Skip to content

Auto Annotation

This page is used to annotate images for object detection tasks. Annotators can draw bounding boxes, rectangles, polygons, lines, or points. This page provides all the necessary tools to create, edit, manage, review, and export annotations efficiently.

Object Detection Workspace

Page Overview

When you open a job, the annotation workspace loads with:

  • Top Toolbar – Navigation, zoom, undo/redo, shortcuts, and submission controls.
  • Left Panel (Job Images) – List of images in the job.
  • Center Canvas – Main image annotation area.
  • Right Panel (Labels & Tools) – Object tools, labels list, and auto-label options.

Top Navigation & Controls

  • Back
    Returns to the previous screen (job or project list).
    If there are unsaved changes, a confirmation popup appears for confirmation. Exit

  • Toggle Sidebar – On click, this hides or shows the Job Images section.

  • Job #68 – Displays the current job ID.

  • Annotation – Indicates that the current task is in annotation mode.

  • Admin – Displays the role of the logged-in user.

Toolbar Controls

  • Pan Tool – Move around the image canvas.
  • Zoom In / Zoom Out – Adjust zoom level for precise annotation.
  • Undo / Redo – Revert or reapply recent annotation actions.
  • Keyboard Shortcuts – Click the keyboard icon to view available shortcuts.
    Keyboard Shortcuts

Prompt Seek - allows users to detect and annotate objects using simple text prompts.

  1. Click Prompt Seek from the toolbar.
  2. Enter the object names you want to detect (e.g., person, face, car).
  3. Adjust the detection threshold if required.
  4. Click Detect Objects.

The system automatically identifies matching objects in the image and creates annotations, helping users annotate images faster and reduce manual effort.

Prompt Seek

Smart Select

Smart Select - is an AI-assisted annotation tool that automatically identifies and selects objects within an image. Instead of manually drawing annotations, users can simply click on an object, and the system intelligently detects its boundaries and creates a more accurate selection.

This feature helps speed up the annotation process, reduces manual effort, and improves annotation accuracy.

  • Submit – Finalizes the annotations. On click, a popup appears; clicking Submit saves the annotations for the job. Submit

  • Toggle Sidebar – On click, this hides or shows the right-side labels panel.

  • Options – Opens additional settings for annotation preferences. Options

The left sidebar displays all images included in the current job.

Features:

  • Job Images (4) – Shows the total number of images in the job.
  • Thumbnail previews allow quick navigation between images.
  • The currently selected image is highlighted.
  • Users can scroll through images and click any thumbnail to load it into the main canvas.

Center Canvas – Annotation Area

This is where annotations are drawn:

  • Each object is marked with colored shapes and labels (e.g., rect, rect3).
  • Selected annotations show resize handles.
  • You can move, resize, or delete any annotation.

Pagination

Navigate through job images using previous and next arrows at the bottom.

Right Panel – Objects & Labels

Objects Tab

Displays the list of all annotations created on the current image.

  • Auto Label – Automatically suggests annotations using an AI model (if enabled for the project). Auto Label

Labels Section

Displays all labels used in the current image:

  • Each label shows:
  • Label name (e.g., rect, rect3)
  • Coordinates

  • Delete icon to remove the annotation

  • Labels are selectable, allowing:

  • Highlighting the corresponding shape on the canvas.
  • Editing or deleting the annotation.

Label Editing & Management

Edit Label (Inline Panel)

When you click on a label or shape, the Edit Label panel opens.

Fields & Options:

  • Object Name Input – Rename the label.
  • Save / Cancel – Apply or discard changes.
  • In This Image – Shows labels used in the current image.
  • Project Labels – Shows all labels available in the project (e.g., car, person, bus, traffic light). Edit Label

Annotation Workflow

  1. Select an Image from the left sidebar.
  2. Choose a Tool (e.g., Rectangle).
  3. Draw on the Canvas over the object of interest.
  4. Assign or Confirm a Label from the labels panel.
  5. Repeat for all objects in the image.
  6. Navigate to the next image using the thumbnails or arrows.
  7. Once all images are annotated, click Submit to complete the job.

Submission & Review

Submit Button

Once all images are annotated, click Submit.

Submit Batch for Review Popup

  • Confirms that the job will move to the Review stage.
  • Warns that annotations cannot be edited after submission.
  • Options:
  • Cancel
  • Yes, Submit Submit

Manage Labels (Global)

From the Options dropdown, select Manage Labels.

Manage Labels Popup

  • View all project labels with usage count.
  • Rename any label globally across all images.
  • Delete unused or incorrect labels.
  • Click Done to save changes. Manage Label

Exporting Annotations

From the Options menu, select Export Data.

Export Job Annotations Popup

  • Export Format – Choose output format (e.g., COCO, YOLO, JSON, etc.).
  • Include Images – Option to bundle images with annotations.
  • Export Dataset Name – Custom name for the export file.
  • Export Destination – Default: Local Storage (Download).
  • Actions:
  • Cancel
  • Export Export_Job

Exit Project Confirmation

If you try to leave the page with unsaved changes:

Exit Project Popup

  • Message: “Are you sure you want to leave the Editor? You will permanently lose all your progress if you haven't saved.”
  • Options:
  • Cancel
  • Exit Exit Project