README.md

metadata

license: creativeml-openrail-m
task_categories:
  - image-to-text
  - image-to-image
  - text-to-image
tags:
  - art
size_categories:
  - 1K<n<10K

Dataset Format

• File Type: .png

Data Sources

1. Online Collection: Images collected from public datasets such as Kaggle and Huggingface.
2. Offline Collection: Hand-drawn sketches by seven-year-old children, gathered through school visits and other offline activities.

Data Acquisition

• Team members downloaded and collected images from online datasets.
• Offline data was collected by visiting schools and gathering hand-drawn sketches from children.

Dataset Characteristics

1. Diverse Artistic Styles: The style dataset includes works inspired by famous artists such as Van Gogh and Monet, as well as various art movements like Impressionism, Cubism, and Abstract art.
2. High Image Quality: All images are of high resolution and clarity.
3. Variety of Image Types: The dataset covers a wide range of subjects, including portraits and landscapes, supporting diverse creative scenarios.
4. Custom Style Samples: To enable user-defined style generation, the dataset includes a rich collection of style samples.

Dataset Scale

• Style Datasets: Includes oil painting, colored pencil, and crayon datasets, each with 150 images.
• Sketch Datasets: Includes digital sketch and pencil sketch datasets, each with 100 - 200 images.

Data Sources

• Kaggle: Real AI Art Dataset (for style datasets except colored pencil and crayon)
• Colored pencil and crayon images were collected from various online sources.