metadata

license: cc-by-4.0
dataset_info:
  features:
    - name: image
      dtype: image
    - name: label
      dtype:
        class_label:
          names:
            '0': banded
            '1': blotchy
            '2': braided
            '3': bubbly
            '4': bumpy
            '5': chequered
            '6': cobwebbed
            '7': cracked
            '8': crosshatched
            '9': crystalline
            '10': dotted
            '11': fibrous
            '12': flecked
            '13': freckled
            '14': frilly
            '15': gauzy
            '16': grid
            '17': grooved
            '18': honeycombed
            '19': interlaced
            '20': knitted
            '21': lacelike
            '22': lined
            '23': marbled
            '24': matted
            '25': meshed
            '26': paisley
            '27': perforated
            '28': pitted
            '29': pleated
            '30': polka-dotted
            '31': porous
            '32': potholed
            '33': scaly
            '34': smeared
            '35': spiralled
            '36': sprinkled
            '37': stained
            '38': stratified
            '39': striped
            '40': studded
            '41': swirly
            '42': veined
            '43': waffled
            '44': woven
            '45': wrinkled
            '46': zigzagged
  splits:
    - name: train
      num_bytes: 226313270.04
      num_examples: 1880
    - name: test
      num_bytes: 172035822
      num_examples: 1880
    - name: validation
      num_bytes: 222278767.48
      num_examples: 1880
  download_size: 629315160
  dataset_size: 620627859.52
task_categories:
  - image-classification
language:
  - en
tags:
  - texture
  - computer-vision
pretty_name: Describable Textures Dataset
size_categories:
  - 1K<n<10K

Dataset Card for Describable Textures Dataset (DTD)

Dataset Description

Homepage: https://www.robots.ox.ac.uk/~vgg/data/dtd/
Repository: https://github.com/mcimpoi/deep-fbanks
Paper: https://openaccess.thecvf.com/content_cvpr_2014/html/Cimpoi_Describing_Textures_in_2014_CVPR_paper.html
Leaderboard: https://paperswithcode.com/sota/image-classification-on-dtd

Dataset Summary

Texture classification dataset; consists of 47 categories, 120 images per class.

Data Splits

Equally split into train, val, test; The original paper proposed 10 splits; recent works (BYOL, arxiv:2006.07733) use only first split.

Licensing Information

Not defined at https://www.robots.ox.ac.uk/~vgg/data/dtd/

Citation Information

@InProceedings{cimpoi14describing, Author = {M. Cimpoi and S. Maji and I. Kokkinos and S. Mohamed and and A. Vedaldi}, Title = {Describing Textures in the Wild}, Booktitle = {Proceedings of the {IEEE} Conf. on Computer Vision and Pattern Recognition ({CVPR})}, Year = {2014}}