Overview
3D perception starts with getting a point cloud on screen and understanding its density. This toolkit compares three libraries on the same scan, a patchwork chair from the IITH LiDAR ground dataset.
The voxelisation script makes the trade-off between detail and size tangible: smaller voxels keep more structure, larger ones leave fewer cubes to process.
How it works
Open3D
Reads the PLY file, checks that points were actually loaded and opens an interactive viewer.
Voxelise
Builds a voxel grid with an adjustable voxel size and reports how many voxels it produced.
PyVista
Renders the points as spheres with a configurable point size.
PyntCloud
Loads the cloud as a pandas-backed structure, pulls the XYZ coordinates into NumPy and hands them to PyVista.