Dataset · huggan
WikiArt paintings with labels
huggan/wikiart
81,444 art images with artist, genre, and style labels.
What's inside
The dataset contains 81,444 visual art pieces from WikiArt.org. Each image has classes: 129 artists, 11 genres, and 27 styles, including Unknown classes. Modality is images; tasks include classification, text-to-image, and image-to-text.
What you can do with it
- Training artist/genre/style classifiers
- Fine-tuning image-to-text models for painting description
- LoRA for text-to-image in artistic styles
- Tasks
- Image classification, Text → image, Image → text
- Size
- 10K–100K rows
- Downloads this month
- 10K
- Likes
- 250
Models trained on it
- ash12321/nano-banana-pro-detector-minimalImage classification 39
- ash12321/flux-detector-vitImage classification 30
- meetran/painting-classifier-keras-v1Image classification 26
- ash12321/seedream-4.5-specialist-h100Image classification 21
- ash12321/sdxl-detector-vitImage classification 16
- ash12321/imagegbt-1.5-specialist-h100Image classification 16
- ash12321/nano-banana-pro-specialist-h100Image classification 12
- sakhmatd/vit-wikiart-finetuned 8
How to load it
from datasets import load_dataset
ds = load_dataset("huggan/wikiart")