available, programs attempting to interpret the scene would have to conclude that the scene fragments would collapse in a jumble. But they won't. We describe a program that exploits commonsense knowledge of naive physics to make sense of scenes with occlusion. Our causal analysis focuses on the static stability of structures: what supports what. Occluded connections in a link-and-junction scene are inferred by determining the stability of each subassembly in the scene, and connecting parts when they are unstable. The causal explanation that is generated reflects a deeper understanding of the scene than mere model matching; it allows the seeing agent to predict what will happen next in the scene, and determine how to interact with it.",
author = "Cooper, {Paul R.} and Birnbaum, {Lawrence A} and Daniel Halabe and Matthew Brand and Prokopowicz, {Peter N.}",
year = "1994",
month = "1",
day = "1",
language = "English (US)",
isbn = "9783540579564",
volume = "800 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "535--540",
booktitle = "Computer Vision — ECCV 1994 - 3rd European Conference on Computer Vision, Proceedings",
address = "Germany",

}

Cooper, PR, Birnbaum, LA, Halabe, D, Brand, M & Prokopowicz, PN 1994, Divided we fall: Resolving occlusions using causal reasoning. in Computer Vision — ECCV 1994 - 3rd European Conference on Computer Vision, Proceedings. vol. 800 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 800 LNCS, Springer Verlag, pp. 535-540, 3rd European Conference on Computer Vision, ECCV 1994, Stockholm, Sweden, 5/2/94.

Divided we fall : Resolving occlusions using causal reasoning. / Cooper, Paul R.; Birnbaum, Lawrence A; Halabe, Daniel; Brand, Matthew; Prokopowicz, Peter N.

Computer Vision — ECCV 1994 - 3rd European Conference on Computer Vision, Proceedings. Vol. 800 LNCS Springer Verlag, 1994. p. 535-540 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 800 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Divided we fall

T2 - Resolving occlusions using causal reasoning

AU - Cooper, Paul R.

AU - Birnbaum, Lawrence A

AU - Halabe, Daniel

AU - Brand, Matthew

AU - Prokopowicz, Peter N.

PY - 1994/1/1

Y1 - 1994/1/1

N2 - An image of a scene with occlusions can yield only partial knowledge about disconnected fragments of the scene. If this were the only knowledge available, programs attempting to interpret the scene would have to conclude that the scene fragments would collapse in a jumble. But they won't. We describe a program that exploits commonsense knowledge of naive physics to make sense of scenes with occlusion. Our causal analysis focuses on the static stability of structures: what supports what. Occluded connections in a link-and-junction scene are inferred by determining the stability of each subassembly in the scene, and connecting parts when they are unstable. The causal explanation that is generated reflects a deeper understanding of the scene than mere model matching; it allows the seeing agent to predict what will happen next in the scene, and determine how to interact with it.

AB - An image of a scene with occlusions can yield only partial knowledge about disconnected fragments of the scene. If this were the only knowledge available, programs attempting to interpret the scene would have to conclude that the scene fragments would collapse in a jumble. But they won't. We describe a program that exploits commonsense knowledge of naive physics to make sense of scenes with occlusion. Our causal analysis focuses on the static stability of structures: what supports what. Occluded connections in a link-and-junction scene are inferred by determining the stability of each subassembly in the scene, and connecting parts when they are unstable. The causal explanation that is generated reflects a deeper understanding of the scene than mere model matching; it allows the seeing agent to predict what will happen next in the scene, and determine how to interact with it.

UR - http://www.scopus.com/inward/record.url?scp=85027603124&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85027603124&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9783540579564

VL - 800 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 535

EP - 540

BT - Computer Vision — ECCV 1994 - 3rd European Conference on Computer Vision, Proceedings

PB - Springer Verlag

ER -

Cooper PR, Birnbaum LA, Halabe D, Brand M, Prokopowicz PN. Divided we fall: Resolving occlusions using causal reasoning. In Computer Vision — ECCV 1994 - 3rd European Conference on Computer Vision, Proceedings. Vol. 800 LNCS. Springer Verlag. 1994. p. 535-540. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).