Segmentation of LiDAR Point Cloud Data in Urban Areas using Adaptive Neighborhood

Dataset

• Vaihingen Dataset
• Toronto-3D Dataset

Vaihingen Classes

• Powerline (label 0)
• Low vegetation (label 1)
• Impervious surfaces (label 2)
• Car (label 3)
• Fence (label 4)
• Roof (label 5)
• Facade (label 6)
• Shrub (label 7)
• Tree (label 8)

Toronto Classes

• Unclassified Point (label 0)
• Ground (label 1)
• Road marking (label 2)
• Tree (label 3)
• Building (label 4)
• Power line (label 5)
• Electrical Pole (label 6)
• Car (label 7)
• Fence (label 8)

Dependencies Library

• Scikit Learn
• Open3D
• Numpy
• Plotly
• Seaborn
• Tqdm

Figure

1. Vaihingen

2. Toronto

Notes

Paper Link - https://doi.org/10.1371/journal.pone.0307138 For Citation(Bibtex)

@article{chakraborty2024segmentation, title={Segmentation of LiDAR point cloud data in urban areas using adaptive neighborhood selection technique}, author={Chakraborty, Debobrata and Dey, Emon Kumar}, journal={PloS one}, volume={19}, number={7}, pages={e0307138}, year={2024}, publisher={Public Library of Science San Francisco, CA USA} }

For the complete code and additional analysis suggestions, please reach out to the author at msse1729@iit.du.ac.bd.