Sampling bedrooms

Luca Del Pero, Jinyan Guan, Ernesto Brau, Joseph Schlecht, Jacobus J Barnard

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

37 Citations (Scopus)

Abstract

We propose a top down approach for understanding indoor scenes such as bedrooms and living rooms. These environments typically have the Manhattan world property that many surfaces are parallel to three principle ones. Further, the 3D geometry of the room and objects within it can largely be approximated by non overlapping simple structures such as single blocks (e.g. the room boundary), thin blocks (e.g. picture frames), and objects that are well modeled by single blocks (e.g. simple beds). We separately model the 3D geometry, the imaging process (camera parameters), and edge likelihood, to provide a generative statistical model for image data. We fit this model using data driven MCMC sampling. We combine reversible jump Metropolis Hastings samples for discrete changes in the model such as the number of blocks, and stochastic dynamics to estimate continuous parameter values in a particular parameter space that includes block positions, block sizes, and camera parameters. We tested our approach on two datasets using room box pixel orientation. Despite using only bounding box geometry and, in particular, not training on appearance, our method achieves results approaching those of others. We also introduce a new evaluation method for this domain based on ground truth camera parameters, which we found to be more sensitive to the task of understanding scene geometry.

Original languageEnglish (US)
Title of host publicationProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
Pages2009-2016
Number of pages8
DOIs
StatePublished - 2011
Event2011 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2011 - Colorado Springs, CO, United States
Duration: Jun 20 2011Jun 25 2011

Other

Other2011 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2011
CountryUnited States
CityColorado Springs, CO
Period6/20/116/25/11

Fingerprint

Sampling
Geometry
Cameras
Pixels
Imaging techniques
Statistical Models

ASJC Scopus subject areas

  • Software
  • Computer Vision and Pattern Recognition

Cite this

Del Pero, L., Guan, J., Brau, E., Schlecht, J., & Barnard, J. J. (2011). Sampling bedrooms. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition (pp. 2009-2016). [5995737] https://doi.org/10.1109/CVPR.2011.5995737

Sampling bedrooms. / Del Pero, Luca; Guan, Jinyan; Brau, Ernesto; Schlecht, Joseph; Barnard, Jacobus J.

Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. 2011. p. 2009-2016 5995737.

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

Del Pero, L, Guan, J, Brau, E, Schlecht, J & Barnard, JJ 2011, Sampling bedrooms. in Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition., 5995737, pp. 2009-2016, 2011 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2011, Colorado Springs, CO, United States, 6/20/11. https://doi.org/10.1109/CVPR.2011.5995737
Del Pero L, Guan J, Brau E, Schlecht J, Barnard JJ. Sampling bedrooms. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. 2011. p. 2009-2016. 5995737 https://doi.org/10.1109/CVPR.2011.5995737
Del Pero, Luca ; Guan, Jinyan ; Brau, Ernesto ; Schlecht, Joseph ; Barnard, Jacobus J. / Sampling bedrooms. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. 2011. pp. 2009-2016
@inproceedings{37111a20c0544abea6b8cc21a9bdc437,
title = "Sampling bedrooms",
abstract = "We propose a top down approach for understanding indoor scenes such as bedrooms and living rooms. These environments typically have the Manhattan world property that many surfaces are parallel to three principle ones. Further, the 3D geometry of the room and objects within it can largely be approximated by non overlapping simple structures such as single blocks (e.g. the room boundary), thin blocks (e.g. picture frames), and objects that are well modeled by single blocks (e.g. simple beds). We separately model the 3D geometry, the imaging process (camera parameters), and edge likelihood, to provide a generative statistical model for image data. We fit this model using data driven MCMC sampling. We combine reversible jump Metropolis Hastings samples for discrete changes in the model such as the number of blocks, and stochastic dynamics to estimate continuous parameter values in a particular parameter space that includes block positions, block sizes, and camera parameters. We tested our approach on two datasets using room box pixel orientation. Despite using only bounding box geometry and, in particular, not training on appearance, our method achieves results approaching those of others. We also introduce a new evaluation method for this domain based on ground truth camera parameters, which we found to be more sensitive to the task of understanding scene geometry.",
author = "{Del Pero}, Luca and Jinyan Guan and Ernesto Brau and Joseph Schlecht and Barnard, {Jacobus J}",
year = "2011",
doi = "10.1109/CVPR.2011.5995737",
language = "English (US)",
isbn = "9781457703942",
pages = "2009--2016",
booktitle = "Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition",

}

TY - GEN

T1 - Sampling bedrooms

AU - Del Pero, Luca

AU - Guan, Jinyan

AU - Brau, Ernesto

AU - Schlecht, Joseph

AU - Barnard, Jacobus J

PY - 2011

Y1 - 2011

N2 - We propose a top down approach for understanding indoor scenes such as bedrooms and living rooms. These environments typically have the Manhattan world property that many surfaces are parallel to three principle ones. Further, the 3D geometry of the room and objects within it can largely be approximated by non overlapping simple structures such as single blocks (e.g. the room boundary), thin blocks (e.g. picture frames), and objects that are well modeled by single blocks (e.g. simple beds). We separately model the 3D geometry, the imaging process (camera parameters), and edge likelihood, to provide a generative statistical model for image data. We fit this model using data driven MCMC sampling. We combine reversible jump Metropolis Hastings samples for discrete changes in the model such as the number of blocks, and stochastic dynamics to estimate continuous parameter values in a particular parameter space that includes block positions, block sizes, and camera parameters. We tested our approach on two datasets using room box pixel orientation. Despite using only bounding box geometry and, in particular, not training on appearance, our method achieves results approaching those of others. We also introduce a new evaluation method for this domain based on ground truth camera parameters, which we found to be more sensitive to the task of understanding scene geometry.

AB - We propose a top down approach for understanding indoor scenes such as bedrooms and living rooms. These environments typically have the Manhattan world property that many surfaces are parallel to three principle ones. Further, the 3D geometry of the room and objects within it can largely be approximated by non overlapping simple structures such as single blocks (e.g. the room boundary), thin blocks (e.g. picture frames), and objects that are well modeled by single blocks (e.g. simple beds). We separately model the 3D geometry, the imaging process (camera parameters), and edge likelihood, to provide a generative statistical model for image data. We fit this model using data driven MCMC sampling. We combine reversible jump Metropolis Hastings samples for discrete changes in the model such as the number of blocks, and stochastic dynamics to estimate continuous parameter values in a particular parameter space that includes block positions, block sizes, and camera parameters. We tested our approach on two datasets using room box pixel orientation. Despite using only bounding box geometry and, in particular, not training on appearance, our method achieves results approaching those of others. We also introduce a new evaluation method for this domain based on ground truth camera parameters, which we found to be more sensitive to the task of understanding scene geometry.

UR - http://www.scopus.com/inward/record.url?scp=80052882477&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80052882477&partnerID=8YFLogxK

U2 - 10.1109/CVPR.2011.5995737

DO - 10.1109/CVPR.2011.5995737

M3 - Conference contribution

SN - 9781457703942

SP - 2009

EP - 2016

BT - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

ER -