This repo contains the altered version of the OpenMVG pipeline, and the code to generate the swig wrapper.

1) Compiling OpenMVG and its wrap
  I reccomand a local installation in a custom directory; I have changed the cmake to install the versions of ceres, osi and other libraries 
  used by the original authors of the pipeline.
  
  a) mkdir build && cd build
  b) cmake -DCMAKE_INSTALL_PREFIX:STRING="$some_path_on_your_machine/my_bin" -DCMAKE_BUILD_TYPE=RELEASE ../src/
  c) cmake --build . --target install
  d) Add the my_build directory to your PATH and LD_PATH
     export PATH=$some_path_on_your_machine/my_bin/bin:$PATH
     export LD_LIBRARY_PATH=$some_path_on_your_machine/my_bin/lib/:$LD_LIBRARY_PATH
     Install the cecessary libraries via apt-get install python-dev
     Force library link updates via sudo ldconfig
  e) cd .. && mkdir build_sfm && cd build_sfm
  f) cmake ../source_sfm  -DPYTHON_INCLUDE_DIR=$(python3 -c "from distutils.sysconfig import get_python_inc; print(get_python_inc())") /
  -DPYTHON_LIBRARY=$(python3 -c "import distutils.sysconfig as sysconfig; print(sysconfig.get_config_var('LIBDIR'))"). && make
  g) You now have the wrapper files, just copy sfm.py and _sfm.so to the pycode directory.
 2) Running the pipeline
  The pycode directory contains everything you need to run the simplified version of the pipeline:
  - sfm_class.py contains the BA code and all the functions used in the pipeline
  - depg.py is the file to launch. Sorry if it is messy, I have added some comment lines to make it usable
  - dataset contains the data for the statue dataset: images, ground truth data and initial estimates before BA.