Back to Search
Start Over
Performance, effectiveness and computational efficiency of powerline extraction methods for quantifying ecosystem structure from light detection and ranging
- Publication Year :
- 2023
- Publisher :
- Zenodo, 2023.
-
Abstract
- This repository contains the input data, output results, and processing code for the manuscript entitled "effectiveness and efficiency of powerline extraction methods for quantifying ecosystem structure from light detection and ranging". The raw point clouds are the Dutch AHN3 data at ten study areas in the Netherlands, and the hand-labeled point clouds are the points manually labeled into six categories: vegetation (1), ground (2), buildings (6), water (9), powerline (14), and others (26) (e.g. bridges, cars). The hand-labeled point clouds are used as ground truth for accuracy assessment. There are 25 LiDAR metrics (GeoTIFF layers at 10 m resolution) calculated based on raw point clouds, ground truth, and three powerline extraction methods (i.e. deep learning, hybrid, and eigenvalue methods). The list of the 25 metrics and their ecological meaning can be found in our previous publications (https://doi.org/10.1016/j.ecoinf.2022.101836, https://doi.org/10.1016/j.dib.2022.108798). For the deep learning method, we provide a Jupyter Notebook for the model training and prediction, also available at GitHub: https://github.com/ShiYifang/Powerline_extraction. We also provide the R code for the implementation of the eigenvalue method using the lidR package (https://github.com/r-lidar/lidR).
Details
- Database :
- OpenAIRE
- Accession number :
- edsair.doi.dedup.....6ac1878b94f85bcfd9c3403c6373bc5e
- Full Text :
- https://doi.org/10.5281/zenodo.7701809