metadata

license: mit
task_categories:
  - image-classification
  - image-to-text
  - image-to-image
  - text-to-image
  - image-feature-extraction
language:
  - en
  - ja
tags:
  - not-for-all-audiences
size_categories:
  - 1M<n<10M

Pixiv 2.6M specific

This dataset contain 2.6M images from Pixiv.

Introduction

This dataset aims to add some concepts which is lacked or banned in danbooru.

Images Content

The images is collected with search cralwer made in selenium and searched following tags:

R-18: popularity sort, 5000pages
ロリ: popularity sort, 2500pages
足: popularity sort, 2500pages
Scenary: popularity sort, 2000pages
男の子: popularity sort, 2000pages

This dataset contain totally 2,639,665 images. (Deduplicated within this set)
This dataset contain totally 747,482 pixiv posts. (Deduplicated within this set)

Format

As my other datasets. This dataset save images in tars which can be read by webdataset. (Or you can refer to my HakuBooru's code)
All the images are resized to max 4M pixels and encoded with webp 90% quality (PIL). Which is 608GiB in total.
Original files(4.29TiB in total) may be uploaded in future if lot of ppl want it.