This fork rewrites tag handling and adds support for lepton files. Only compatible with linux unfortunately. (If you can get exempi to compile on windows then it may work but I gave up)
Uses IPTC Keywords for jpgs and lepton files and XMP Subjects for png files
Automatically adds booru style tags to an image or directory of images by using this neural net model: https://github.com/KichangKim/DeepDanbooru
You will need:
- lepton https://github.com/dropbox/lepton, follow instructions and build, then put in auto-anime-tag folder
- pre-trained model deepdanbooru-v3 from https://github.com/KichangKim/DeepDanbooru, grab the model from the 'releases' section, or train your own, put the files in
auto-anime-tag/deep
- Exempi, install with your package manager
- Python 3.8 (3.9 will not work for now)
git clone https://github.com/winterNebs/auto-tag-anime.git
cd auto-tag-anime
python3.8 -m pip install -r requirements.txt
tag one image:
python3.8 auto-tag-anime.py "example.jpg"
tag multiple images in a directory (saves on model loading overhead):
python3.8 auto-tag-anime.py "/path/to/directory/"
Skip tagged images:
python3.8 auto-tag-anime.py --skip "/path/to/directory/"
display help message:
python3.8 auto-tag-anime.py -h
- See a list of tags the model will predict in 'tags.txt' inside of the deepdanbooru-v3 folder
- checks for images in subdirectories