Data Annotation Tool
Data Annotation Tool
Professional Computer Vision Workflows • v1.0.0
1

Model Configuration

Load pre-trained YOLO or custom .pt weights for auto-annotation.

Pre-trained Base Model
OR
Local Model File (.pt, .onnx, .engine)
Inference Device
Detecting hardware devices…
Detected Classes
2

Dataset Management

Open an existing dataset or ingest raw images to begin.

Existing Dataset
OR CREATE NEW
Source Images Directory
Type
Split (T / V / Ts)
Linking images…0 / 0
Active Dataset Loaded
image.jpg
Size: 0x0 | Split: train
Objects in Image
Help & Shortcuts

🔄 End-to-End Workflow

  1. Select Model: Pick any supported model file (.pt, .onnx, etc.). The tool auto-detects your GPU and extracts model classes.
  2. Ingest Images: Select a source folder and click "Load Images". Hardlinking prevents duplicating image data.
  3. Review & Annotate: YOLO will pre-annotate images in the background. Use the draw mode and hotkeys to perfect your dataset.
  4. Auto-Split Dataset: Configure your Train/Valid/Test ratios and execute "Auto-Split Dataset".
  5. Train Model: Go to the Training panel, adjust hyperparameters, and click "Start Training".
  6. Export: Return to the Grid View and click "Export Dataset" to download the finalized package.

⌨️ Essential Keyboard Shortcuts

  • D Toggle Draw mode
  • F Fit image to screen
  • C Copy boxes from previous
  • J Jump to next unannotated
  • Q / E Cycle box class
  • 1 - 9 Set class 1-9
  • ← / → Prev / Next image
  • Enter Save & jump to next
  • Del Delete box
  • Ctrl+Z / Y Undo / Redo
  • Wheel Zoom in/out
  • Mid-drag Pan canvas