Research Group of Prof. Dr. C. Sminchisescu
Mathematical Sciences




CPMC: Constrained Parametric Min-Cuts for Automatic Object Segmentation

Joao Carreira and Cristian Sminchisescu

Version 1.0, released on 28 of September 2010
Last patch on 13th of February 2011




This is the implementation of CPMC used in our VOC2010 segmentation challenge submissions. It is an extention of the algorithm used in our VOC2009 submission, which is described in Constrained Parametric Min-Cuts for Automatic Object Segmentation, CVPR 2010 [1]. Changes over [1] are in reducing computation time and the usage of subframes to better segment small objects. The learning code is not yet provided, but the package includes two pretrained ranking models.

The system has been tested on MATLAB 7.9 and 7.10 on Linux 32bits and 64bits machines with 10 Gb of memory.
For details see the README file in the package below.

Download CPMC: [code].
This package is free for academic use only. You run it at your own risk.



References

Constrained Parametric Min-Cuts for Automatic Object Segmentation
J. Carreira and C. Sminchisescu
In IEEE International Conference on Computer Vision and Pattern Recognition, June 2010. See also the extended journal version.

If you use our system, please cite reference [1], and the following website reference for the particular release:

@misc{cpmc-release1,
author = "J. Carreira and C. Sminchisescu",
title = "Constrained Parametric Min-Cuts for Automatic Object Segmentation, Release 1",
howpublished = "http://sminchisescu.ins.uni-bonn.de/code/cpmc/"}



Related References


For generating full image segmentations using CPMC segments read:

Image Segmentation by Figure-Ground Composition into Maximal Cliques,
A. Ion, J. Carreira and C. Sminchisescu.
In International Conference on Computer Vision, November 2011.

For an earlier version please see:
Image Segmentation by Discounted Cumulative Ranking on Maximal Cliques,
J. Carreira, A. Ion, and C. Sminchisescu.
Technical Report 06-2010 (arXiv:1009.4823), Computer Vision and Machine Learning Group, Institute for Numerical Simulation, University of Bonn, June 2010.
Also available at http://arxiv.org/abs/1009.4823.


For doing recognition using CPMC segments read:

Object Recognition as Ranking Holistic Figure-Ground Hypotheses
F. Li, J. Carreira, and C. Sminchisescu.
In IEEE International Conference on Computer Vision and Pattern Recognition, June 2010. See also the extended journal version.