Project-AgML/siri_whu_scene_classification

Siri Whu Scene Classification

SIRI-WHU scene classification: 2,400 high-resolution RGB image tiles of 200 x 200 pixels at 2 m resolution, cut from Google Earth imagery of urban and rural Wuhan, China, 200 per class across 12 land-use classes (agriculture, commercial, harbor, idle land, industrial, meadow, overpass, park, pond, residential, river, water). Released by the RS_IDEA group at Wuhan University with Zhao et al. (2016); this copy is the CC BY 4.0 Figshare re-host used by the Remote Sensing study HFCC-Net. The dataset contains 2,400 images across 12 classes: agriculture, commercial, harbor, idle_land, industrial, meadow, overpass, park, pond, residential, river, water.
Images per class:

• agriculture: 200
• commercial: 200
• harbor: 200
• idle_land: 200
• industrial: 200
• meadow: 200
• overpass: 200
• park: 200
• pond: 200
• residential: 200
• river: 200
• water: 200

This dataset is indexed on https://project-agml.github.io/ as part of the AgML python library.

Citation

@article{zhao2016dirichlet,
  title={Dirichlet-derived multiple topic scene classification model for high spatial resolution remote sensing imagery},
  author={Zhao, Bei and Zhong, Yanfei and Xia, Gui-Song and Zhang, Liangpei},
  journal={IEEE Transactions on Geoscience and Remote Sensing},
  volume={54},
  number={4},
  pages={2108--2123},
  year={2016},
  doi={10.1109/TGRS.2015.2496185}
}

The dataset itself can be cited as:

Zafar, B., & Nouman Ali. (2019). SIRI_WHU Dataset [Dataset]. figshare. https://doi.org/10.6084/M9.FIGSHARE.8796980.V2

This dataset was reformatted from its original format to match HuggingFace standards.