-
Install Python 3.6, 3.7 or 3.8
- We have tested the pipeline with Python 3.8
- The library versions in
SatelliteSurfaceReconstruction/requirements.txt
might differ for Python 3.6 and 3.7
-
Install the freeimage plugin for imageio
- Activate your conda environment (e.g.
conda activate <ssr_conda_environment_name>
) - Start an interactive python session (i.e. by typing
python
) - Download and install the freeimage plugin
>>> import imageio
>>> imageio.plugins.freeimage.download()
- Activate your conda environment (e.g.
-
Install VisSat
-
Install ColmapForVisSat or ColmapForVisSatPatched
- Option 1: ColmapForVisSatPatched (RECOMMENDED)
- Follow the install instructions
- sudo apt-get install libmetis-dev (Latest tested version: 5.1.0.dfsg-5)
- To avoid that incorrect libraries are detected by CMake, consider to disable the active anaconda/miniconda environment with
conda deactivate
- Follow the official install instructions of Colmap (for linux)
- Latest tested version: commit 31df46c
- Follow the install instructions
- Option 2: ColmapForVisSat
- Follow the install instructions
- Make sure you have installed a compatible cuda version (e.g.
Cuda 10.0
)Cuda 10.0
can be installed following this link.
- To avoid that incorrect libraries are detected by CMake, consider to disable the active anaconda/miniconda environment with
conda deactivate
- Clone the repository
git clone https://github.com/Kai-46/ColmapForVisSat.git
- Execute
ubuntu1804_install_dependencies.sh
- Execute
ubuntu1804_install_colmap.sh
- Make sure you have installed a compatible cuda version (e.g.
- Follow the install instructions
- Option 1: ColmapForVisSatPatched (RECOMMENDED)
-
Follow the install instructions of VisSatSatelliteStereo
- Latest tested version of VisSatSatelliteStereo: commit c6cb1b4
- Install GDAL
- Option 1: (Recommended)
conda install gdal libgdal
- Activate (or re-activate) your conda environment and make sure gdal environment variables are correctly configured
echo $PROJ_LIB
- RESTART Pycharm to parse the current environment variables!
- Activate (or re-activate) your conda environment and make sure gdal environment variables are correctly configured
- Tested versions: GDAL 3.3.2 with
Python 3.8
- Option 2:
- Follow the original GDAL install instructions here
- Tested versions: GDAL 2.4.2 with
Python 3.6
, released 2019/06/28
- Option 1: (Recommended)
- Clone the repository
git clone https://github.com/Kai-46/VisSatSatelliteStereo
pip install -r requirements.txt
- For Python 3.8 use our
SatelliteSurfaceReconstruction/requirements.txt
instead ofVisSatSatelliteStereo/requirements.txt
- For Python 3.8 use our
- Make sure to ADJUST the
--SiftExtraction.num_threads 32
parameter in colmap_sfm_commands.py- See this issue
-
-
Install at least one of the following MVS & surface reconstruction libraries
- ColmapForVisSat / Colmap
- See installation instructions above
- MVE with FSSR
- Follow the installation instructions of MVE
- Note: FSSR is part of the MVE library
- Latest tested version: commit 354a652
- GDMR (requires MVE)
- Follow the installation instructions of MVE
- Follow the installation instructions of GDMR
- Latest tested version: 0.2.0
- OpenMVS
- Follow the installation instructions of OpenMVS
- Latest tested version: v2.1.0
- ColmapForVisSat / Colmap
-
Install MVS-Texturing (requires MVE)
- Follow the installation instructions of MVS-Texturing
- Latest tested version: commit 1a8603e
-
Install Meshlab
- Make sure to install Meshlab-2020.09 or older
- For example download MeshLabServer2020.09-linux.AppImage
- Snap installations (e.g.
MeshLab2020.09-linux.snap
) might not work properly because of snap confinement
- Newer versions do not include the required meshlabserver anymore
- For a list of releases see here
- Make sure to install Meshlab-2020.09 or older
-
Make VisSatSatelliteStereo and
SatelliteSurfaceReconstruction
available to your python interpreter-
Option 1: Add
VisSatSatelliteStereo
andSatelliteSurfaceReconstruction
to the python path- Add
ssr
(in theSatelliteSurfaceReconstruction
directory)export PYTHONPATH="${PYTHONPATH}:/path/to/SatelliteSurfaceReconstruction"
- Add
lib
etc. (in theVisSatSatelliteStereo
directory)export PYTHONPATH="${PYTHONPATH}:/path/to/VisSatSatelliteStereo"
- This allows to run your pipeline (later) with
cd /path/to/SatelliteSurfaceReconstruction/ssr
python run_pipeline.py
(see the sectionRun the SSR Pipeline
below)
- Add
-
Option 2: Use and IDE like Pycharm to manage the project dependencies
- For example in
Pycharm
File
/Open...
/ SelectSatelliteSurfaceReconstruction
File
/Open...
/ SelectVisSatSatelliteStereo
and attach it to the currentSatelliteSurfaceReconstruction
project- Go to
File / Settings / Project: SatelliteSurfaceReconstruction / Project Dependencies
- Select
SatelliteSurfaceReconstruction
- Make sure that
VisSatSatelliteStereo
is selected
- Select
- For example in
-