README.md

---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: label
    dtype:
      class_label:
        names:
          '0': brick
          '1': carpet
          '2': ceramic
          '3': fabric
          '4': foliage
          '5': food
          '6': glass
          '7': hair
          '8': leather
          '9': metal
          '10': mirror
          '11': other
          '12': painted
          '13': paper
          '14': plastic
          '15': polishedstone
          '16': skin
          '17': sky
          '18': stone
          '19': tile
          '20': wallpaper
          '21': water
          '22': wood
  splits:
  - name: train
    num_bytes: 2017774670.25
    num_examples: 48875
  - name: test
    num_bytes: 240662928
    num_examples: 5750
  - name: validation
    num_bytes: 135991437.125
    num_examples: 2875
  download_size: 2350770974
  dataset_size: 2394429035.375
license: cc-by-4.0
task_categories:
- image-classification
language:
- en
size_categories:
- 10K<n<100K
---
# Materials in Context Dataset (MINC-2500)

## Dataset Description
- **Homepage:** http://opensurfaces.cs.cornell.edu/publications/minc/
- **Paper:** https://openaccess.thecvf.com/content_cvpr_2015/html/Bell_Material_Recognition_in_2015_CVPR_paper.html

## Dataset Summary
(from the website)
MINC-2500 is a patch classification dataset with 2500 samples per category 
(Section 5.4 of the paper). This is a subset of MINC where samples have been
sized to 362 x 362 and each category is sampled evenly. The original resolution
images are not needed as we include the extracted patches in the archive.