Skip to content

ODIN Model

Segment and label 2D images and 3D point clouds in one model

Visit official site

What ODIN Model is

The model is described as an omni-dimensional instance segmentation approach that uses a transformer to segment and label 2D RGB images and 3D point clouds.

What you can do with it

  • Segment and label 2D RGB images
  • Segment and label 3D point clouds
  • Alternate fusion within 2D views and across 3D views

Pros & cons

✓ What we like

  • One model covers both 2D and 3D perception
  • Code and checkpoints are stated to be on the project site

! What to watch out for

  • Feature operations are separated between 2D and 3D views
  • Benchmark results depend on the point cloud input used

FAQ

What inputs does it handle?

The material names 2D RGB images and 3D point clouds.

Where is the code?

The material states code and checkpoints are on the project website.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review