pyqt-image-annotation-tool 0.1.0

Desktop tool for whole-image classification and bounding-box image annotation

PyQt Image Annotation Tool

A desktop app for labeling images in a folder — either assigning whole-image classes, or drawing bounding boxes. Built with PySide6.

Whole-image classification	Bounding boxes

Features

Whole-image classification

• Assign one or more labels to each image
• Move or copy labeled images into per-label sub-folders, or leave files untouched
• Export to CSV and/or XLSX

Bounding boxes

• Draw, move, resize, and delete rectangular regions per image, each with its own label
• Export to COCO JSON and/or YOLO txt (with classes.txt)

Both modes

• Zoom & pan on the image (mouse wheel to zoom, drag to pan)
• Thumbnail filmstrip showing every image's labeling status
• Undo (Ctrl+Z)
• Dark theme
• Drag & drop a folder onto the app to open it
• Autosave: progress is saved continuously and offered back to you if you reopen the same folder, so a crash or an accidental close doesn't lose your work

Installation

From PyPI:

pip install pyqt-image-annotation-tool
annotator

Standalone build (no Python required): download the archive for your OS from the latest release and run the annotator executable inside it.

From source:

git clone https://github.com/nikuznetsov/PyQt-image-annotation-tool.git
cd PyQt-image-annotation-tool
pip install -e .
annotator            # or: python -m annotator

Usage

1. Pick the folder containing your images (or drag & drop it onto the window).
2. Choose an annotation type — whole-image classification or bounding boxes.
3. For classification, also choose a mode:
   • csv — labels are tracked and exported to CSV/XLSX; files aren't moved.
   • copy / move — a folder is created per label and labeled images are copied/moved into it.
4. Enter your label names (or load them from a text file, one label per line).
5. Click Next and start labeling.

Keyboard shortcuts

Key	Action
→ / ←	Next / previous image
1–9, 0	Select label (wraps after the 9th)
Ctrl+Z	Undo
Mouse wheel	Zoom
Middle-drag (bbox mode) / drag (classification mode)	Pan
Delete (bbox mode)	Remove selected box

Development

pip install -e ".[dev]"
pytest
ruff check .

See CONTRIBUTING.md for more.

Contributing

Pull requests are welcome.

License

MIT