Dataset · huanngzh
Edit3D-Bench 3D model editing benchmark
huanngzh/Edit3D-Bench
A set of 100 3D models with English text prompts and annotations for evaluating localized 3D editing.
What's inside
The dataset was created for the VoxHammer paper and includes 100 3D models from the Google Scanned Objects and PartObjaverse-Tiny collections. For each model, 3 English editing prompts are prepared along with a full set of labeled 3D assets (total size ranging from 10 to 100 thousand entries).
What you can do with it
- Evaluating algorithms and neural networks for text-based 3D object editing
- Comparing 3D scene generation and modification methods (text-to-3d, image-to-3d)
- Tasks
- Image → 3d, Text → 3d
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 21K
- Likes
- 12
How to load it
from datasets import load_dataset
ds = load_dataset("huanngzh/Edit3D-Bench")