The code implement the method described in the 3DGP paper published in CVPR13 (see README for full title).
Matlab C M C++ Objective-C Java Other
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
3rdParty
common
evaluation
experimental
figures
geometry
ijcv_tests
image
inference
model
objmodel
potentials
preprocess
scripts
train
visualize
COPYING
README
addPaths.m
addVarshaPaths.m
demo_test.m

README

Information
===========

Project webpage: http://www.eecs.umich.edu/vision/3DGP/

This package is an implementation of the indoor scene understanding method described in 
(W. Choi, Y. -W. Chao, C. Pantofaru and S. Savarese "Understanding Indoor Scenes using 3D Geometric Phrases", CVPR 2013). 

For questions concerning the code please contact Wongun Choi at <wgchoi AT umich DOT edu>.

The system is implemented in Matlab. 

The code is tested on Linux (Ubuntu 10.10) with a Matlab version R2011b. 

Basic Usage
===========

Currently, only testing part of the method is released. Training algorithm will be made available soon as well. 

To test:
--------

The package include a script "demo_test.m" that replicates the results reported in our CVPR paper. The code does, 
1. downloading dataset
2. preprocessing images - object detection (Felzenszwalb et al, PAMI 2010), scene classification (Lazebnik et al, CVPR 2006) and indoor layout estimation (Hedau et al, ICCV 2009)
3. running 3DGP model
4. evaluation (detection, scene classification, layout estimation)
5. visualization of examplar results


How to Cite
===========
When citing our system, please cite our following CVPR 13 publication:

@InProceedings{choi_cvpr13,
  author       = "W. Choi and Y. -W. Chao and C. Pantofaru and S. Savarese",
  title        = "Understanding Indoor Scenes Using 3D Geometric Phrases",
  booktitle    = "CVPR",
  year         = "2013",
}