Image & Video Tagger

A Python desktop application for automatically tagging images and videos using ONNX models. Supports adding tags to text files, image metadata (EXIF, IPTC, XMP), or video metadata. Built with PyQt6, ONNX Runtime, and ExifTool.

Screenshot

Features

Process images (jpg, png, webp, etc.) and videos (mp4, mkv, mov, webm)
Extract frames from videos for tagging
Supports multiple ONNX tagging models from Hugging Face
Write tags to:
- Text files
- Image metadata (EXIF/IPTC/XMP)
- Video metadata (MP4, WebM)
Customizable thresholds for character/general tags
Option to hide rating tags or prioritize character tags
Extra tags and ignore lists
Recursive folder processing
Live preview of images/frames
Stop button to cancel processing

Requirements

Python 3.10+
PyQt6
numpy
pandas
Pillow
onnxruntime
huggingface_hub
piexif
exiftool installed and in your system PATH
ffmpeg installed and in your system PATH

All Python dependencies are listed in requirements.txt.

Installation

1. Clone the repository

git clone https://github.com/ThesiiNCey/IVTagger.git
cd IVTagger

2. Create a virtual environment (recommended)

python -m venv venv

3. Activate the virtual environment

Windows (cmd):

venv\Scripts\activate

Windows (PowerShell):

venv\Scripts\Activate.ps1

Linux/macOS:

source venv/bin/activate

4. Install dependencies

pip install -r requirements.txt

5. Ensure external tools are installed

ExifTool: Must be available in PATH
FFmpeg: Must be available in PATH

Usage

python ivtagger.py

Select a folder containing images or videos
Configure thresholds, output type, extra tags, ignore tags, and model
Click Start
Progress will be displayed in the right panel, along with a live preview

Output options

Text File: Writes tags to a .txt file with the same name as the media
Metadata: Writes tags to the image metadata (EXIF/IPTC/XMP)
Video Metadata: Writes tags to video metadata (comment field)

Models

Available tagging models (loaded from Hugging Face):

wd14-vit.v1
wd14-vit.v2
wd14-eva02.v3.large
wd-v1-4-vit-tagger.v3
wd14-vit.v3.large

The model CSV (selected_tags.csv) contains tag names and categories (general, character, rating).

Notes

PNG, JPEG, and WebP images are fully supported for metadata writing.
Video frame extraction uses FFmpeg; large videos may take time.
If using Overwrite Metadata, previous tags are replaced. Otherwise, new tags are appended.
Stop button can be used to abort processing at any time.

Contact

For questions, feedback, or issues, you can contact me via:

Email: [email protected]
Matrix: @aiiko:matrix.mochiart.moe

License

Apache 2.0 License

Acknowledgements

ONNX Runtime for fast inference
Hugging Face for hosting tagging models
ExifTool for robust metadata handling
PyQt6 for GUI