We used a Point Cloud Transformer (PCT) network as the backbone and tested different Self-Attention modules for 3D Hypercloud data in Geological application. Find our paper here.
Download Tinto data from RODARE.
First prepare KNN points:
python find_kNN_for_training.py
python find_kNN_for_testing.py
Then train and test all models:
Train_and_Test.sh
Cite the paper here:
A. Rizaldy, A. J. Afifi, P. Ghamisi and R. Gloaguen, "Transformer-Based Models for Hyperspectral Point Clouds Segmentation," 2023 13th Workshop on Hyperspectral Imaging and Signal Processing: Evolution in Remote Sensing (WHISPERS), Athens, Greece, 2023, pp. 1-5, doi: 10.1109/WHISPERS61460.2023.10431346. keywords: {Point cloud compression;Geology;Benchmark testing;Signal processing;Transformers;Noise measurement;Hyperspectral imaging;Point cloud;Hyperspectral;Transformer;Attention;Classification},