An anime image taggger trained with cleaned danbooru2021.
ML-Danbooru uses the structure of multi-scale recognition, which has better detail recognition ability and more accurate than wd14.
使用这一插件需要修改WebUI的requirements_versions.txt
文件,
将其中的einops==0.4.1
替换为einops>=0.6.0
!
To use this extension, you need to modify the requirements_versions.txt
file of WebUI.
Replace einops==0.4.1
with einops>=0.6.0
!