Deep-learning semantic segmentation of aerial LiDAR point clouds with a 3D SegFormer model (TreeAIBox / NRCan).
Classifies aerial LiDAR point clouds (LAS / LAZ / COPC) into
Ground, Vegetation, Building, Wires / Powerlines and Poles
(standard ASPRS codes) using the UrbanFiltering 3D SegFormer
model from NRCan's TreeAIBox project. Runs on NVIDIA GPU (CUDA),
Apple Silicon (MPS) or CPU; spatial tiling handles files larger
than RAM. A one-click installer sets up an isolated Python venv
on first use, with no impact on QGIS's own Python. The model is
distributed under CC BY-NC 4.0; full documentation, the GPU
driver table and the CLI / Processing usage are on the project
homepage: https://github.com/akharroubi/AerialLidarClassifier
Plugin Tags