README.md

metadata

annotations_creators:
  - expert-generated
language:
  - en
license: mit
multilinguality:
  - monolingual
size_categories:
  - 10K<n<100K
task_categories:
  - image-classification
  - tabular-classification
task_ids:
  - land-cover-classification
pretty_name: 6-Band Hyperspectral Benchmark Classification Dataset

6-Band Multispectral/Hyperspectral Classification Dataset

Dataset Description

This dataset comprises a calibrated 6-band spectral imaging benchmark covering a 330 x 307 spatial scene across 7 ground-truth land-cover classes.

Dataset Specifications

• Spatial Resolution: 330 x 307 grid (101,310 total spatial pixels)
• Spectral Bands: 6 optical channels
• Target Classes: 7 annotated classes (Water, Vegetation, Soil, Built-up, Agriculture, etc.)
• Files:
  • data/data.csv: Continuous pixel reflectance feature vectors across 6 spectral channels
  • data/ref.csv: Ground-truth discrete class labels

Citation & Attribution

• Dataset Curator: Danesh Selwal (selwaldanesh@gmail.com)
• Co-authored by: Claude (claude@anthropic.com)