Dataset · Voxel51
Damage and structural element segmentation for concrete bridges
Voxel51/dacl10k
1 to 10 thousand real bridge images with multi-label segmentation across 19 damage and structural classes.
What's inside
The dacl10k (damage classification) dataset is gathered from engineering bureau and government agency databases following real concrete bridge inspections. It contains between 1,000 and 10,000 English images annotated for multi-label semantic segmentation, classification, and detection tasks. Annotations cover 19 classes: 13 damage types and 6 structural elements.
What you can do with it
- Training computer vision models for automated infrastructure crack and defect detection
- Testing detection and segmentation systems during drone bridge inspections
- Benchmarking multi-label segmentation algorithms in construction
- Tasks
- Image classification, Object detection
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 85.5K
- Likes
- 9
How to load it
from datasets import load_dataset
ds = load_dataset("Voxel51/dacl10k")