Skip to content

Objaverse

A large dataset of 800K+ annotated 3D objects

Visit official site

What Objaverse is

Objaverse is described as a large-scale dataset of 800K+ annotated 3D objects, each carrying a name, description, tags and other metadata. The collection covers many object types and varied visual styles.

What you can do with it

  • Use the objects for 3D model generation
  • Apply the collection as augmentation for 2D instance segmentation
  • Work on open-vocabulary embodied AI
  • Study the robustness of CLIP

Dataset contents

  • Static objects and animated objects
  • Characters with part annotations and decomposable models
  • Indoor and outdoor environments
  • A range of varied visual styles

Pros & cons

✓ What we like

  • Objects come with names, descriptions and tags
  • Visual styles and object types are varied

! What to watch out for

  • The material describes a dataset of objects rather than a single model
  • Named uses centre on generation, segmentation and robustness research

FAQ

How many objects are in the dataset?

The material states 800K+ annotated 3D objects.

What metadata comes with each object?

The material lists a name, description, tags and other metadata for every object.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review