Deep learning-based segmentation of skid trails from airborne LiDAR data
TrailScan is a QGIS plugin for automatic mapping of skid trails from Airborne Laser Scanning (ALS) data in forest environments. Input: ALS point clouds (.las / .laz) from aircraft, helicopters, or UAVs. Workflow: 1. Preprocessing – Convert the point cloud into a normalized multi-band raster (DTM, CHM, MRM, VDI). 2. Inference – Apply the pretrained TrailScan deep learning model to generate a probability map ("Trailmap"). The trained TrailScan model is available at https://doi.org/10.25625/GEIP6T. Output: A georeferenced raster ("Trailmap" with values 0–1) indicating the likelihood of skid trail presence, ready for further GIS analysis.
Plugin Tags