README.md

metadata

task_categories:
  - mask-generation
language:
  - en
tags:
  - Forgery
  - Detection
  - Localization
  - Inpainting
pretty_name: tiny-audits
size_categories:
  - 10K<n<100K
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
dataset_info:
  features:
    - name: image
      dtype: image
    - name: label
      dtype:
        class_label:
          names:
            '0': manipulated
            '1': mask
            '2': original
  splits:
    - name: train
      num_bytes: 700387952.5
      num_examples: 48500
  download_size: 2668321513
  dataset_size: 700387952.5