GSoC 2012 - SfM: adapt libmv for OpenCV

This is a collaborative effort to port libmv to OpenCV. libmv is a Structure from Motion (SfM) library, which is divided into different modules (image/detector/descriptor/multiview) that allow to resolve part of the SfM process. We are focused on the muliview functionalities, trying to port them to OpenCV.

Quick links:

• Most of the code developed during GSoC 2012: ./src/libmv_opencv/
• GSoC 2012 progress: https://github.com/wg-perception/libmv/wiki/Progress

Members:

• Mentors: Vincent Rabaud, Ilya Lysenkov
• Students: Pablo Speciale, Srimal Jayawardena

Github repository

https://github.com/wg-perception/libmv

Compile

To compile CMake is used: http://www.cmake.org

Prerequisite

• OpenCV 2.4.9: http://opencv.org
• Qt4 (optional): http://qt.nokia.com
• Sphinx (optional): http://sphinx.pocoo.org

How to compile

On Linux,

mkdir bin-opt
cd bin-opt
cmake ../src/
make

An alternative way is using the top-level Makefile, just doing:

make

Compiling for debug,

make debug

Documentation (internal)

make documentation
cd bin-opt/doc/
open 'index.html' file

Documentation content:

• ToDo
• Detailed TODO
• Docs on what has been done
• Guidelines for coding
• Datasets to use
• Progress

Click here to see a preview of our internal documentation.

Note: github does not include recursively '.rst' files, so this preview is incomplete. See important sections like: Detailed TODO and Progress.

Test cases

make
cd bin-opt
./bin/tests/sfm_tests

Note 1: the test case Sfm_simple_pipeline.backyard takes time, so it can be skipped using gtest options:

make
cd bin-opt
./bin/tests/sfm_tests --gtest_filter=-Sfm_simple_pipeline.backyard

Note 2: if you want to see the data with blender, take a look at:

blender ./src/libmv_opencv/testdata/cv/sfm/backyard.blend

Troubleshooting

If you have troubles compiling the project, use the development version of OpenCV: https://github.com/Itseez/opencv

There was a bug in OpenCV, and it was fixed here: https://github.com/Itseez/opencv/commit/1baf5209c4df292922f78e2757321c359f8b28e8

Mail list

Google Group: https://groups.google.com/forum/?hl=en#!forum/libmv_gsoc