img2tags 1.2.0

Tag images

img2tags

Tag images by using ONNX Runtime. The current default model is the converted WD 1.4 ConvNext Tagger V2 on https://huggingface.co/shirayu/img2tags. It works fine for anime images.

Setup

# For CPU
pip install -U 'img2tags[cpu]'

# For GPU
pip install -U 'img2tags[gpu]'

Run

# Generate .txt file
img2tags -i input_dir

# Generate .json file in another directory with CPU
img2tags --model shirayu/img2tags/SmilingWolf__wd-v1-4-convnext-tagger-v2 \
    --batch 4 \
    --ext json \
    --cpu \
    -i input_dir \
    -o output_dir

The values of default thresholds are stored in config.json in each model like this. You can set values in JSON format like --th {"0": 0.3, "4":0.2, "9":0.3}.

LICENSE

Apache 2.0

Some codes are based on codes in the following project.

• https://github.com/kohya-ss/sd-scripts

References

• Convert onnx models to change batch size