Dataset · allenai
800K+ annotated 3D objects
allenai/objaverse
Collection of over 800k 3D objects with annotations for training 3D models.
What's inside
Objaverse is a massive 3D object dataset; the card indicates 800K+ annotated objects. The license for the entire set is ODC-By v1.0, individual objects are distributed under Creative Commons: CC-BY 4.0 - 721K, CC-BY-NC 4.0 - 25K, and others. Modality is 3D, and the language is specified as en.
What you can do with it
- Train a 3D object classification or generation model
- Test a 3D detector or segmenter
- Create embeddings for 3D model search
- Languages
- English
- Downloads this month
- 2.4M
- Likes
- 506
Models trained on it
- stabilityai/TripoSRImage → 3d 237.7K
- stabilityai/stable-fast-3dImage → 3d 27.9K
- tencent/Hunyuan3D-Part 5.5K
- stabilityai/stable-point-aware-3dImage → 3d 4.4K
- zxhezexin/openlrm-base-obj-1.0Image → 3d 2.1K
- zxhezexin/openlrm-small-obj-1.0Image → 3d 2.1K
- cocktailpeanut/sf3dImage → 3d 1.2K
- zxhezexin/openlrm-large-obj-1.0Image → 3d 375
How to load it
from datasets import load_dataset
ds = load_dataset("allenai/objaverse")