Here is Gwern Danbooru 2018 dataset with 2.536.329 Danbooru images
till 01.01.2019 rating:safe resized to 512x512 px with some meta-information
used for image recognition training in zipped format, acceptible to all torrent clients.
Meta information included in “initial” JSON format and “normalized” 3-tables CSV
(posts with some additional stats, taglist with some additional info, tags occurrences in posts).
There is the next volume for 2019-2021.
NOTE BOORU CHARS - my compilation of 1.227.622 thumbnails (also 512x512px)
for best art images from several sources (only ~360.000 taken from this release)
enriched with much more calculated metadata, including face detected.
Also I support a BOORU CHAR dataset with 1280px samples release 2021 and release 2015 , more to come.
This is my mainstream.
Comments - 2
Astral
Neat.
SomaHeir
Thanks for this update!