Skip to content

Dataset · Amber-River

2.6M Pixiv illustrations and tags collection

Amber-River/Pixiv-2.6M

Large image corpus from Pixiv in English and Japanese for generation and tagging.

What's inside

The dataset contains 2.6 million images collected from the Pixiv platform using a Python and Selenium crawler. The author purposefully gathered posts by tags missing or banned on Danbooru (such as landscapes and NSFW content). It is accompanied by annotations in English and Japanese.

What you can do with it

  • Training or fine-tuning text-to-image models and LoRA in anime style
  • Training automatic taggers and image-to-text description models
  • Extracting visual features and art classification
Tasks
Image classification, Image → text, Image → image, Text → image, Image feature extraction
Languages
English, Japanese
Size
1M–10M rows
Downloads this month
1.3K
Likes
48
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("Amber-River/Pixiv-2.6M")