image-labeling-yolo

Lightweight Tkinter GUI for reviewing and editing YOLO-format bounding boxes. Point it at your dataset and quickly add, edit, and save labels.

Features

• Simple, fast GUI for YOLO labels (class id, x_center, y_center, width, height)
• Keyboard-friendly: A add box, D delete, S save, digits 0–9 to set class
• Reads class names from yolo_dataset/data.yaml (names: list) when present
• Works with images under yolo_dataset/images/ and matching labels under yolo_dataset/labels/

Install

pip install image-labeling-yolo

Python 3.9+ is supported. Tkinter must be available in your Python installation (most CPython builds include it by default).

Quick start

1. Prepare folders (minimal structure):

yolo_dataset/
  images/
    img001.jpg
    img002.jpg
  labels/
    img001.txt   # optional; created on save if missing
    img002.txt
  data.yaml      # optional; provides class names via `names:`

2. Launch the GUI:

label-review

If you prefer, you can also run:

python -m review_gui

3. Use the folder pickers (right panel) to select your images/ and labels/ directories if the defaults are not detected.

Usage notes

• Navigation: Left/Right arrows or N/P
• Add a box: Press A, then click two opposite corners; drag handles to resize
• Delete selection: D
• Save: S (auto-saves when you navigate)
• Set class: Digits 0–9 (or choose from the dropdown)
• Class names: Provide a yolo_dataset/data.yaml with a names: list to show readable labels; otherwise numeric IDs are shown

Troubleshooting

• Tkinter missing: On some Linux distributions you may need to install Tk packages (e.g., sudo apt-get install python3-tk), or use a Python build that includes Tkinter.
• Headless servers: Run the GUI on a machine with a display or use X forwarding/remote desktop options.

License

MIT

Links

• Source and issues: https://github.com/mateus558/image-labeling-yolo