Voxelize point clouds

Tool to voxelize annotated point clouds.

Dependencies

• Eigen >= 3.2
• boost >= 1.54
• QT >= 5.2

Build

On Ubuntu 22.04/20.04, most of the dependencies can be installed from the package manager:

sudo apt install build-essential cmake libeigen3-dev libboost-all-dev qtbase5-dev libglew-dev

Use the usual cmake out-of-source build process:

mkdir build && cd build
cmake ..

Now the project root directory should contain a bin directory containing the voxelizer.

Usage

In the bin directory, just run ./voxelizer to start the voxelizer.

In the settings.cfg files you can change the followings options:

max scans: 500    # number of scans to load for a tile. (should be maybe 1000), but this currently very memory consuming.
min range: 2.5    # minimum distance of points to consider.
max range: 50.0   # maximum distance of points in the point cloud.
ignore: 0,250,251,252,253,254  # label ids of labels that should be ignored when building a voxelgrid.

To generate the data by iterating over a sequence directory, call ./gen_data in the bin directory.

Folder structure

When loading a dataset, the data must be organized as follows:

point cloud folder
├── velodyne/             -- directory containing ".bin" files with Velodyne point clouds.   
├── labels/               -- label directory, will be generated if not present.    
├── calib.txt             -- calibration of velodyne vs. camera. needed for projection of point cloud into camera.  
└── poses.txt             -- file containing the poses of every scan.