---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: label
    dtype:
      class_label:
        names:
          '0': agricultural
          '1': airplane
          '2': baseball_diamond
          '3': beach
          '4': buildings
          '5': chaparral
          '6': dense_residential
          '7': forest
          '8': freeway
          '9': golf_course
          '10': harbor
          '11': intersection
          '12': medium_residential
          '13': mobile_home_park
          '14': overpass
          '15': parking_lot
          '16': river
          '17': runway
          '18': sparse_residential
          '19': storage_tanks
          '20': tennis_court
  splits:
  - name: train
    num_bytes: 252492192
    num_examples: 2100
  download_size: 252492968
  dataset_size: 252492192
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
license: cc-by-4.0
task_categories:
- image-classification
size_categories:
- 1K<n<10K
---
# Ucm Land Use Classification

UC Merced Land Use (UCM) scene classification: 2,100 aerial RGB image tiles of 256 x 256 pixels at about 0.3 m (1 foot) resolution, 100 per class across 21 land-use classes (agricultural, airplane, baseball diamond, beach, buildings, chaparral, dense/medium/sparse residential, forest, freeway, golf course, harbor, intersection, mobile home park, overpass, parking lot, river, runway, storage tanks, tennis court). The tiles were cut by Yang and Newsam (2010) from USGS National Map Urban Area Imagery of US towns; this copy is the CC BY 4.0 Figshare re-host used by the Remote Sensing study on federated fine-tuning (Fed-RSAdapter). A handful of tiles are a few pixels smaller than 256 x 256, as in the original release. The dataset contains 2,100 images across 21 classes: agricultural, airplane, baseball_diamond, beach, buildings, chaparral, dense_residential, forest, freeway, golf_course, harbor, intersection, medium_residential, mobile_home_park, overpass, parking_lot, river, runway, sparse_residential, storage_tanks, tennis_court.  
Images per class:
- agricultural: 100
- airplane: 100
- baseball_diamond: 100
- beach: 100
- buildings: 100
- chaparral: 100
- dense_residential: 100
- forest: 100
- freeway: 100
- golf_course: 100
- harbor: 100
- intersection: 100
- medium_residential: 100
- mobile_home_park: 100
- overpass: 100
- parking_lot: 100
- river: 100
- runway: 100
- sparse_residential: 100
- storage_tanks: 100
- tennis_court: 100

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

## Citation

```bibtex
@inproceedings{yang2010bag,
  title={Bag-of-visual-words and spatial extensions for land-use classification},
  author={Yang, Yi and Newsam, Shawn},
  booktitle={Proceedings of the 18th SIGSPATIAL International Conference on Advances in Geographic Information Systems},
  pages={270--279},
  year={2010},
  doi={10.1145/1869790.1869829}
}
```


The dataset itself can be cited as:

Nouman Ali, &amp; Zafar, B. (2018). *UCM image dataset* [Dataset]. figshare. https://doi.org/10.6084/M9.FIGSHARE.6085976.V2


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