Ticino-RS-Dataset

Ticino dataset is a novel multi-modal remote sensing dataset specifically tailored for semantic segmentation tasks. It covers an area of about 1332 $km^2$.

It incorporates five distinct modalities:

• RGB
  • spatial resolution: 1.86m/px (vertical) - 2.64m/px (horizontal)
  • 3 bands
• Digital Terrain Model (elevation of the ground)
  • spatial resolution: 5m/px
  • 1 band
  • from 51.86 to 124.75 meters
• Panchromatic
  • spatial resolution: 5m/px
  • 1 band [400-700 nm]
• Hyperspectral
  • VNIR
    • spatial resolution: 30m/px → 5m/px (pansharpened)
    • 63 bands [400-1010 nm] → 60 bands (cleaned)
  • SWIR
    • spatial resolution: 30m/px → 5m/px (pansharpened)
    • 171 bands [920-2500 nm] → 122 bands (cleaned)

It includes two pixel-level labelings:

• Land Cover:
  • 8 classes:
    • 0 Background
    • 1 Building
    • 2 Road
    • 3 Residential
    • 4 Industrial
    • 5 Forest
    • 6 Farmland
    • 7 Water
• Soil Agricultural Use:
  • 10 classes:
    • 0 Background
    • 1 Other agricultural crops
    • 2 Forage crops
    • 3 Corn
    • 4 Industrial plants
    • 5 Rice
    • 6 Seeds
    • 7 Man-made areas
    • 8 Water bodies
    • 9 Natural vegetation

How to download

To download the entire dataset, go to this webpage and follow the instructions..

Citation

if you use this dataset, please cite us :)

@article{barbato2024ticino,
  title={Ticino: A multi-modal remote sensing dataset for semantic segmentation},
  author={Barbato, Mirko Paolo and Piccoli, Flavio and Napoletano, Paolo},
  journal={Expert Systems with Applications},
  volume={249},
  pages={123600},
  year={2024},
  publisher={Elsevier}
}