A Comparison of Deep Learning Methods for Airborne Lidar Point Clouds Classification

The success achieved by deep learning techniques in image labeling has triggered a growing interest in applying deep learning for three-dimensional point cloud classification. To provide better insights into different deep learning architectures and their applications to ALS point cloud classificati...

Full description

Saved in:
Bibliographic Details
Published in:IEEE journal of selected topics in applied earth observations and remote sensing 2021, Vol.14, p.6467-6486
Main Authors: Li, Nan, Kahler, Olaf, Pfeifer, Norbert
Format: Article
Language:English
Subjects:
ALS point clouds
Classification
comparison
Computation
Computational modeling
Computer architecture
Computing time
Convolution
Deep learning
Image classification
Learning
Lidar
Machine learning
Parameter sensitivity
Parameters
Solid modeling
Three-dimensional displays
Training
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The success achieved by deep learning techniques in image labeling has triggered a growing interest in applying deep learning for three-dimensional point cloud classification. To provide better insights into different deep learning architectures and their applications to ALS point cloud classification, this article presents a comprehensive comparison among three state-of-the-art deep learning networks: PointNet++, SparseCNN, and KPConv, on two different ALS datasets. The performances of these three deep learning networks are compared w.r.t. classification accuracy, computation time, generalization ability as well as the sensitivity to the choices of hyper-parameters. Overall, we observed that PointNet++, SparseCNN, and KPConv all outperform Random Forest on the classification results. Moreover, SparseCNN leads to a slightly better classification result compared to PointNet++ and KPConv, while requiring less computation time and memory. At the same time, it shows a better ability to generalize and is less impacted by the different choices of hyper-parameters.
ISSN:1939-1404
2151-1535
DOI:10.1109/JSTARS.2021.3091389