Welcome in repository focused on processing AR devices recordings.
That main goal of this project is to implement all the research codes in single place that allow easy setup, easy buils, and remove the time that is required to reinstall all the repositories, writing the format conversion, and wrappers for deployment. The overview of the codes will be here (TODO), however, you can simply run the GUI and load the pipeline that are already done, e.g., localization, dense and sparse mapping, downloading of the AR device recordings, dense point clouds alignment, etc..
HOW TO INSTALL ME
1) Requirements (install in advance)
2) Clone this repository and its submodules
3) Create or copy the Docker/Singularity containers (copy the containers if available, otherwise use code in init.sh) 4) Create the conda environment (use code in init.sh) 5) Compile the C++ codes (use code in init.sh) 6) Download pre-trained weights for NetVLAD (use code in init.sh)
sh ./init.sh
(building the containes may take few hours)7) Activate conda enviroment
conda activate meshroom
F1 -> Python:Select Interpreter -> meshroom
8) If debuging in VS Code
launch.json
, for example:
{
"version": "0.2.0",
"configurations": [
{
"name": "Meshroom GUI",
"type": "python",
"request": "launch",
"program": "<path to hololens_mapper>/third_party/meshroom/meshroom/ui",
"console": "integratedTerminal",
"env": {
"PYTHONPATH": "<path to hololens_mapper>/third_party/meshroom;<path to hololens_mapper>"
}
},
{
"name": "Meshroom batch",
"type": "python",
"request": "launch",
"program": "<path to hololens_mapper>/third_party/meshroom/bin/meshroom_compute",
"console": "integratedTerminal",
"env": {
"PYTHONPATH":
"<path to hololens_mapper>/third_party/meshroom/meshroom;<path to hololens_mapper>/third_party/meshroom;<path to hololens_mapper>"
},
"args": [
"<path to hololens_mapper>/pipelines/<your pipeline name.mg>",
"--cache",
"<your output cache folder>/<evaluation folder>"
]
}
]
}
export PYTHONPATH="$PYTHONPATH;<path to hololens_mapper>/third_party/meshroom;<path to hololens_mapper>"
python <path to hololens_mapper>/third_party/meshroom/meshroom/ui
HOW TO RUN ME
If you finished previous section, you can create, build, modify, and test your pipelines in Meshroom GUI.
<path to hololens_mapper>/pipelines
tmp_<your private pipeline name>.mg
HOW TO CONTRIBUTE
If you would like to add your nodes:
<path to hololens_mapper>/third_party/meshroom/meshroom/nodes/Alignment/DensePonitcloudsConcatenator.py
<path to hololens_mapper>/third_party/meshroom/meshroom/nodes
(please follow reasonable naming)To add your own source codes:
<path to hololens_mapper>/src
(please, follow reanable naming)sh ./init.sh