Skip to content

README.md

Latest commit

 

History

History
36 lines (29 loc) · 1.54 KB

README.md

File metadata and controls

36 lines (29 loc) · 1.54 KB
Raw

Dataset acquisition

Raw hyperspectral cubes and ground-truth maps are deliberately excluded from this repository. They are third-party research datasets and retain the terms set by their original providers.

The code loader expects the public mirror filenames below. Running scripts/download_data.py calls the same loader used by the experiments and stores files below this directory. Mirror availability does not establish a redistribution licence; users must consult the original provider pages before use.

Dataset Expected files used by the loader
Salinas-A SalinasA_corrected.mat, SalinasA_gt.mat
Indian Pines Indian_pines_corrected.mat, Indian_pines_gt.mat
KSC KSC/KSC_data.mat, KSC/KSC_gt.mat
Pavia University PaviaU.mat, PaviaU_gt.mat
WHU-Hi LongKou WHU-Hi-LongKou/WHU-Hi-LongKou.{bsq,hdr} and _gt.{bsq,hdr}
WHU-Hi HongHu WHU-Hi-HongHu/WHU-Hi-HongHu.{bsq,hdr} and _gt.{bsq,hdr}
WHU-Hi HanChuan WHU-Hi-HanChuan/WHU-Hi-HanChuan.{bsq,hdr} and _gt.{bsq,hdr}
Botswana Botswana/Botswana_data.mat, Botswana/Botswana_gt.mat
Trento TrentoRepo/Italy_hsi.mat, TrentoRepo/allgrd.mat

Examples:

python scripts/download_data.py --list
python scripts/download_data.py --dataset trento --data-dir data
python scripts/download_data.py --all --data-dir data

To work offline, place files using the paths above and run the environment and shape checks. More detailed dimensions, class counts, provenance notes, and licensing caveats are in docs/DATASETS.md.