Dataset · dronefreak
BDD100K road scenes for segmentation
dronefreak/BDD100K-Semantic-Segmentation
Unofficial mirror of a 10k image subset of BDD100K for road semantic segmentation.
What's inside
The repository hosts an unofficial copy of a subset of the well-known BDD100K dataset from UC Berkeley researchers. Inside there are from 10 to 100 thousand (10 thousand indicated in the description) road scene images with semantic segmentation markup in English. Distributed for non-commercial, educational, and research tasks.
What you can do with it
- Training and testing semantic segmentation models for autonomous transport
- Quick downloading of a BDD100K subset for educational and research experiments
- Benchmarking computer vision neural networks in difficult road conditions
- Tasks
- Image segmentation
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 34.9K
- Likes
- 2
How to load it
from datasets import load_dataset
ds = load_dataset("dronefreak/BDD100K-Semantic-Segmentation")