Ctrlk
Contact iMeritAngo Hub WebsiteLog In to Ango Hub
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Image Viewer (IViewer)

The Image Viewer is a feature that displays camera sensor images, each showing the projection of a selected cuboid. It helps users analyze how an object appears across multiple camera angles, using calibration data to align point cloud annotations with images.

lightbulb-on

Image Viewer helps users analyze how an object appears across multiple camera angles.

Use Cases

  • Multi-camera review: See how an object is captured by different cameras at the same frame.

  • Annotation verification: Check the accuracy of cuboid placement across views.

  • Tracking moving objects: Follow an object entering one view and exiting another within the same frame.

  • Sensor layout: Identify the most helpful camera view based on clarity or proximity.

Benefits

  • Provides a complete visual context of the object from various angles.

    • Supports easier tracking of moving objects across views.

  • Allows custom arrangement of camera feeds by:

    • Best view

    • Closest to the ego vehicle

    • Locking preference places key views at the front for quicker access.

  • Enhances accuracy by relying on point cloud-to-image calibration.

Steps to Use

1

Enable IViewer

Enable the IViewer from the top left of the menu.

2

Place and Resize IViewer

Adjust the IViewer to a location on the tool that is most convenient and resize it to the dimensions that help identify objects.

3

Select a Cuboid

Select a cuboid in the 3D point cloud.

4

View Projections

The IViewer displays all camera images where the cuboid is projected.

5

Rearrange and Lock Sensors

Rearrange sensors to prioritize views by quality or distance.

Lock important sensors so they always appear first.

6

Validate and Track

Observe how the object appears in each camera to validate, adjust, or track annotations effectively.

The Image Viewer displays the projection of the selected cuboid based on the calibration data provided in the dataset - not the camera sensors the respective rectangles are created on. It automatically shows the cuboid on all camera views where it is projected using the available calibration.

Last updated