Tagging Mode

Draw bounding boxes and classify road assets

Overview

Tagging Mode allows you to draw bounding boxes around road assets in video frames and classify them with AI assistance. Tagged assets are stored with GPS coordinates and can be exported for ML training or used as your road asset inventory.

Entering Tagging Mode

To start tagging assets:

Keyboard

Press T to toggle tagging mode

Click

Click the Tag button in the video toolbar

Note: Tagging mode captures a still frame from the video. Navigate to the desired position before entering tagging mode.

Drawing Bounding Boxes

Click and drag on the frame to draw a box around an asset:

  1. Click where you want to start the box (corner)
  2. Drag to the opposite corner while holding the mouse button
  3. Release to complete the box (minimum size: 2% of frame)

Box colors:

  • Dashed blue: Currently drawing
  • Solid green: Saved boxes
  • Solid blue: Selected box

Managing boxes:

  • Click a box to select it
  • Click the red X to delete a selected box
  • Double-click to open classification modal

Road Side Detection

The road side is automatically detected based on box position:

LHS

Left Hand Side

Center X < 35%

CENTER

Road Center

35% ≤ X ≤ 65%

RHS

Right Hand Side

Center X > 65%

A grid overlay on the canvas shows these zones. You can manually override the detected side in the classification modal.

Asset Classification

When you select a box, the classification modal opens with these fields:

Asset Type

Over 40 types available:

  • Traffic signs (warning, regulatory, guide)
  • Road markings (lines, symbols)
  • Poles and posts
  • Barriers and guardrails
  • Drainage structures
  • And more...

Condition

  • Good
  • Fair
  • Poor
  • Very Poor

Additional Fields

  • GPS coordinates (auto-filled from video)
  • Road side (LHS/CENTER/RHS)
  • Chainage (position on road)
  • Notes (free text)
  • Sign code/description (for traffic signs)

AI-Assisted Classification

When you select a bounding box, AI automatically analyzes the cropped image:

What AI provides:

  • Suggested asset type
  • Confidence score (0-100%)
  • Extracted text (for signs)
  • Sign code matching (for traffic signs)

Note: AI suggestions are pre-filled in the form but you can always override them. The AI confidence score is saved with the asset for quality filtering.

Tip: For traffic signs, AI uses visual similarity matching against a reference database to identify the exact sign code.

Keyboard Shortcuts

KeyAction
TEnter/exit tagging mode
EscExit tagging mode
Ctrl + SSave all changes
?Show help modal

Related Guides

enmsid