Skip to content

Segmentation Cascades with Localized Priors and Efficient Learning

License

Notifications You must be signed in to change notification settings

djweiss/scalpel

Repository files navigation

SCALPEL: Segmentation Cascades with Localized Priors and Efficient Learning

This is the source code and data files needed to reproduce the experiments from the paper,

  • SCALPEL: Segmentation Cascades with Localized Priors and Efficient Learning. David Weiss and Ben Taskar. CVPR2013.

It includes all third-party tools and dependencies. It has only been tested on 64-bit Unix MATLAB R2012a systems.

Rather than include even more third-party tools (DPM, Objectness, etc.) to generate the bounding boxes used in the experiments of the paper, we include bounding boxes used as input to SCALPEL for every image in the PASCAL VOC2010 validation set. By running the included code with these boxes, you should reproduce the results of the experiments in the paper.

Instructions (important)

  • Download the code and extract to a directory.
  • If cloning from git, make sure to run "git submodule init" and "git submodule update" to get the latest dependencies.
  • If downloading the zip archive, make sure to also download and manually extract the matlab-utils submodule from https://github.com/djweiss/matlab-utils/archive/master.zip. Note: You need to rename the extracted contents to matlab-utils for this to work.
  • Go to that directory and start MATLAB.
  • Open scalpel_demo.m and follow the instructions there.

About

Segmentation Cascades with Localized Priors and Efficient Learning

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published