Skip to content

Computer vision package for algorithms built on convex pose estimation

Notifications You must be signed in to change notification settings

KelseyHorowitz/Cvx_Pose

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

43 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Cvx_Pose

Computer vision package for algorithms built on convex pose estimation. Wrappers for PCL and Matlab are included. Compiled and tested with gcc.

Authored by Matanya Horowitz.
Based on the work by Matanya Horowitz, Nikolai Matni under the guidance of Joel Burdick
http://matanyahorowitz.com/
http://www.cds.caltech.edu/~nmatni/

Academic papers:

  • Convex Relaxations of SE (2) and SE (3) for Visual Pose Estimation. http://arxiv.org/abs/1401.3700
  • A Convex Approach to Consensus on SO(n) (Under review)
  • Convex Iterative Closest Point (cICP) (In preparation for CVPR 2015)
  • Analytic Solutions to Parallelized Pose Estimation with Outlier Rejection (In preparation for CVPR 2015)
  • Convex Model Predictive Control for Vehicular Systems (In preparation)

The CVX_CV package currently contains solutions based on the convex relaxation of SO(n) for n=2,3 for the following problems:

  • ICP
  • Pose Estimation

Future versions will include support for Structure from Motion (SfM), Simultaneous Localization and Mapping (SLAM), and more.

The package currently includes a number of examples (in progress):

  • Head tracking for the cardboard virtual reality headset
  • Pose estimation of juggling balls

Dependencies:

  • PCL 1.7 +
  • Eigen 3.2 +
  • OpenCV 2.4 + - for running some examples based on visual features
  • Mosek or SDPA - for running convex estimation, include point-to-plane ICP

Notes

Installation

Linux (Ubuntu):

  1. First install CMake. Open a command prompt and type sudo apt-get install cmake
  2. Download Eigen library from http://eigen.tuxfamily.org/index.php. Create a folder in the download directory called build_dir, and move into it. Then type cmake .. sudo make install
  3. Install PCL. In the command prompt type sudo add-apt-repository ppa:v-launchpad-jochen-sprickerhof-de/pcl sudo apt-get update sudo apt-get install libpcl-all
  4. Install git: sudo apt-get install git
  5. Install SDPA sudo apt-get install sdpa libsdpa-dev Mac:

Use Homebrew to get the latest version of PCL

Windows:

Once the dependencies have been installed, move into your chosen directory and type git clone git@github.com:matanyahorowitz/Cvx_Pose.git

Compiling

Navigate to build/ and then type cmake .. make

About

Computer vision package for algorithms built on convex pose estimation

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C 79.9%
  • MATLAB 17.7%
  • C++ 2.2%
  • Other 0.2%