Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
image
imagewidth (px)
512
3.4k
label
class label
2 classes
0images_raw
0images_raw
0images_raw
0images_raw
0images_raw
0images_raw
0images_raw
0images_raw
0images_raw
0images_raw
0images_raw
0images_raw
1images_square_512
1images_square_512
1images_square_512
1images_square_512
1images_square_512
1images_square_512
1images_square_512
1images_square_512
1images_square_512
1images_square_512
1images_square_512
1images_square_512

Cleveland Museum of Art CC0 Style Reference Benchmark Dataset

Scope & Exclusions: This dataset contains ONLY public domain style reference images from the Cleveland Museum of Art. It does NOT contain any COCO content images, WikiArt, Fons, or third-party proprietary assets.

Dataset Structure

.
├── README.md                           # Dataset Card
├── LICENSE                             # Dual-boundary license declaration
├── data/
│   ├── style_registry.csv              # Primary metadata table with accession numbers and SHA256 hashes
│   ├── images_raw/                     # Original high-resolution public domain artwork images
│   └── images_square_512/              # Derived 512x512 inputs (deterministic centre crop + Lanczos)
├── metadata/
│   ├── source_credits.csv              # Attribution and museum source records
│   └── preprocessing.json              # Preprocessing policy and PIL version record
└── scripts/
    └── prepare_square_style_candidate_assets.py # Reproducible asset preparation script

Dataset Contents & Metadata

The dataset consists of public domain artworks representing diverse artistic styles (e.g., De Stijl, Expressionism, Cubism, Vorticism, Impressionism, Post-Impressionism, Naïve Art, Pointillism).

Stable ID Accession Number Title Artist License
B01 1967.215 Composition with Red, Yellow, and Blue Piet Mondrian CC0
B02 1966.49 Wrestlers in a Circus Ernst Ludwig Kirchner CC0
B03 2006.139 Night Café Louis Marcoussis CC0
B04 1987.50 Illustration (Typhoon) Edward Wadsworth CC0
B05 1960.81 Water Lilies (Agapanthus) Claude Monet CC0
B06 1947.209 The Large Plane Trees Vincent van Gogh CC0
B07 1926.1653 On the Beach, No. 3 Maurice Prendergast CC0
B08 1983.185 Woman with Red Hair and Green Eyes Edvard Munch CC0
B09 1953.212 Under the Trees Édouard Vuillard CC0
B10 1929.951 Outskirts of Paris Henri Rousseau CC0
B11 1958.344 At the Concert Parisien Georges Seurat CC0
B12 1958.21 Mount Sainte-Victoire Paul Cézanne CC0

Preprocessing & Verification Pipeline

  1. Source Download: Original images downloaded directly from the Cleveland Museum of Art Open Access API print/web image endpoints.
  2. Derivation Policy: 512×512 square inputs generated via deterministic centre crop followed by Lanczos resampling (centering=(0.5, 0.5)), preventing aspect ratio distortion.
  3. Integrity Checks:
    • SHA256 cryptographic hashes verified for both raw and 512×512 derived images.
    • PIL image loading verified for valid RGB/JPEG channels and dimensions.
    • Per-artwork share_license_status verified directly against the Cleveland API.

Intended Use & Known Limitations

  • Intended Use: Evaluation of training-free multimodal diffusion models (e.g., BAGEL, ConsistCompose) on dual-reference style transfer tasks (content reference + style reference -> stylized output).
  • Known Limitations:
    • Style reference images may contain distinct spatial layouts, human figures, or semantic subjects (e.g., Kirchner's wrestlers, Munch's portrait, Prendergast's beach scene). Evaluators MUST conduct explicit reference leakage audits to ensure the model transfers rendering style rather than copying style-image semantics/layouts.
    • This dataset does NOT claim to represent an exhaustive distribution of historical art styles, nor does it represent all digital art or watercolor media.

Citation & Attribution

When using this dataset, please credit the Cleveland Museum of Art Open Access Initiative:

The Cleveland Museum of Art Open Access Initiative (CC0 1.0 Universal)
https://clevelandart.org/open-access
Downloads last month
56