ultimate-ai-labeller 0.1.1

AI Labelling tool for computer vision

Ultimate AI Labeller

Desktop image-annotation tool for computer vision datasets, built with Tkinter.

Features

• Fast bounding-box labeling with mouse drag, resize handles, and move operations
• Keyboard-first workflow (F next, D previous, A auto red-detection, Space fuse boxes)
• Undo/redo support (Ctrl+Z, Ctrl+Y)
• YOLO-assisted detection (Ultralytics) with confidence control
• Box fusion utilities for overlapping/nearby boxes
• Scrollable right settings panel for small window sizes
• Remove bad frames from current split (train/val/test)
• Restore removed frames from a selection dialog
• Image jump selector (dropdown) to move directly to any image in the split
• Session resume memory (reopens last project/split/image)
• Language and theme switch support in-app

Dataset Layout

The app expects a YOLO-style folder structure:

your_project/
  images/
    train/
    val/
    test/
  labels/
    train/
    val/
    test/

Supported image extensions: .png, .jpg, .jpeg
Label format: YOLO .txt (class cx cy w h normalized).

Removed frames are moved to:

your_project/
  removed/
    train|val|test/
      images/
      labels/

Install

pip install -e .

or:

pip install .

Run

After install:

ai-labeller

Or directly:

python src/ai_labeller/main.py

Shortcuts

• F: save and next image
• D: previous image
• A: auto red detection
• Space: fuse boxes
• Ctrl+Z: undo
• Ctrl+Y: redo
• Delete: delete selected box

Notes

• First YOLO use requires model weights (default: yolov8n.pt) available in your working directory or configured path.
• Session state is saved to ~/.ai_labeller_session.json.