Dataset · Voxel51
35k LVIS image subset
Voxel51/LVIS
A 35,000-image English subset for object detection tasks, tailored for the FiftyOne library.
What's inside
Contains 35,000 English-annotated images for object detection. This is a reduced version of the full LVIS dataset, prepared for loading and processing via the FiftyOne framework. The card also mentions a Jupyter notebook recipe to assemble both this sample and the full original dataset.
What you can do with it
- Rapid prototyping and debugging of object detection pipelines
- Training and validation of lightweight computer vision models on limited data volume
- Tasks
- Object detection
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 239.6K
- Likes
- 11
How to load it
from datasets import load_dataset
ds = load_dataset("Voxel51/LVIS")