[Point cloud] Investigate on classification libraries

[allyoucanmap]

We should investigate in open source solution to classify point cloud data in las format.

Some links that could be of interest:

• https://github.com/NRCan/geo-deep-learning
• https://github.com/yangyanli/PointCNN
• https://github.com/pyg-team/pytorch_geometric
• ...

The output of this investigation should be a report and if possible a proof of concept with approaches and code snippets.

We could start analyzing some data available on open portals:

• Genova data: https://mappe.comune.genova.it/MapStore2/#/viewer/1000003072

  • eg. .laz tile url: https://mappe.comune.genova.it/gis/rilievo/LAS/LAS3710.las (already classified)

• Belgium data https://remotesensing.vlaanderen.be/apps/openlidar/

  • eg. .laz tile url https://remotesensing.vlaanderen.be/download/openlidar/LiDAR_DHMV_2_V2/P4/ATL12415/LiDAR_DHMV_2_P4_ATL12415_ES_20131216_15058_4/Tiles/LiDAR_DHMV_2_P4_ATL12415_ES_20131216_15058_4_148000_210500.laz