metadata

dataset_info:
  features:
    - name: image
      dtype: image
    - name: label
      dtype:
        class_label:
          names:
            '0': cat
            '1': dog
            '2': elephant
            '3': fish
            '4': giraffe
            '5': horse
            '6': lion
            '7': penguin
            '8': rabbit
            '9': tiger
  splits:
    - name: train
      num_bytes: 14149466600
      num_examples: 10000
    - name: test
      num_bytes: 7115540480
      num_examples: 5000
  download_size: 23294717535
  dataset_size: 21265007080
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*
license: other
task_categories:
  - image-classification
tags:
  - biology

Fake Animals

This dataset is a collection of 10 types of animal images created using Stable Diffusion 3.5 Large.

Dataset Details

Dataset Description

All images have a resolution of 1024x1024 and are divided into training and test sets with 1000 and 500 images, respectively. The classes are as follows:

'0': cat
'1': dog
'2': elephant
'3': fish
'4': giraffe
'5': horse
'6': lion
'7': penguin
'8': rabbit
'9': tiger

The data was generated according to the prompt "a photo of {animal}."

Uses

from datasets import load_dataset
ds = load_dataset('Prgckwb/fake-animals')
ds_train = ds['train']
ds_test = ds['test']

License

Since this dataset was created using Stable Diffusion 3.5 Large, it should be used for non-commercial purposes and in accordance with this license.